FudanCVL/Cut-VOS

---
license: apache-2.0
size_categories:
- n<1K
---

The Cut-VOS is a comple Multi-shot Video Object Segmentation (MVOS) benchmark, aims to segment the target object throughout a video sequence given the first mask prompt.
Cut-VOS contains 100 videos, 174 annotated objects, and 10.2K high-quality masks with 9 different types of shot transitions.
The target objects are accross 11 different categories, involving both actors and static objects. Please cite our work Segment Anything Across Shots:
A method and Benchmark if the dataset helps your research.

```bibtex
@inproceedings{SAAS2025,
title={{S}egment {A}nything {A}cross {S}hots: {A} Method and {B}enchmark},
author={Hu, Hengrui and Ying, Kaining and Ding, Henghui},
booktitle={AAAI},
year={2026}
}