Deep-Representation-of-Visual-Descriptions

The project is ongoing, so changes will be frequent!!

Why?

Documentation

Aim

Translating text in form of single statement human written descriptions directly into image through following steps.

1. Make machines understand how humans represent an abstract image.
2. To make an image generator out of visual descriptions.
3. To make the model platform independent.

Deep Convolutional Networks and RNN have already yielded discriminative and generalizable text representation of images. We also have DCGAN to generate natural images using competing Discriminative and Generator functions.

A basic pictorial representation

Dependencies

1. Python
2. Pytorch
3. torchfile
4. nltk- ('punkt')
5. pandas
6. scikit-learn
7. python-dateutil
8. easydict

Data Download

Repo :

Models :

Trained models:

Eval models

Current Outputs

Text : flat screen television on top of an old tv console

Text : a large red and white boat floating on top of a lake

TEXT :this bird is red and white in color with a stubby beak and red eye rings

Text : this bird is yellow with black on its head and has a very short beak

Caption Generation with Attention:

Contributors

Team Members	Github	LinkedIN
Ashutosh Mishra LEAD	ASH1998	Ashutosh's LinkedIN
V. Praveen	praveenvoona	Praveen's LinkedIN
Deepak Kumar Behera	Github	LinkedIN
Madan Mohan Mohapatra	Github	LinkedIN

Contributing :

Website :

https://ash1998.github.io/Deep-Representation-of-Visual-Descriptions/