This repo is being used to play around with different implementations of various CNNs to help understand which one we will ultimately choose for our capstone project. I am looking at incorporating GRADCAM, GRADCAM++ and Saliency maps to explain the classification. The goal of this repo is to arrive at a final pipeline that we can feed our private dataset into.

So far the implementations tested out on our private dataset have been:

1. EfficientNet in PyTorch
2. ResNet50 in PyTorch
3. AlexNet in PyTorch
4. VGG11 in PyTorch

Looking to try various models in TensorFlow, found in keras.applications...

Transfer Learning article from TensorFlow.

Guide on loading and preprocessing images in Tensorflow.

Here is a pretty comprehensive article on using pre-trained EffecientNet (all variants B0 to B7).