img
images listlengths 1
1
| question
stringlengths 1.75k
2.06k
| answer
dict | match_rate
float64 0.5
1
|
|---|---|---|---|
This image was generated from the prompt: A man in a dark, textured coat stands in a misty jungle, holding a brown, aged map in his hands. The map, outlined in red, is unfurled and appears worn, with faint lines suggesting a route. He faces right, examining the map closely, surrounded by lush green ferns and towering trees, with soft light filtering through the canopy..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
532,
369,
901,
785
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A man in a black wetsuit surfs on a white board, riding a wave in choppy gray-green water. To his left and slightly above, a white seagull with gray-tipped wings and an orange beak flies mid-air, wings outstretched, captured within a red bounding box..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
80,
110,
305,
411
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A tall, dark silhouette of a figure stands within a red bounding box, leaning forward with arms extended toward a small child beside them. Both are silhouetted against a glowing blue, futuristic cityscape with circular holographic patterns on surrounding walls. Their reflections appear on the shiny floor. The figure is positioned to the left of the child, appearing to reach out protectively..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
151,
47,
722,
680
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A ginger cat with striking yellow eyes sits upright on a wooden surface, enclosed in a red bounding box. To its right, a brass-colored mantel clock with a cream face and black numerals displays the time. The cat’s fur is soft and striped, and it gazes directly forward. Both objects rest against a neutral beige wall..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
36,
47,
488,
929
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A white worksheet titled "Tsunami Physics WebQuest Waves Worksheet" is centered within a red bounding box. It includes fields for Name, Period, and Date at the top, followed by three numbered questions about traveling vs. standing waves, mechanical wave examples, and vacuum propagation. The worksheet is developed by Isadel Eddy, March 2009, and intended for a waves review..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
3,
3,
996,
995
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A Luigi figurine in green hat and blue overalls points left, attached to a silver keyring. To the right, a plain white rectangular tag with rounded corners hangs from the ring, enclosed in a red bounding box. The tag is smooth, unmarked, and positioned vertically beside the character..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
566,
283,
881,
698
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A brown fried chicken drumstick is centered within a red bounding box, floating against a cosmic background of swirling red and blue nebulae. The drumstick has a glossy, textured surface with visible skin and bone, illuminated by dramatic lighting that highlights its curved shape. The background features scattered stars, enhancing the surreal, space-like setting..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
238,
70,
864,
900
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A red baseball cap with a flat, dark brown brim is centered in the image. The cap features a diamond-shaped brown leather patch on the front, embroidered with a circular logo and the word "UPFRONT" below it. The cap’s fabric has a textured weave, and it is displayed against a plain white background, fully enclosed by a red bounding box..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
21,
116,
990,
915
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A man stands with hands on hips, wearing a blue soccer jersey inside the red bounding box. The jersey features white Adidas stripes, a "Jeep" sponsor logo, the Juventus crest with three stars, and a UEFA Champions League patch. He also wears blue goalkeeper gloves. A hashtag "#WELCOMEWOLCIECH" appears at the bottom..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
63,
232,
916,
992
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A green abstract background with smooth, flowing wave-like curves. The design features a gradient from light green at the top to dark green at the bottom, creating a soft, luminous effect. The red bounding box encloses the entire image, highlighting the full abstract composition. No other objects are present..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
2,
4,
996,
994
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A baby with raised arms and tattooed forearms wears a white shirt and blue denim pants, highlighted by a red bounding box. The pants feature a front button, belt loops, and visible stitching, fitting snugly below the shirt. The baby’s face shows an open-mouthed expression, set against a gradient teal background..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
187,
557,
795,
995
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: Two skiers in colorful jackets and helmets descend a snowy slope under a clear blue sky. In the foreground, a brown lizard with a long tail lies motionless on the snow, positioned between their skis. The lizard’s textured scales contrast with the smooth, sunlit snow..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
154,
775,
482,
937
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A vintage car is centered within a red bounding box, viewed from the front. The car features a rounded hood, vertical grille, round headlights, and whitewall tires. Bold black text reading “HOT FRESH” overlays the car. The image has a sepia tone with a textured, grainy appearance..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
130,
81,
994,
889
