Request for Google Drive/Hugging Face mirror for pretrained checkpoints

[aksharag13]

Hello authors,

Thank you for releasing the code for RAGTrack. I am trying to reproduce your work using the pretrained checkpoints for evaluation purposes.
Unfortunately, the pretrained model is currently hosted only on Baidu Netdisk, and I have been unable to access/download the files from outside China despite multiple attempts.

Would it be possible to provide an alternative download link, such as Google Drive, Hugging Face, or another international mirror, for the pretrained checkpoints?

Thank you very much for your time and for sharing your excellent work.
Best regards.

[IdolLab]

Hi @aksharag13,

Thank you for your interest in our work. We have updated the README with a Google Drive link for the pretrained checkpoints. Please check the README for the alternative download link.

Let us know if you encounter any other issues.

Best regards,
Hao Li

[aksharag13]

Hi @IdolLab,

Thank you for updating the README and providing the Google Drive links for the pretrained checkpoints.

I am currently trying to reproduce the experiments reported in the RAGTrack paper. I have obtained the publicly available RGBT datasets, but I would like to confirm whether the official versions of GTOT, RGBT210, RGBT234, and LasHeR were used directly, or if any modifications were made for the RAGTrack benchmarks.

In addition, could you please share the corresponding textual annotations (e.g., visible_description.txt and class.txt) used for GTOT and LasHeR, or provide the Google Drive links if they are already available? I would also like to confirm that the released checkpoints were trained using these same annotations.

Thank you very much for your help and for open-sourcing your work.

Best regards,
Akshara

[IdolLab]

Hi @aksharag13,

Thank you for your follow-up.

The image data we used are the official versions of GTOT, RGBT210, RGBT234, and LasHeR without any modification. The corresponding textual annotations are available on Google Drive:

🔗 https://drive.google.com/file/d/18GOOF6YKekhIwodJMI0ZQPwkRJHFuahI/view?usp=drive_link

Please feel free to reach out if you need any further assistance.

Best regards,
Hao Li

[aksharag13]

Hi @IdolLab,

Thank you for updating the README and providing the Google Drive links for the pretrained checkpoints.

As I continue working on reproducing the RAGTrack experiments, I have a few questions regarding the benchmark datasets:

1. The README mentions that GTOT, RGBT210, RGBT234, and LasHeR were used to construct the RGB-T-L benchmarks. Could you please share the official download sources (or confirm the exact versions) of the GTOT and LasHeR datasets used in your experiments? I was able to locate RGBT210 and RGBT234, but I had difficulty finding the corresponding GTOT and LasHeR datasets.

2. The Google Drive link for textual annotations currently contains folders for GTOT and LasHeR. Could you also share the textual annotations for RGBT210 and RGBT234, or clarify whether they are required for reproducing the reported results?

Thank you very much for your time and assistance.

Best regards,
Akshara

[IdolLab]

Hi @aksharag13,

Thanks for reaching out again.

Textual annotations: I just double-checked the Google Drive link. The released package contains textual annotations for all four benchmarks — GTOT, RGBT210, RGBT234, and LasHeR. Please download the entire zip file to your local machine and extract it rather than previewing online; the folders for RGBT210 and RGBT234 are included inside.

Let me know if you still cannot locate specific folders after extraction.

Best regards,
Hao Li