Visualizing what ConvNets learn

This Github repository showcases the activations of a Convolutional Neural Network (CNN) used for classifying images of cats and dogs.

The purpose of this project is to provide a detailed explanation of how CNNs work, specifically how the different activation functions impact the CNN's performance and ability to classify images. By visualizing the different activations, convnet filters and heatmaps of class activation in an image of the CNN model, I aim to make it easier for developers and machine learning enthusiasts to understand how a CNN works under the hood.

Visualizing intermediate layers activations:

for more details check the visualizing intermediate layers activations directory.

Click to view intermediate layers activations

Visualizing Heatmaps of class activations using GradCAM:

for more details check the Visualizing Heatmaps of class activations directory.

Click to view heatmaps of class activations

Visualizing Filters of a CNN:

for more details check the Visualizing Filters of a CNN directory.

Click to view samples of filters of a pre-trained VGG19