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A rider in black gear leans forward on a blue Yamaha sport motorcycle, enclosed in a red bounding box. The bike features a sleek design with visible "YAMAHA" branding, blue wheels, and a clear windshield. It travels on a paved road with a blurred desert landscape and distant mountains in the background..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
427,
424,
935,
872
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A large gray elephant with wrinkled skin, white tusks, and large ears walks on sandy ground within a red bounding box. Behind a metal railing, a group of people, including adults and children, observes from a grassy area. Lush green trees form the background. The elephant’s trunk hangs downward as it moves forward..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
2,
177,
814,
833
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A family of four elephants—two adults and two calves—walks across a paved road. To the right, a yellow diamond-shaped wildlife sign with a black elephant silhouette and the word "WILDLIFE" is mounted on a post, bordered in red. Lush green trees line the background..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
819,
329,
969,
519
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A dense green forest with tall, leafy trees forms the background, enclosed by a red bounding box. In the foreground, a military armored vehicle sits on a dirt path, surrounded by grazing cattle of various colors. The forest extends behind the vehicle, creating a contrast between nature and machinery. The sky above is pale blue with light clouds..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
2,
66,
997,
816
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A dark silhouette of a girl, wearing a hat and a flared dress, is captured mid-leap within a red bounding box. She stands on a tiled floor surrounded by an intricate network of large, rusted metal pipes. The scene is dimly lit with a bluish tone, emphasizing the industrial setting. The girl’s arms are outstretched, and her legs are bent, suggesting dynamic movement. The pipes form a dense, maze-like structure around her, with no visible interaction between her and the environment..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
390,
344,
613,
692
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A green tape dispenser with a white tape roll sits on a polished wooden surface, enclosed by a red bounding box. To its left, a red, rounded object resembling a button or knob rests on the same surface. The dispenser features a cutout for tape dispensing and a small metal blade for cutting. Both objects are positioned side by side, with the dispenser slightly closer to the foreground..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
469,
253,
906,
669
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A blue Volvo XC90 SUV is centered in a red bounding box, viewed from the rear three-quarter angle. It features a roof rack, silver side steps, and dual exhaust pipes. The vehicle has tinted windows, alloy wheels, and a European-style license plate reading "JGC 558". The background is a smooth gray gradient..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
141,
235,
851,
777
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A beige vintage hatchback car, enclosed in a red bounding box, is positioned on the left side of the image. To its right, a coral-colored hairdryer stands upright, its nozzle facing the car. Both objects rest on a smooth, uniform beige surface under soft, even lighting. The car features round headlights, a black grille, and silver hubcaps..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
2,
361,
744,
815
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A black lantern with a glowing flame sits on a wooden surface, enclosed in a red bounding box. It features a glass chamber, metal frame, and handle. To its right, a brown canvas bag with leather handles and a front pocket rests nearby. The scene is dimly lit, emphasizing the lantern’s warm glow..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
62,
152,
380,
869
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A black chess pawn lies on its side within a red bounding box, emitting blue flames and wisps of smoke. It rests on a dark square of a chessboard, with other white chess pieces blurred in the background, arranged in rows behind it. The scene is bathed in cool blue lighting, emphasizing the fallen pawn’s dramatic, smoldering state..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
196,
571,
722,
876
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A gray bird with a red crown and long black legs stands centered on a wooden floor, framed by a red bounding box. Behind it, a large window reveals a starry night sky over a distant blue horizon. The room is painted in deep blues, with textured brushstrokes creating a moody, introspective atmosphere. The bird faces right, its long neck curved slightly, casting a faint shadow on the floor..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
386,
198,
608,
875
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A green ball with a glossy finish sits on a blue felt surface, highlighted by a red bounding box. To its left are an orange and a purple ball, while two granite curling stones with wooden handles rest to its right. All objects are stationary, arranged in a scattered pattern on the smooth, dark blue surface..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
278,
300,
458,
481
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A yellow cow stands on wet pavement under a blue umbrella. The cow has short horns, perked ears, and a calm expression, facing forward. The umbrella, with a black handle, is positioned above the cow, slightly to its left. A gray concrete wall forms the background. The cow is fully enclosed within the red bounding box..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
