search

sczhou / ProPainter

[ICCV 2023] ProPainter: Improving Propagation and Transformer for Video Inpainting
https://shangchenzhou.com/projects/ProPainter/
Other
5.46k stars 645 forks source link

How to inference on my own data? #31

Open mahimairaja opened 11 months ago

mahimairaja commented 11 months ago

1. I could see from the example video running_car, you are using a single image as mask. In that case, video completion is happening using RAFT, am I right?

I could see that ProPainter could accomplish:

  1. Video Completion
  2. Object Removal
  3. Video Restoration
  4. Watermark and Logo Removal

In the inference code, I could see only two modes: video_inpainting, video_outpainting

2. What is the difference between these two modes?

3. How can I perform Object Removal with my own video data?

Thanks to the team and NTU for such a wonderful project!

Kudos 🥇