A project that generates caption for the given input image.
Dataset used:- Flickr8K
Framework used:- Tensorflow
The project workflow is divided into two portions:-
- Generating features from images and cleaning descriptions
- Modelling and evaluation using bleu scores
The features are extracted using VGG model and stored in a pickle format. While the descriptions are saved as txt file.
The model used is a simple LSTM model.
For more info on bleu scores:-