Who Is Who in Judo? Role-Aware Detection in Match Video

• Lucas Raniére J. Santos UFCG
• Everton Kauan S. Oliveira UFCG
• Everton Leandro G. Alves UFCG
• João Brunet UFCG

DOI: https://doi.org/10.5753/eniac.2026.31584

Resumo

Computer vision enables scalable solutions for sports performance analysis and competition management. We present publicly available fine-tuned YOLO models for role-aware actor detection in judo match videos, trained on amateur and professional competition data and supported by a replicable frame curation pipeline. The models distinguish referees and athletes according to white and blue judogi roles. Our best model achieves a precision of 0.834, a recall of 0.923, and an mAP50 of 0.854, outperforming a naive baseline across all metrics. These results demonstrate that domain-specific fine-tuning with role-aware annotations improves detection accuracy and robustness, supporting automated sports analytics and referee assistance applications.

Referências

Bhat, H., Gupta, V., Mundada, P., and Mali, H. (2024). Analyzing sports videos using computer vision methods. In 2024 Intelligent Systems and Machine Learning Conference (ISML), pages 63–72. IEEE.

Fujii, K. (2025). Computer Vision for Sports Analytics, pages 21–57. Springer Nature Singapore, Singapore.

Host, K. and Ivaŝić-Kos, M. (2022). An overview of human action recognition in sports based on computer vision. Heliyon, 8(6).

Kato, S. and Yamagiwa, S. (2022). Predicting successful throwing technique in judo from factors of kumite posture based on a machine-learning approach. Computation, 10(10).

Lai, C., Mo, J., Xia, H., and fang Wang, Y. (2024). FACTS: Fine-grained action classification for tactical sports.

Lin, T.-Y., Maire, M., Belongie, S., Bourdev, L., Girshick, R., Hays, J., Perona, P., Ramanan, D., Zitnick, C. L., and Dollár, P. (2015). Microsoft COCO: Common objects in context.

Mendes-Neves, T., Meireles, L., and Mendes-Moreira, J. (2023). A survey of advanced computer vision techniques for sports.

Miyaguchi, A., Moutahir, J., and Sutar, T. (2024). Annotation techniques for judo combat phase classification from tournament footage.

Nilsson, J. and Akenine-Möller, T. (2020). Understanding SSIM.

Quinn, E. and Corcoran, N. (2022). Automation of computer vision applications for real-time combat sports video analysis. In European conference on the impact of artificial intelligence and robotics, volume 4, pages 162–171.

Quinn, E. and Corcoran, N. (2024). The comparative analysis of YOLOv5/v8/v9 for object detection, tracking, and human action recognition in combat sports. In Proc. Int. Conf. on AI Research (ICAIR).

Redmon, J., Divvala, S., Girshick, R., and Farhadi, A. (2016). You only look once: Unified, real-time object detection. In 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages 779–788.

Rezatofighi, H., Tsoi, N., Gwak, J., Sadeghian, A., Reid, I., and Savarese, S. (2019). Generalized intersection over union: A metric and a loss for bounding box regression. In 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 658–666.

Salim, M., Lim, H., Salim, M., and Baharuddin, M. (2010). Motion analysis of arm movement during badminton smash. In 2010 IEEE EMBS Conference on Biomedical Engineering and Sciences (IECBES), pages 111–114.

Shariatmadar, K. and Osman, A. (2025). Ai-enhanced precision in sport taekwondo: Increasing fairness, speed, and trust in competition (fst.ai).

Tsirtsakis, P., Zacharis, G., Maraslidis, G. S., and Fragulis, G. F. (2025). Deep learning for object recognition: A comprehensive review of models and algorithms. International Journal of Cognitive Computing in Engineering, 6:298–312.