115,
385,
907,
957
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A brown architectural sketch within a red bounding box depicts a multi-level, arched interior with balconies, columns, and skylights. Light patches suggest windows or openings. The foreground shows an open space with small figures. The sketch uses charcoal or pencil with white highlights, rendered on aged paper. The artist’s signature appears in the lower right corner..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
54,
35,
915,
948
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A small brown pufferfish with spiky white underbelly and large round eye is enclosed in a red bounding box. It floats near a human finger on the right, with soft, warm lighting and blurred green foliage in the background. The fish’s fins are translucent, and its body is inflated, appearing round and defensive..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
84,
290,
654,
666
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A black hammer with a wooden handle stands upright on a wooden surface, enclosed in a red bounding box. To its right is a black traffic light with three unlit circular lenses. Both objects are silhouetted against a warm, glowing circular backdrop resembling a sun or moon, casting soft light and shadows..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
188,
352,
495,
943
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A white Minecraft skeleton made of pixelated beads is centered in a red bounding box. It features black square eyes and a rectangular mouth, with blocky arms and legs. Above it, bold white text reads “Make a MINECRAFT Skeleton.” Below, a small cartoon boy in a superhero cape stands beside the URL “BrainPowerBoy.com,” all set against a black background..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
289,
325,
717,
919
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A brown tarsier with large, round, golden-brown eyes and prominent ears perches on a tree branch within a red bounding box. Its fur is soft and grayish-brown, and its hands grip the branch firmly. The scene is set at night under a full moon, with dark blue sky and silhouetted tree branches in the background..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
233,
136,
856,
926
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A pink ceramic bowl sits in the left side of a red sink, which features a matching red faucet with a cross-shaped handle. The sink is set against a red tiled backsplash with white grout. The bowl is smooth, rounded, and empty, positioned near the drain on the right..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
199,
482,
665,
822
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A bright blue vintage sedan is positioned on a neutral gray surface, viewed from the front-left side. In the foreground, to the right of the car, sits a glossy red ceramic mug with a C-shaped handle, outlined by a red bounding box. The mug is unadorned and stands alone, casting a soft shadow..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
679,
605,
996,
902
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A dark green pumpkin with a rough, ribbed texture and a short, woody stem sits on rocky ground, highlighted by a red bounding box. To its right, a metallic green weevil stands on the rocks, while a brown duck-like creature observes from behind. In the foreground, a sturgeon lies on the ground. Snow-capped mountains rise in the background under a clear blue sky..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
8,
367,
440,
690
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: Two men stand in a dimly lit, arched stone hall. The man on the left, in a dark jacket, holds a metallic dragonfly in his right hand. The dragonfly, with intricate veined wings and a dark, segmented body, is highlighted in a red bounding box. The man on the right, wearing glasses and a cable-knit sweater, observes it closely. Both men have serious expressions..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
338,
438,
808,
714
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A man with a mustache, wearing a flat cap and a button-up shirt, stands centered in a red-bordered frame. The black-and-white image uses dramatic lighting, highlighting his facial features and textured clothing. He gazes upward, with smoke-like shadows in the dark background. His right arm is slightly bent, hand near his side..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
220,
2,
864,
995
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A framed business license with a red border is mounted on a light-colored wall. The license features a golden seal at the top, Chinese text, a QR code, and a red official stamp. It is encased in a dark wooden frame with a cream-colored mat. The document includes registration details, date, and address..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
114,
12,
870,
984
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A red velvet armchair with tufted backrest and rounded arms sits centered in a dimly lit room. It rests on four dark, tapered legs. Warm, golden light illuminates the left side, casting soft shadows on the right. The chair is the sole focus, framed by a red bounding box, with a plain, gradient background transitioning from orange to dark brown..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
111,
165,
919,
987
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A pink pig with a curly tail, rosy cheeks, and a smiling expression walks forward, enclosed in a red bounding box. To its left, a small green turtle follows closely. Both are drawn in a soft, watercolor-like style on a beige background, with the pig slightly ahead, suggesting movement and companionship..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
196,
180,
937,
776
