This project is about car classification for stanford car dataset. The Cars dataset contains 16,185 images of 196 classes of cars. The data is split into 8,144 training images and 8,041 testing images, where each class has been split roughly in a 50-50 split. Here I use transfer learning to speed up the training process.