Neural network created using Keras to experiment with different architectures and optimizers.
It trains on the Homus dataset which contains musical notes and classifies them with a 94% accuracy.
The Homus dataset was augmented using my processImages script.
To train the network use the homus_cnnNoCV.py
To train the network with k-fold crossvalidation use homus_cnnCV.py
To load a pretrained model you can find the h5 files in each of the different folders for the different optimizers.