Datasets:

Modalities:
Image
ArXiv:
Libraries:
Datasets
License:
Mask_Perception / README.md
yayafengzi's picture
Update README.md
6300e46 verified
metadata
license: apache-2.0

This is a 10K test dataset for Mask Perception, a mask-to-text task which aims for fine-grained understanding. It is built based on RefCOCO/+/g. You can use it in [📂 GitHub].

If you find this project useful in your research, please consider citing:

@article{wang2025himtok,
  title={HiMTok: Learning Hierarchical Mask Tokens for Image Segmentation with Large Multimodal Model},
  author={Wang, Tao and Cheng, Changxu and Wang, Lingfeng and Chen, Senda and Zhao, Wuyue},
  journal={arXiv preprint arXiv:2503.13026},
  year={2025}
}