The dataset used on this project is the "animal_10", which contains 10 classes for animal classification. The classification has around 28K animal images and the 10 classes are dog, cat, horse, spyder, butterfly, chicken, sheep, cow, squirrel, elephant.
The objective of this project is to apply the CNN concepts learned creating 2 CNN's architecture, training and evaluating the model, and after analyse 2 models using transfer learning using the same conditions as the previous ones created.
- Part A: CNN architecture and Training from scratch
- Part B: CNN and Training using transfer learning
https://pytorch.org/vision/stable/transforms.html
https://www.geeksforgeeks.org/how-to-normalize-images-in-pytorch/
https://pytorch.org/vision/stable/models.html#classification
https://papers.nips.cc/paper/2012/hash/c399862d3b9d6b76c8436e924a68c45b-Abstract.html
https://pytorch.org/vision/main/models/generated/torchvision.models.alexnet.html