Downloads · 30 days
0
chess-board-photo-onnx
cstr/chess-board-photo-onnx
chess-board-photo-onnx is a machine learning model from cstr. Use it for the machine learning task on the model card, and read the license before you ship it in a product. It is set up for onnx. The card lists the license as mit.
Two MobileNetV3-Small classifiers for reading a photo of a physical chessboard, used by CrispChess together with a Dart port of chesscog's board localisation and crops (Georg Wölflein and Ognjen Arandjelović, J. Imagi…
Access
Public
Updated Sep 24, 2026
Repo size
12.2 MB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.onnx12.2 MB · 100%
At a glance
- Library
- onnx
- License
- mit
- Access
- Public
- Created
- Sep 24, 2026
- Updated
- Sep 24, 2026
- SHA
- 4690fd54
Model card
From the Hugging Face model README
Chess board photo classifiers, ONNX
Two MobileNetV3-Small classifiers for reading a photo of a physical chessboard, used by CrispChess together with a Dart port of chesscog's board localisation and crops (Georg Wölflein and Ognjen Arandjelović, J. Imaging 2021, MIT).
| File | Input | Output |
|---|---|---|
board_photo_occupancy.onnx | 100×100 square crops (chesscog's occupancy crop) | empty / occupied |
board_photo_pieces.onnx | 100×200 piece crops (chesscog's piece crop) | 12 piece classes |
Trained for CrispChess (MIT) from torchvision's ImageNet-initialised MobileNetV3-Small (BSD-3) on:
- chesscog's synthetic renders (OSF xf3ka), CC BY 4.0;
- samryan18/chess-dataset, © 2019 Samuel Ryan, Mukund Venkateswaran, Kurt Convey, Michael Deng, MIT;
- Roboflow 100 "chess pieces" (
chess-pieces-mjzgj), CC BY 4.0.
Accuracy (end to end with the localiser): 99.6–99.9% of squares on held-out photos of the piece sets seen in training; 85% of squares (every error a piece type or colour) on chesscog's photos of an unseen wooden set. Unusual sets need checking.
- Library
- onnx
- License
- mit
- Created
- Sep 24, 2026
- Updated
- Sep 24, 2026