AI-Image-Caption-Generator

An automated image caption generator built using the Flickr-8k dataset

This project implements an Image caption generator using Deep Learning. We use Flickr-8K dataset to train our model. Each image in the dataset comes with 5 manually generated descriptions of the image. For ensuring syntactic correctness, we use GloVe vectors. Our model consists mainly of 2 components: First is CNN based image feature extractor ( we use Inception V3 model for this). Second part is sentence generator, which is RNN based(We use LSTMs for this). We remove the last layer of the Inception V3 model and feed it to the LSTM network along with a dummy start token. End of the generated sequence is marked by a special end token.