metadata
license: apache-2.0
This is a multi-class version of open vocabulary segmentation dataset by randomly merging annotations from several classes, including ADE20K(A-150), PASCAL Context59(PC-59), and PASCAL VOC20(PAS-20). You can use them according to [📂 GitHub].
If you find this project useful in your research, please consider citing:
@article{wang2025alto,
title={ALTo: Adaptive-Length Tokenizer for Autoregressive Mask Generation},
author={Wang, Lingfeng and Lin, Hualing and Chen, Senda and Wang, Tao and Cheng, Changxu and Zhong, Yangyang and Zheng, Dong and Zhao, Wuyue},
journal={arXiv preprint arXiv:2505.16495},
year={2025}
}