💻 Chess-diagram-recognition

Detect and identify a 2D chessboard and the configuration of its pieces through the
application of image processing.

🎥 Screenshot

📄 Description

First, we need to load an image:

Next job is to extract a 2D diagram:

Then, we will divide the diagram into 64 squares,
and identify each piece:
==> White queen!

🔧 Tools

• Python
• NumPy
• OpenCV
• Tkinter
• Tensorflow
• Keras
• And many more ...

📌 Tasks

• Migrate Python2 to Python3
• Refactor tools/raw_diagrams_to_diagrams.py
• Refactor tools/diagrams_to_squares.py
• Improve detection (Detect even rotated diagrams)
• Build Application.py
• Build JavaScript Application
• Data augmentation
• Play with different learning models (use transfer learning)
• Collect more labeled data (both diagrams and squares)

📚 Datasets

• Kaggle
• diagrams - cropped diagrams from images
• squares - divided into black and white subsets, each with 13 classes

🐛 Known bugs

• For now, these are all features :)

🎓 Author

Stefan Ilić