image_captioning_project

•Replicated image feature extraction model using ResNet50 and embedding matrix using globe6B50d vocabulary for our own custom vocabulary from Flickr8K dataset
•Implemented a image captioning model comprising ResNet50 model, embedding layer and LSTM layer by using categorical cross entropy for loss function and adam as optimizer
•Analyzed the model for various hyper-parameters such as no. of epochs, batch size, vocabulary size formulated a model which is computationally much cheaper and faster