The project focusses on classification of Buddhist heritage sites from fake with a state of the art dataset. We collected a dataset of positive samples from surroundings of Taxila of over 180 videos. Currently we are trying to use GANs to generate negative samples and train the discriminator for accurate classification.