Spaces:
Running
on
T4
Apply for community grant: Academic project
Hi,
I'm a computer engineering student from Izmir, Turkey. Me and my teammate are making a senior project for our graduation. We have been following hugging face platform for a while since stable diffusion becomes so popular. We wanted to make an demo in spaces for our project. Our project topic is Video Inpainting using Deep Learning Models and we have researched sota models for this application. We have found SiamMask and FGT models for our application. The SiamMask model is used for tracking and segmentation parts of the selected object in the video. SiamMask gives us a masked frames for our selected objects. After the SiamMask part we are using these masked frames and our original frames to run FGT model. FGT model is used for inpainting part and gives us an object removed video. All of these processes can be made with CPU also (which are taking too much time). We are trying to debug and make tests on our project for now. If GPU access was possible, we would be really appreciated. There is a workflow for our application.
Thank you,
Sincerely.
Oğuz Akif