We were provided with a training set and a test set of dog images. The dataset consisted of 120 dog breeds and the goal was to create a classifier capable of determining a dog's breed from a photograph. We converted the images into tensors, adjusted there shape, grouped the images and trained them using a transfer learning model. We used the same process with the test dataset and finally fit the model.