Retrieving Any Relevant Moments: Benchmark and Models for Generalized Moment Retrieval
Paper • 2605.02623 • Published • 1
Soccer-GMR NDA Form
Please fill in the following NDA form with accurate information and agree to the non-commercial, no-redistribution, and rights-compliance terms before requesting access.
By requesting access to Soccer-GMR, you agree to use the dataset only for non-commercial research or educational purposes. You also agree not to redistribute, mirror, sell, or make the dataset publicly downloadable. The dataset may contain or reference third-party football match content,and you are responsible for complying with all applicable rights, licenses,and platform terms.
Log in or Sign Up to review the conditions and access this dataset content.
Access and NDA Terms
Soccer-GMR is available through gated access on Hugging Face. Access is manually reviewed after the requester completes the Soccer-GMR NDA form. Commercial use, redistribution, public hosting, or sharing access links is not permitted.
Soccer-GMR
This repository hosts large assets for Soccer-GMR, a dataset introduced in Retrieving Any Relevant Moments: Benchmark and Models for Generalized Moment Retrieval.
- Paper: https://arxiv.org/abs/2605.02623
- Code: https://github.com/dymm9977/generalized-moment-retrieval
- Project page: https://dymm9977.github.io/generalized-moment-retrieval/
Contents
Currently available:
| File | Description |
|---|---|
SoccerReplay-GMR.tar |
Processed Soccer-GMR video clips constructed from the SoccerReplay-1988 source data. |
SportsMoments-GMR.tar |
Processed Soccer-GMR video clips constructed from the SportsMoments source data. |
WC2022-GMR.tar |
Processed Soccer-GMR video clips constructed from the World Cup 2022 source data. |
Planned uploads:
| File / Directory | Description |
|---|---|
checkpoints/ |
Model weights released with the paper. |
Labels, evaluation code, data format documentation, and usage examples are maintained in the GitHub repository.
Note
Soccer-GMR is built by adapting soccer video sources and their timestamp/caption annotations into the Generalized Moment Retrieval format. Please follow the usage terms of the corresponding upstream sources when using the video archives.
Citation
@article{ding2026retrieving,
title={Retrieving Any Relevant Moments: Benchmark and Models for Generalized Moment Retrieval},
author={Ding, Yiming and Cao, Siyu and Jiao, Luyuan and Li, Yixuan and Wang, Zitong and Liu, Zhiyong and Zhang, Lu},
journal={arXiv preprint arXiv:2605.02623},
year={2026},
doi={10.48550/arXiv.2605.02623}
}
- Downloads last month
- 106