Implemented two image captioning models trained on Flickr dataset [ 6000 training images ] Baseline model impleemented with Xception CNN and LSTM for decoder architecture Improved model by fusing CNN features with Scene graph embeddings obtrained from torchvision
