sort mask from left to right

app.py

@@ -107,7 +107,8 @@ def process_image(image_input, image_url, task_prompt, text_prompt=None, dilate=
             (image_width, image_height) = image_input.size
             bboxes = detections.xyxy
             merge_mask_image = np.zeros((image_height, image_width), dtype=np.uint8)
-            
+            # sort from left to right
+            bboxes = sorted(bboxes, key=lambda bbox: bbox[0])
             for bbox in bboxes:
                 x1, y1, x2, y2 = map(int, bbox)
                 cv2.rectangle(merge_mask_image, (x1, y1), (x2, y2), 255, thickness=cv2.FILLED)

utils/florence.py

@@ -49,6 +49,7 @@ def run_florence_inference(
     else:
         prompt = task
     inputs = processor(text=prompt, images=image, return_tensors="pt").to(device)
+    print(inputs)
     generated_ids = model.generate(
         input_ids=inputs["input_ids"],
         pixel_values=inputs["pixel_values"],

utils/sam.py

@@ -37,7 +37,10 @@ def run_sam_inference(
 ) -> sv.Detections:
     image = np.array(image.convert("RGB"))
     model.set_image(image)
-    mask, score, _ = model.predict(box=detections.xyxy, multimask_output=False)
+    # from left to right
+    bboxes = detections.xyxy
+    bboxes = sorted(bboxes, key=lambda bbox: bbox[0])
+    mask, score, _ = model.predict(box=bboxes, multimask_output=False)
 
     # dirty fix; remove this later
     if len(mask.shape) == 4: