README.md

metadata

annotations_creators:
  - manual
language:
  - en
license: mit
multilinguality: monolingual
pretty_name: Chess Pieces Detection (YOLO)
size_categories:
  - 1K<n<10K
source_datasets:
  - original
  - extended
task_categories:
  - object-detection

Chess Pieces Detection Dataset (YOLO format)

📦 Overview

This dataset is designed for object detection of chess pieces on a chessboard using YOLOv5/YOLOv10.

• Classes: 12 (6 white + 6 black pieces)
• Format: YOLO (txt annotations)
• Split: Train/Validation
• Image count: ~2208 images

🗂️ Structure

images/train/
images/val/
labels/train/
labels/val/
dataset.yaml

🏷️ Class Names

0 = white pawn
1 = white rook
2 = white knight
3 = white bishop
4 = white queen
5 = white king
6 = black pawn
7 = black rook
8 = black knight
9 = black bishop
10 = black queen
11 = black king
12 = empty square (not used)

🧠 Train with YOLOv10

# Train
yolo train model=yolov10n.pt data=dataset.yaml epochs=100 imgsz=640

# Predict
yolo predict model=<mybestmodel.pt> source=path/to/chessboard-sample.jpg

📄 License

MIT License

🙏 Acknowledgments