- Download the Flickr8k dataset and store the images in the 'Flicker8k_Dataset' folder.
- Run each cell of the .ipynb file to view output generated at every step and to generate checkpoints. When training the model, a new checkpoint folder will be created and the 5 most recently trained checkpoints are saved.
- Run the 'app_flask_new.py' file to execute the application.
On uploading images to the application, the generated captions along with the image name is saved as a JSON object and image is stored in a 'gallery' folder.