hkchengrex/MiVOS
A CVPR 2021 research project for interactive video object segmentation combining user interactions with mask propagation and difference-aware fusion.
Not currently ranked — collecting fresh signals.
star history
MiVOS implements modular interactive video object segmentation, allowing users to provide sparse interactions that are converted to masks and then propagated across video frames. The system combines an interaction-to-mask module, a mask propagation module, and a difference-aware fusion module. It supports both semi-supervised VOS evaluation on DAVIS/YouTube and interactive evaluation with a user GUI. The project is implemented in Python using PyTorch.
Frequently asked
- What is hkchengrex/MiVOS?
- A CVPR 2021 research project for interactive video object segmentation combining user interactions with mask propagation and difference-aware fusion.
- Is MiVOS open source?
- Yes — hkchengrex/MiVOS is open source, released under the MIT license.
- What language is MiVOS written in?
- hkchengrex/MiVOS is primarily written in Python.
- How popular is MiVOS?
- hkchengrex/MiVOS has 488 stars on GitHub.
- Where can I find MiVOS?
- hkchengrex/MiVOS is on GitHub at https://github.com/hkchengrex/MiVOS.