surawut/chess-move-tracking-yolo11
Object Detection
β’
Updated
β’
278
image
imagewidth (px) 640
640
|
|---|
This dataset is designed for Keypoint Detection (Pose Estimation) tasks, specifically to detect the 4 semantic corners of a chessboard from various camera angles.
It is intended to be used as Phase 1 of a Chess Move Tracking pipeline. The detected keypoints are used to calculate the Homography matrix for Perspective Warping (straightening the board).
a1, h1, a8, h8)The model is trained to detect 1 Class (Chessboard) with 4 Keypoints:
a1 (Bottom-Left in White perspective)h1 (Bottom-Right in White perspective)a8 (Top-Left in White perspective)h8 (Top-Right in White perspective)| Split | Images | Description |
|---|---|---|
| Train | 3390 | Extracted from Various Chess Play Videos with augmentations |
| Valid | 290 | Extracted from Various Chess Play Videos (Unseen angles) |