https://github.com/y-t-g/blur-anything
Track and blur any object or person in a video.
https://github.com/y-t-g/blur-anything
blurring mobilesam segment-anything track-anything xmem
Last synced: over 1 year ago
JSON representation
Track and blur any object or person in a video.
- Host: GitHub
- URL: https://github.com/y-t-g/blur-anything
- Owner: Y-T-G
- License: mit
- Created: 2023-05-05T15:13:06.000Z (about 3 years ago)
- Default Branch: main
- Last Pushed: 2024-02-10T07:53:22.000Z (over 2 years ago)
- Last Synced: 2025-03-19T00:38:54.163Z (over 1 year ago)
- Topics: blurring, mobilesam, segment-anything, track-anything, xmem
- Language: Python
- Homepage:
- Size: 163 KB
- Stars: 10
- Watchers: 2
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- Changelog: CHANGELOG.md
- License: LICENSE
Awesome Lists containing this project
README
# Blur Anything For Videos
Blur Anything is an adaptation of the excellent [Track Anything](https://github.com/gaomingqi/Track-Anything) project which is in turn based on Meta's Segment Anything and XMem. It allows you to blur anything in a video, including faces, license plates, etc.
## Get Started
```shell
# Clone the repository:
git clone https://github.com/Y-T-G/Blur-Anything.git
cd Blur-Anything
# Install dependencies:
pip install -r requirements.txt
# Run the Blur-Anything gradio demo.
python app.py --device [cpu|cuda:0|cuda:1|...] --sam_model_type [vit_t| vit_b|vit_h| vit_l] [--backend [onnx|openvino]]
```
## Features
- FastSAM with ONNX and OpenVINO support.
- Lower memory usage.
## To Do
- [x] Add a gradio demo
- [ ] Add support to use YouTube video URL
- [ ] Add option to completely black out the object
- [ ] Convert XMem to ONNX
## Acknowledgements
The project is an adaptation of [Track Anything](https://github.com/gaomingqi/Track-Anything) which is based on [Segment Anything](https://github.com/facebookresearch/segment-anything) and [XMem](https://github.com/hkchengrex/XMem).
Thanks to [PIMS](https://github.com/soft-matter/pims) which is used to process video files while keeping memory usage low.