ThinkVG (ACM MM 2025 Dataset Track):

For images, please refer to MIMIC-CXR-JPG(https://physionet.org/content/mimic-cxr-jpg/2.1.0/). After downloading, pad the shorter side with zeros and then resize the image to 336 × 336.

If you find any incorrect data, please contact me immediately for correction. Thank you 😊!

Reference:

If you find ThinkVG useful in your research, please consider citing the following paper:

@misc{liu2025gemexthinkvg,
    title={GEMeX-ThinkVG: Towards Thinking with Visual Grounding in Medical VQA via Reinforcement Learning},
    author={Bo Liu and Xiangyu Zhao and Along He and Yidi Chen and Huazhu Fu and Xiao-Ming Wu},
    year={2025},
    eprint={2506.17939},
    archivePrefix={arXiv},
    primaryClass={cs.CV},
    url = {https://arxiv.org/abs/2506.17939}
}