Image Captioning

This program lets you generate captions using a Neural Network.

Dependencies

Tensorflow
Keras
Pillow
Pickle
Numpy

Description

Program takes an image as an input and generates a caption for the image. Program is set to train on Flickr 8k dataset. VGG16 model is used to encode images, then encoed data is passed through three layered RNN to generate captions.

preprocess.py is used to preprocess data.
train_model.py builds the model. It uses the data generated by preprocess.py to train the model.
Caption_generator.py takes image as an input and generates caption for it.

Structure of the model used: