chess-vision's Introduction

Chess-Vision

The goal of this project is to convert a video of a live chess game to a pgn file.

Demo

ScaledYOLOv4_tiny_detection_result:

Installation

Clone Project

git clone https://github.com/jkner/Chess-Vision.git

Install Anaconda:
```
https://www.anaconda.com/
```
Install pip:
```
conda update --all
conda install pip
```
Create an environment, activate it then install tensorflow-gpu:

Tensorflow Gpu :: Anaconda.org
```
conda install tensorflow-gpu
```
Install requirements:
```
pip install -r requirements.txt
```

Evaluation Results:

Evaluation results(GTX3060Ti,[email protected]):

model	Chess Pieces
Scaled-YoloV4-tiny(Multi-Scale)	0.995
Scaled-YoloV4-tiny(Single-Scale)	0.989

Detection

To run the code:

python3 cv_chess.py

Detection Result

Video Demo

References

Board Detection & Piece Localization

Chess piece detection - digitalcommons.calpoly.edu. (n.d.). Retrieved May 14, 2022, from https://digitalcommons.calpoly.edu/cgi/viewcontent.cgi?article=1617&context=eesp

Underwood, A. (2020, October 22). Board Game Image Recognition Using Neural Networks. Medium. Retrieved May 13, 2022, from https://towardsdatascience.com/board-game-image-recognition-using-neural-networks-116fc876dafa

Training and Validating Model:

wangermeng2021. (n.d.). Wangermeng2021/scaled-yolov4-tensorflow2: A tensorflow2.x implementation of scaled-yolov4 as described in scaled-yolov4: Scaling cross stage partial network. GitHub. Retrieved May 13, 2022, from https://github.com/wangermeng2021/Scaled-YOLOv4-tensorflow2

Recommend Projects