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A black and white dog with brown markings around its eyes and muzzle sits in the foreground, framed by a red bounding box. Its fur is thick and glossy, with a white chest and a black nose. Behind it, a misty forest with tall pine trees fades into a cool blue-gray background. The dog gazes directly forward with calm, expressive brown eyes..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
123,
42,
859,
993
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A lone figure in a dark, narrow canoe paddles through turbulent, dark waters under a starlit, stormy sky. The canoe, highlighted by a red bounding box, is positioned centrally in the lower half of the image. The figure, cloaked and hooded, holds a paddle, facing forward into the misty, jagged landscape. Above, the title "SPIRIT WALKER" looms in pale, textured letters..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
308,
824,
651,
954
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A black smartphone with a white screen lies on green grass in the foreground, outlined by a red box. Above it, a red and orange dragon with green-tipped wings flies in a cloudy sky, tethered by thin white strings to the phone. Lush green trees form the background..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
504,
478,
803,
939
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A blue octagonal stop sign is centered in the image, with white text reading "STOP THE ILLINOIS TOLL TAX." It is framed by a red bounding box and placed over a blurred background showing cars on a road. The sign is part of a news article webpage from examiner.com, positioned below the headline and above the article text..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
275,
517,
731,
897
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A toasted sandwich with ham, cheese, tomato, and lettuce sits on a wooden surface, enclosed in a red bounding box. To its left, a pair of scissors with black handles and metallic blades rests on the same surface. The sandwich features golden-brown bread, layered with green lettuce, red tomato slices, pale yellow cheese, and pink ham..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
320,
77,
990,
692
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A brown bear with thick, dark fur is centered within a red bounding box, crouched low on a dark, grid-patterned surface. The bear’s head is lowered, eyes nearly closed, and front paws extended forward. The grid glows with blue and red lines, creating a digital, futuristic environment. The bear appears motionless, illuminated by subtle light from above..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
238,
109,
831,
788
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A golden-brown dog lies on a light gray carpet in front of a beige sofa, which is framed by a red bounding box. The sofa features two matching beige cushions and two off-white throw pillows. The dog rests with its front paws extended forward, gazing directly at the camera. The scene is well-lit, with soft, neutral tones throughout..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
2,
41,
996,
684
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A white circular pin with a glossy finish is centered on a black textured fabric background. The pin features a purple banner with yellow text reading "SOUTH SHORE LINE," outlined in yellow. The red bounding box highlights the pin, which is the sole object in focus..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
122,
123,
871,
871
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A transparent orange screwdriver with a ribbed handle and metallic shaft lies diagonally within a red bounding box. The screwdriver’s flat tip points toward the bottom-right. It rests on a textured, cloud-like background with warm orange and purple hues, illuminated by soft, directional light casting subtle shadows..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
215,
88,
785,
889
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A man in a white t-shirt and black shorts is lunging forward on a grassy field, reaching for a white frisbee highlighted in a red box. A woman in a blue shirt and black shorts, wearing a black cap, is behind him, also in motion. Trees and open green space are visible in the background under a clear sky..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
4,
428,
234,
535
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A blue and yellow shield with a white feather emblem lies on sandy ground, outlined by a red box. To its right, a wooden xylophone with two mallets rests beside a brown bear sitting upright, facing the instruments. Sunlight casts long diagonal shadows across the scene..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
156,
323,
273,
638
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: The book cover features white text reading "CLINICAL PHARMACOLOGY IN PSYCHIATRY" and "STRATEGIES IN PSYCHOTROPIC DRUG DEVELOPMENT" on a dark navy background. Below the text, a red-bordered square contains a geometric pattern composed of orange, teal, brown, and beige shapes, including squares, triangles, and curved segments, arranged in a mosaic-like design..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
149,
425,
881,
992
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A man with a split face—half realistic with a mustache, half glowing blue neon—wears a dark suit. In the lower-left corner, a blue bicycle with black tires and handlebars is framed by a red bounding box, positioned in front of his torso. The bicycle appears stationary, with visible spokes and a black seat..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
2,
664,
447,
994
