Simple (and explanatory) Pytorch implementation of Dog Breed Classifier using Transfer Learning

The task is to perform comparison of different CNN models for Fine-grained Classification task on 133 class Dog breed dataset. The models used are as follows:

• Custom 6-layer CNN model
• VGG-16 (trained from scratch)
• VGG-16 (pre-trained)
• ResNet101 (pre-trained)

Data can be downloaded from here. Weights for the trained model can be found on the project space here on Google Drive.