A webui for Propainter

简体中文
A webui that you can easily pick up objects from the video and eliminate them.

Demo

Getting started

Fork: english UI.
If you don't want to install the environment, you can download the package directly.
link 百度网盘
tutorial bilibili youtube

clone repo

git clone https://github.com/Katehuuh/ProPainter-Webui.git && cd ProPainter-Webui

create enviroment

python -m venv venv && venv\Scripts\activate

install dependencies

Just follow the instructions in Segment-ant-Track-Anything and ProPainter（P.S.Don't need to install groundingdino, I have put it in the project.）

Install PyTorch and follow the instructions, click to see more details

pip3 install -r ProPainter\requirements.txt
pip3 install -e ./sam
pip install -r requirements.txt

# Optional
git clone https://github.com/ClementPinard/Pytorch-Correlation-extension.git && cd Pytorch-Correlation-extension
python setup.py install
cd ..

prepare pretrained models

Download all the needed models for propainter, Models auto-download on first run.
propainter
raft-things
recurrent_flow_completion
i3d_rgb_imagenet

Download all the needed models for segment-and-track-anything
SAM-VIT-B (sam_vit_b_01ec64.pth)
R50-DeAOT-L (R50_DeAOTL_PRE_YTB_DAV.pth)
GroundingDINO-T (groundingdino_swint_ogc)\

The directory structure will be arranged as：

ckpt
   |- bert-base-uncased
   |- groundingdino_swint_ogc.pth
   |- R50_EdAOTL_PRE_YTB_DAV.pth
   |- sam_vit_b_01ec64.pth
...
ProPainter/weights
   |- ProPainter.pth
   |- recurrent_flow_completion.pth
   |- raft-things.pth
   |- i3d_rgb_imagenet.pt (for evaluating VFID metric)
   |- README.md

quick start

python app.py

Reference

• Segment-ant-Track-Anything
• ProPainter

Star History