Keras and PyTorch implementation of the Capsule Network architecture for handwritten digit recognition
The folder mnist holds notebooks implementing the CapsNet architecture on the classic MNIST dataset
The folder kannada mnist holds notebooks implementing the CapsNet architecture on the Kannada MNIST and DIG-10K datasets
The folder src holds PyTorch code implementing the CapsNet architecture on the MNIST dataset(inference TBD)
Zip all the py files in src into CapsNet.zip and upload the zip to colab nb provided in src and run the cells in order
PyTorch implementation is almost entirely based on this notebook