Cycle GAN: unsupervised video-to-video translation

1.Abstract

Research about image-to-image translation has been widely studied. The power of Cycle GAN that can generate image from unpaired dataset help researchers to jump to the advanced technology. But, research in the field of video-to video translation is still limited. Some challenges due to flickering, unsynchronization of frame time, etc. in this field need to be studied more deeply.

2.Keywords
Cycle GAN
3.Objective

We want to generate video from day to night by training some day and night images using Cycle GAN.

4.Methodology

Cycle GAN

5.Team

Irfan Dwi Bhaswara
Sriram Natarajan

6.Computation plan (required processor core hours, data storage, software, etc)

Processor: 16 core
Storage: 20 GB
Software: Python 3.7, Pytorch, Open CV

7.Source of funding
8.Target/outputs
final report that can be submitted for publication
9.Date of usage
07/02/2019 - 13/02/2019
10.Gpu usage
use gpu
11.Supporting files
12.Created at
07/02/2019
13.Approval status
approved