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A glazed donut with a bite taken out of it sits on a white napkin, enclosed in a red bounding box. To its right is a black plastic fork, and further right, a paper coffee cup with a brown sleeve labeled "MIRANDA JOE'S COFFEE" and a black lid. All items rest on a wooden table surface..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
61,
460,
461,
783
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A large green fly with iridescent metallic body, translucent wings, and prominent reddish-brown compound eyes hovers centrally in a red bounding box. It is positioned above a turbulent, misty chasm flanked by dark, jagged rock cliffs under a stormy, overcast sky..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
261,
228,
726,
788
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A silhouetted man in a hat and long coat walks forward along a concrete path between stacked shipping containers. He is centered in the frame, framed by a red bounding box, with industrial cranes visible in the background. The scene is monochromatic, with strong backlighting creating deep shadows and emphasizing his outline..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
499,
266,
655,
880
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A silver laptop with a black screen and keyboard sits on a light wooden desk, enclosed by a red bounding box. To its right is a Diet Coke bottle. Behind it, a black desktop monitor and tower are visible. Two black cables connect to the laptop’s left side. The scene is well-lit with neutral background..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
26,
157,
747,
820
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A woman with curly dark hair and a warm smile wears a blue heathered crew-neck shirt, highlighted by a red bounding box. The shirt covers her torso and upper arms, with visible stitching around the neckline. She stands outdoors on a paved street, with blurred buildings in the background..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
51,
569,
969,
995
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A black motorcycle with chrome accents speeds along a dirt road, kicking up dust. The bike features a round headlight, dual mirrors, and a V-twin engine. It is positioned slightly left of center, with a blurred green hillside and blue sky with clouds in the background. The road curves gently, and the motorcycle leans slightly into the turn..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
150,
179,
842,
784
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A large, leafy green plant with deeply lobed leaves sits in a black plastic pot, outlined by a red bounding box. The pot has vertical ridges, drainage holes at the base, and a white label on its side. It rests on a light brown wooden floor, positioned in front of a white shelf unit..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
323,
656,
664,
963
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A man with short dark hair and a light beard wears a black baseball cap. The cap, highlighted by a red bounding box, features a detailed portrait of a young man with dark hair and a tan shirt, printed on the front panel. The man’s face is visible below the cap, looking directly at the camera against a neutral beige background..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
218,
26,
791,
465
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A person in a teal jacket holds a small, fluffy brown poodle in their arms. The poodle has curly fur, dark eyes, and floppy ears, positioned centrally in the red bounding box. To the right, another person in a brown hat and scarf gestures toward the poodle. The scene is set against a cloudy sky with golden lighting..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
176,
491,
548,
968
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A gray pigeon with iridescent green and purple neck feathers stands on concrete, framed by a red bounding box. Behind it, a skateboard with orange wheels and black grip tape rests on the pavement. The pigeon faces right, casting a shadow, while the skateboard lies horizontally in the background..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
3,
443,
601,
936
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A white sailboat named "NAVY FLOW" sails on calm blue waters under a clear sky with scattered clouds. The boat features two large beige sails, a tall mast, and a cabin with windows. It moves forward, creating a small wake, with the red bounding box highlighting the entire vessel..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
97,
9,
869,
901
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A weathered hand rests near a metallic blue emblem embedded in a dark woven surface. The emblem, highlighted by a red box, features an angular, maze-like design with reflective edges. The hand, with visible wrinkles and short nails, lies to the left, partially covering the woven texture. The background is blurred, showing dim blue tones and bokeh lights..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
433,
533,
852,
862
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A white airplane with blue tail and landing gear extended is shown in a red bounding box, positioned to the right of a blue-and-white floral porcelain vase resting sideways on a cream-colored mantelpiece. The vase has gold-trimmed edges and is set against a beige wall..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
766,
340,
971,
438
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A metallic frog with textured, bronze-like skin and large, reflective eyes sits on a stone floor within a red bounding box. To its right lies a dark feather quill. The background features a dim, gothic-style stone staircase with arches, creating a mysterious, shadowy atmosphere. The frog is positioned in the foreground, slightly left of center, facing right..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
30,
516,
626,
934
