Skip to content

UniRef++

A unified model for image and video object segmentation

Visit official site

What UniRef is

The model is described as a unified approach to referring object segmentation across images and video, built around the UniFusion module that injects reference information into a base network.

What you can do with it

  • Carry out referring image segmentation
  • Handle few-shot segmentation tasks
  • Segment objects across video frames
  • Add semantic segmentation to a base model

Model notes

  • RIS, FSS, RVOS and VOS are named tasks
  • Models trained on benchmark datasets are provided
  • The code is open for research use

Pros & cons

✓ What we like

  • Works as a plug-in with base models such as SAM
  • Several segmentation tasks are supported

! What to watch out for

  • The listed tasks centre on object segmentation
  • Use as a plug-in depends on a base model

FAQ

Which tasks are supported?

RIS, FSS, RVOS and VOS are listed.

Is code available?

The material states pre-trained models and open source code for research.

Last reviewed: 2026-11-15

More AI video generation tools

View all →

How we review