README.md

Data and code

We have released the features of all videos in the dataset:

1 Video timesformer features of NBA-ldentity datasets: https://pan.baidu.com/s/1RfS5u2z-HKqtAaHlefQd9w?pwd=ytx7

---

About our codes:

1 Supplement files: https://pan.baidu.com/s/18hGAtFZB5Ab4FhL-gmKiWA?pwd=mbuu

2 Stage_one for player identification network: player_train_timesformer.py

3 Stage_two for identity-aware sports video captioning: train_LLM_VC_1.py

Our work has been accepted by ICCV 2025

Player-Centric Multimodal Prompt Generation for Large Language Model Based Identity-Aware Basketball Video Captioning

Cite this paper:

@inproceedings{xi2025player,
  title={Player-Centric Multimodal Prompt Generation for Large Language Model Based Identity-Aware Basketball Video Captioning},
  author={Xi, Zeyu and Sun, Haoying and Wu, Yaofei and Yan, Junchi and Zhang, Haoran and Wu, Lifang and Wang, Liang and Chen, Changwen},
  booktitle={Proceedings of the IEEE/CVF International Conference on Computer Vision},
  pages={24330--24339},
  year={2025}
}

Acknowledgements

Many thanks to the code bases from Detector-free, MatchTime, Clip4Caption, and NSVA.

Contact

If you have any questions, please feel free to contact Zeyu Xi.