PokeShadowBench

GitHub Repo

https://github.com/freddiev4/pokeshadowbench

Dataset Summary

This dataset contains silhouette images from the "Who's That Pokémon?" segments from the Pokémon anime series. Each entry includes:

• A silhouette image of a Pokémon
• The Pokédex number
• The name of the Pokémon

The dataset is focused on the Indigo League, and includes 61 Pokemon.

Usage

from datasets import load_dataset

# Load the dataset
dataset = load_dataset("freddie/pokeshadowbench")

# Access images and metadata
test_dataset = dataset["test"]

# Get a specific Pokemon's images
bulbasaur = test_dataset.filter(lambda x: x["name"] == "bulbasaur")

Dataset Structure

The dataset is organized in the following structure:

test/
├── bulbasaur/
│   └── 001-bulbasaur.png
├── ivysaur/
│   └── 002-ivysaur.png
...

Each Pokémon has its own directory containing its silhouette image, along with a metadata.jsonl file containing Pokémon information.

Data Fields

Each entry in metadata.jsonl contains:

• file_name: Path to the image file
• name: Name of the Pokémon
• dex_number: Pokédex number of the Pokémon

Supported Tasks

• Image Classification: Identify Pokémon from their silhouettes
• Pokémon Recognition: Match silhouettes with their corresponding Pokémon names and numbers

Source Data

The images are sourced from the "Who's That Pokémon?" segments in the Pokémon anime series. Each image is:

• Screenshotted from the anime
• Cropped to focus on the Pokémon
• Named according to the format: XXX-name.png where:
  • XXX is the Pokédex number
  • name is the Pokémon name
• Source: https://pocketmonsters.net/segments

Licensing Information

This dataset is released under the Apache 2.0 license.

Citation Information

@misc{who-that-pokemon,
  author = {Freddie Vargus},
  title = {PokeShadowBench},
  year = {2025},
  publisher = {Hugging Face},
  journal = {Hugging Face Hub},
  howpublished = {\url{https://huggingface.co/datasets/freddie/pokeshadowbench}}
}