This project is done for a deep learning competition held at hackerearth.com and its about classifying images which has a very small dataset using transfer learning and image augmentation. I used pre-models like VGG16,ResNet50 for transfer learning. Although model has been optimized for specific image classification task, this code can be used as a generic image classifier for this 4 types of image classes.

Requirements: This project was built using Windows 10, Ananconda, Keras, and Tensorflow. The code has been tested with both CPU and GPU (2x Nvidia GeForce MX150).

Acknowledgement: A significant portion of this script came from Coursera deeplearning speacialization course which i completed. My main contribution is to make it work with all keras pre-train applications models, and add a higher level of abstraction.