This paper proposes YOLO-Ball, an improved real-time detection model addressing tennis ball detection challenges including small target size, motion blur, and player/racket occlusion. The main contributions include: (1) A multi-branch occlusion-aware attention mechanism for dynamic multi-scale feature fusion, (2) A dual-flow shallow fusion pyramid combining P2 features with bidirectional fusion to enhance small target and blur handling, and (3) A dynamic balance loss integrating Normalized Gaussian Wasserstein Distance (NWD) and Intersection over Union (IoU) with learnable alignment weights. Experiments on the constructed tennis dataset show 82.2% precision and a mean Average Precision at an Intersection over Union threshold of 0.5 (mAP@0.5) of 70.9%, outperforming YOLOv8/v10/v11 by up to 12.5%. YOLO-Ball also demonstrates generalization for volleyball and football detection, thereby providing a robust detection framework for scenarios involving high-speed, small targets.
KalakeLWanWHouL. Analysis based on recent deep learning approaches applied in real-time multi-object tracking: a review. IEEE Access2021; 9: 32650–32671.
2.
ArchanaMKalaiselviMG.
An efficient ball and player detection in broadcast tennis video. In: BerrettiSThampiSMSrivastavaPR (eds) Intelligent systems technologies and applications, Vol. 1. Springer, 2016, pp.427–436.
3.
HashmiMFNaikBTKeskarAG. BDTA: events classification in table tennis sport using scaled-yolov4 framework. J Intell Fuzzy Syst2023; 44(6): 9671–9684.
4.
MesselodiSModenaCMRopeleV, et al. A low-cost computer vision system for real-time tennis analysis. In: Image analysis and processing–ICIAP 2019: 20th international conference, Trento, Italy, September 9–13, 2019, Proceedings, Part I 20. Springer, 2019, pp. 106–116.
5.
MaCXiaHXingS. Research on “AI referee assistant system” promoting the intelligent development of table tennis events. In: Proceedings of the 2025 sports social science conference (Part I), Anhui, China. Anhui Normal University, 2025, pp. 568–570.
6.
OwensNHarrisCStennettC. Hawk-eye tennis system. In: 2003 international conference on visual information engineering VIE 2003. IET, 2003, pp. 182–185.
7.
GirshickR. Fast r-cnn. In: Proceedings of the IEEE international conference on computer vision, Santiago, Chile, 2015, pp. 1440–1448.
8.
ZhaoYLvWXuS, et al. DETRS beat YOLOS on real-time object detection. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2024, pp. 16965–16974.
9.
GhahremaniAAdamsSDNortonM, et al. Detecting defects in solar panels using the yolo v10 and v11 algorithms. Electronics2025; 14(2): 344.
10.
JilongZZhaoJLiJ. Improved dangerous goods detection in x-ray images of yolov7. Comput Eng Appl2024; 60(10): 266–275.
11.
LuoXShaoBCaiZ, et al. A lightweight yolov5-ffm model for occlusion pedestrian detection. arXiv preprint arXiv:2408.06633, 13August2024.
12.
XuYLiuSWangX, et al. Lightweight online de detection method for potato surface defects based on the improved yolov8n mode. Trans Chin Soc Agric Eng2025; 41(05): 135–144.
13.
LiYMaMLiuS, et al. YOLO-Drone: a scale-aware detector for drone vision. Chin J Electron2024; 33(4): 1034–1045.
14.
HongdanYANGFuGShaoH, et al. Small object detection in aerial imagery using multi-scale hierarchical feature fusion based approach. Comput Eng Appl2025; 61(09): 230–241.
15.
XuXJiangYChenW, et al. Damo-yolo: a report on real-time object detection design. arXiv preprint arXiv:2211.15444, 2022.
16.
LiCLiLGengY, et al. Yolov6 v3.0: a full-scale reloading. arXiv preprint arXiv:2301.05586, 13January2023.
17.
MaoMHongM. Yolo object detection for real-time fabric defect inspection in the textile industry: a review of yolov1 to yolov11. Sens2025; 25(7): 2270.
18.
KangMTingCMTingFF, et al. Asf-yolo: A novel yolo model with attentional scale sequence fusion for cell instance segmentation. Image Vis Comput2024; 147: 105057.
19.
ZhouFGuoDWangY, et al. Vehicle detection algorithm based on improved yolov8 in traffic surveillance. Comput Eng Appl2024; 60(6): 110–120.
20.
TaoLILingRENBinHU, et al. Grading detection of tomato hole-pan seedlings using improved yolov5s and transfer learning. Trans Chin Soc Agric Eng2023; 39(23): 174–184.
21.
JunWChongHXinliangW. Defect detection method of insulator and grading ringbased on improved yolov7. J Optoelectron 2024: 1–14.
22.
ZhaoZ. Improvement of tomato fruit detection model based on yolov5s. Autom Instrum2025; 40(04): 109–112.
23.
KangZChaG. Research on sar ocean internal wave detection method based on improved yolo11-obb. 计算机工程与应用2024: 1–20.
24.
ChenWHuangHPengS, et al. Yolo-face: a real-time face detector. Vis Comput2021; 37: 805–813.
25.
XuCWangWYangW, et al. Detecting tiny objects in aerial images: a normalized Wasserstein distance and a new benchmark. J Photogramm Remote Sens2022; 190: 79–93.
26.
WangCHeWNieY, et al. Gold-yolo: Efficient object detector via gather-and-distribute mechanism. Adv Neural Inf Process Syst2023; 36: 51094–51112.
27.
CaiXLaiQWangY, et al. Poly kernel inception network for remote sensing detection. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition (CVPR), June 2024, pp. 27706–27716.
28.
PengYChenDZSonkaM. U-net v2: Rethinking the skip connections of u-net for medical image segmentation. In: 2025 IEEE 22nd International Symposium on Biomedical Imaging (ISBI). IEEE, 2025, pp. 1–5.
29.
XiaoJZhaoTYaoY, et al. Context augmentation and feature refinement network for tiny object detection. Remote Sens2021; 13(7): Article no. 1351.
