CAVIS (ICCV 2025)

CAVIS: Context-Aware Video Instance Segmentation Paper | Project

News

• Our paper has also been accepted to HiCV: 2nd Human-inspired Computer Vision at ICCV 2025. 🎉 2025.09.16
• Check out our new ICCV 2025 paper LOMM, setting a new SOTA (54.0 AP) on YouTube-VIS 2022!
• CAVIS has been accepted to ICCV 2025 🎉 2025.06.25
• We release our code with trained models 2024.12.05

Data preparation & Installation

See Datasets & Installation

Getting Started

See Getting Started with CAVIS

Model Zoo

We provide our trained models here

Citation

@article{lee2024context,
  title={Context-Aware Video Instance Segmentation},
  author={Lee, Seunghun and Seo, Jiwan and Han, Kiljoon and Choi, Minwoo and Im, Sunghoon},
  journal={arXiv preprint arXiv:2407.03010},
  year={2024}
}