---
license: cc
language:
- en
- zh
pretty_name: ViCaS
configs:
- config_name: default
  data_files:
  - split: train
    path: "annotations/v0.1/*.json"
---

# ViCaS: A Dataset for Combining Holistic and Pixel-level Video Understanding using Captions with Grounded Segmentation

[Project](https://ali2500.github.io/vicas-project/) | [GitHub](https://github.com/Ali2500/ViCaS) | [Arxiv](https://arxiv.org/abs/2412.09754)

For details docs and usage guide, refer to the [GitHub repo](https://github.com/Ali2500/ViCaS).