Fashion Recommendation System developed by Pranjal Subhash (https://github.com/Pranjal1510) and Shraddha Srivastava for their 6th Sem Minor Project under the guidance of Dr. Rajdeep Chatterjee
We use Keras for tensorflow and apply transfer learning through different models (VGG16, ResNet50 and DenseNet121) as the base of the network and compare them. Transfer learning is basically taking features learned on one problem, and leveraging them on a new, similar problem. We used a Sequential model as it is appropriate for a plain stack of layers where each layer has exactly one input tensor. We then use Cosine similarity algorithm to find similarity.
