Image Captioning with Tensorflow and Keras and deployment using Flask and Docker in Heroku

This repository contains two parts:

1)Train image captioning model- check the current directory - :

-the model trained in Flicker8k_Dataset which contains around 40 thousand image and their corresponding captions-you can download the dataset from here here-.
-have used glove embeddings and LSTM layer to capture text information.
-have uesd inceptionv3 pretrained model - it was trained in imagenet - to capture image information.
-the model trained in GCP for 3 epchos and the output model is saved-caption_model.model-.
-you can pass training process and use my trained model directly but be aware the if you can train the model for more epochs that will increase the performance a lot,actually i did`t do that because it will cost me a lot :"D.

2)Deploy this model into production using Flask , Docker and Heroku- check the deployment directory -

-you can visit demo here https://imagecaptioningdemo.herokuapp.com/