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A green fabric sofa with tufted backrest and three seat cushions is centered in the image, framed by a red bounding box. It features rounded armrests and tapered wooden legs. The sofa sits on a light beige floor against a plain off-white wall, with no other objects or interactions visible..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
17,
278,
984,
777
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A pair of blue jeans stands upright in a twilight forest, outlined by a red bounding box. The jeans are cuffed at the ankles, with visible back pockets and stitching. Glowing mushrooms surround the base, and dark trees frame the scene under a dusky purple-blue sky..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
368,
63,
659,
923
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A woman in a white lab coat holds a black and yellow spider in her palm, its legs splayed and abdomen patterned. To her left, a woman with glasses observes intently. Behind them, a man in a fedora and a blue-skinned humanoid stand under a starry sky with green auroras. The spider is the focal point, highlighted by a red box..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
286,
615,
541,
903
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A couple rides a colorful jet ski on the ocean, laughing as water sprays around them. To the left, a yellow rubber duck wearing black sunglasses floats in the waves, highlighted by a red bounding box. The man sits behind the woman, both wearing life vests, with the duck positioned in the background near the water’s surface..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
24,
202,
225,
409
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A beige speckled ceramic bottle with a narrow neck and brown unglazed base stands on a textured surface. To its left, a wooden spoon leans against the same neutral-toned backdrop. The bottle is highlighted by a red bounding box, positioned centrally in the frame, with soft lighting casting subtle shadows..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
421,
133,
774,
914
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A black and gray Asics sneaker is shown in a side-back view within a red bounding box. The shoe features a suede upper, gray Asics tiger stripe, black laces, a pull tab at the heel, and a textured gray midsole with black outsole. The brand name "ASICS" and "GEL" are visible on the heel. The sneaker is isolated on a plain white background..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
75,
466,
924,
949
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A dark brown wooden cabinet with arched glass doors and two lower drawers stands on the left, framed by a red bounding box. To its right, a vintage-style metal fan with a protective cage and four blades rests on a pedestal. Both objects are set against a deep blue, starry night sky background with soft light rays..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
99,
129,
612,
992
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A white and orange soccer ball with black accents rests on the forest floor, highlighted by a red bounding box. To its right sits a smooth, tan gourd. Both objects are illuminated by dappled light filtering through dense, dark green foliage above, casting soft shadows on the mossy ground..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
196,
485,
494,
784
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A white horse with a flowing mane and tail stands in a shallow stream, bending its head to drink. The horse is framed by a red bounding box, positioned centrally. Surrounding the stream are rocks, grass, and trees with leafy canopies in the background, under a sky with fluffy clouds..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
201,
224,
887,
843
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A silhouetted figure in a black coat and hat stands against a starry night sky, holding a glowing spiral shell at chest level. The hat, outlined in a red bounding box, sits atop the figure’s head, casting shadow over their face. The shell emits a warm, golden light, contrasting with the cool blue background and distant mountain silhouette below..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
357,
127,
639,
239
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A brown doormat with "WELCOME" embossed in dark letters lies centered in a red bounding box. It has a textured, fibrous surface and a darker border. The mat floats against a cosmic background of swirling red, orange, and dark nebulae, with beams of light shining from above. No other objects interact with it..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
98,
362,
920,
803
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A green chameleon with textured, scaly skin and a large, round eye perches on a stone ledge within a red bounding box. To its left, a weathered stone tower with arched windows stands beside a large, golden quill. A full moon glows in the dark, cloudy sky above, casting a soft light over the scene..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
537,
303,
995,
977
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A large dark bull with prominent curved horns stands in shallow water at sunset, framed by a red bounding box. To its left, a young child with a quiver of arrows looks down. The bull’s dark, textured hide contrasts with the warm golden sky and distant hills. Two armored figures stand to the right, one holding a teapot, observing the scene..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
22,
41,
492,
755
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A coiled blue snake with a yellow underbelly rests on a stone surface in the foreground, enclosed by a red bounding box. To its right sits a mound of pale yellow, textured substance. Behind, a bearded man in robes sips from a cup, while three figures—two in dark clothing and one in a draped toga—stand in the dimly lit background, gazing upward..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
87,
680,
513,
956
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A silver fork with four tines and a tapered handle lies vertically on a wooden surface, enclosed in a red bounding box. To its right sits a closed, brown hardcover book with a textured fabric cover and cream-colored pages. The fork is positioned to the left of the book, both resting flat on the wood grain table..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
106,
84,
259,
902
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A sleek silver high-speed train travels along parallel tracks, framed by a red bounding box. To its left, a stone maze with several tiny figures navigating its paths sits adjacent to the rails. The scene is bathed in golden morning light, with misty forest trees surrounding the area, creating a serene, surreal atmosphere..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
537,
264,
906,
632
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A black circuit board labeled "ECLYPSE Z7" by Digilent is enclosed in a red bounding box on the left. It features gold-plated connectors, a central cooling fan, USB ports, an SD card slot, and various integrated circuits. To the right, text announces the ECLYPSE Z7 as a new tool for accelerating design flow by combining FPGA, software, and high-speed I/O..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
10,
57,
481,
955
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A beige moisturizer jar with a matching lid and embossed "MOISTURIZER" text sits in the foreground, highlighted by a red bounding box. Behind it, to the left, is a beige box with the same text. Both objects rest on a neutral surface under soft, even lighting. The jar is rounded and matte, positioned slightly forward of the box..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
482,
360,
880,
768
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A translucent beige plastic bag, crumpled and weathered with visible creases and small dark spots, floats centrally within a red bounding box. It is illuminated from behind by warm, golden light. The background features dark, glossy, undulating surfaces with a glowing orange vent on the right, suggesting a futuristic or industrial setting..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
148,
139,
627,
843
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A woman in dark, armored attire sits on a patterned rug in a sunlit meadow, holding a brown acoustic guitar with a glossy wooden body and black pickguard. Her left hand presses the strings while her right hand strums. A man in a black suit sits beside her, leaning forward with his hand resting on the rug. The guitar is positioned across her lap, angled toward the man..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
2,
437,
665,
818
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: Three people stand in a dimly lit room, facing a large screen displaying a green, spiky fruit and a coiled cobra. The fruit, highlighted in a red box, has a textured, bumpy surface with yellow-green spots and a short stem. The cobra is to its right. Tall speakers flank the screen. The central figure wears a dark jacket with glowing blue and red lines..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
234,
128,
501,
622
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A black and white pinto horse, framed by a red bounding box, stands on grass with a knight in full metallic armor mounted atop. The horse wears a bridle and saddle, while the knight holds a lance upright. In the background, spectators sit behind a wooden fence under green trees..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
80,
176,
895,
994
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A red frisbee with concentric ridges lies on green grass, enclosed by a red bounding box. To its right, a black smartphone with a dark screen rests on the grass. Both objects are positioned side by side, viewed from above, with natural lighting highlighting their surfaces..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
57,
195,
608,
736
],
"bool": true
}
| 0.5
|
|
This image was generated from the prompt: A green zucchini with a speckled dark green skin and a bumpy green stem is enclosed in a red bounding box on the left. To its right lies a shiny, heart-shaped red apple. Both rest on a light wooden surface with visible grain. The zucchini is elongated and smooth, while the apple has a glossy finish..
Please carefully evaluate whether the image is strictly aligned with the prompt,
with particular attention to objects, quantities, attributes, spatial relationships,
and other key semantic details.
If the image accurately matches the prompt, answer 'true'. Otherwise, answer 'false'.
Respond strictly in the following JSON format:
{
"answer": true/false,
"bbox": [x1, y1, x2, y2],
"explanation": "If answer is false, briefly summarize the main mismatch."
}
Bounding box (bbox) specification:
- (x1, y1) and (x2, y2) are pixel coordinates of the region requiring refinement,
where (x1, y1) is the top-left corner and (x2, y2) is the bottom-right corner.
- The image coordinate system is defined with (0, 0) at the top-left
and (999, 999) at the bottom-right, with x increasing horizontally
and y increasing vertically.
Meaning of bbox:
A bbox identifies the region that needs refinement, including one of:
1. add — add a missing object within the bbox.
2. delete — remove an incorrect object in the bbox.
3. modify — adjust an object within the bbox (e.g., color, size, position, attributes).
If the image is correct (answer = true), return an empty list for "bbox".
|
{
"bbox": [
120,
72,
395,
903
],
"bool": true
}
| 0.5
|
End of preview. Expand
in Data Studio
README.md exists but content is empty.
- Downloads last month
- 169