Tensorflow project for predicting hiragana characters.
The model is trained from fonts (WIP), these are saved into image maps through make scrape. Found out about the MNIST dataset after building this 🙃.
make train - Train the model based on scraped fonts
make api - Start a flask server that runs predictions
make client - Start the canvas application