surawut/chess-pieces-dataset-yolo

Chess Pieces Detection Dataset (Warped View) ♟️🔍

1. Dataset Description

This dataset contains images of chess pieces taken from a Top-Down (Warped) perspective. It is designed for Object Detection models (YOLO) to recognize chess pieces and hands on a standardized 640x640 grid.

It is intended to be used as Phase 2 of a Chess Move Tracking pipeline, receiving input from the Phase 1 Perspective Warper.

• Task: Object Detection
• Format: YOLOv8 / YOLO11 Detection Format
• Input: Warped/Rectified chessboard images (640x640)
• Classes: 13 Classes (12 Pieces + Hand)

2. Dataset Composition

The dataset includes diverse scenarios to ensure robustness:

1. Standard Gameplay: Frames extracted from real gameplay videos.
2. Occlusions: Frames where hands are covering pieces (labeled as Hand).
3. Negative Samples: Empty chessboards (only background) to reduce false positives (~10% of dataset).
4. Lighting Variations: Augmented brightness and exposure to simulate different environments.

3. Classes

The dataset uses the following class mapping:

ID	Class Name	Description
0	Hand	Human Hand
1	bB	Black Bishop
2	bK	Black King
3	bN	Black Knight
4	bP	Black Pawn
5	bQ	Black Queen
6	bR	Black Rook
7	wB	White Bishop
8	wK	White King
9	wN	White Knight
10	wP	White Pawn
11	wQ	White Queen
12	wR	White Rook