Text to Image/Video synthesis using GANs.

This repository contains all the code that was developed during eyantra summer internship 2018 at IITB.

Objective

In this project we have developed models that can generate dataset of new images, images based on text description and video dataset based on men-tioned category.

Authors

• Deval Srivastava
• Aishwarya J Kalloli

Mentors

• Aditya Panwar
• Kalind Karia

Folders

• mnist: Contains all the code for mnist dataset generation
• texttoimage: Contains all the code for generation of images from text
• videogeneration: Contains all the code for the generation of videos of KTH dataset
• documentation: Contains all the progress presentations and posters

software and libraries to be installed

• Python3
• CudaNN
• Cuda
• lua
• torch

python requirements

• Pytorch
• torchvision
• Numpy
• Matplotlib
• Jupyter Notebook
• Pandas
• torchfile
• requests
• tensorboardx
• imageio
• cv2
• zipfile
• PIL
• h5py

WARNING: Please ensure that you have a cuda enabled GPU as all of the code in the repository has been designed to be run on it