This Jupyter Notebook Shows a Neural Image Captioning model using Merge Architecture in keras which generates captions for given image.
Transfer Learning was used to extract image features and formulate word relations using InceptionV3 and Glove 200d word embeddings.
The model was trained on Flickr8k Dataset and Inference was done using Greedy Search Decoder and Beam Search Decoder