LT-YOLO: An Improved Lightweight Detection Algorithm Based on YOLOv11

DOI: https://doi.org/10.62517/jbdc.202501217

Author(s)

Zhidi Cao1, Zixiang Wu1, Jiachun Li1, Dexin Zhang2, Wanwan Wang2

Affiliation(s)

1School of Artificial Intelligence and Big Data, Henan University of Technology, Zhengzhou, Henan, China 2IFLYTEK CO., LTD., Anhui, Hefei, China

Abstract

In the fields of medical image analysis and wildlife monitoring, conventional object detection algorithms often suffer from high complexity and computational overhead, making them inadequate for real-time pipeline weld defect inspection requirements. To address this challenge, we propose LT-YOLO, a lightweight object detection model built upon the YOLOv11 framework. The model incorporates a BRA module to enhance the detection capability for minute defects. Traditional convolutions in the baseline model are replaced with ADown convolutional modules to further improve recognition accuracy. A C3k2_SCConv composite module is introduced to strengthen feature representation in complex backgrounds and occluded scenarios. For model lightweighting, a Lightweight Asymmetric Multi-level Compressed Detection Head (LADH) is employed to reduce training complexity while accelerating inference. Experimental results demonstrate that the proposed model achieves 41.6% mAP@0.5:0.95 (14.3% improvement over baseline) in brain tumor detection and 76.2% mAP@0.5:0.95 in African wildlife monitoring, while reducing computational complexity by 31.3% compared to the base model. These results demonstrate that LT-YOLO maintains detection quality while fulfilling precision requirements for medical imaging and wildlife monitoring applications, achieving the model lightweighting objectives.

Keywords

LT-YOLO; YOLOv11; Lightweight; Object Detection; Dynamic Attention

References

[1]Norouzzadeh M S, Nguyen A, Kosmala M, et al. Automatically identifying, counting, and describing wild animals in camera-trap images with deep learning. Proceedings of the National Academy of Sciences, 2018, 115(25): E5716-E5725. [2]Wang C H, Huang K Y, Yao Y, et al. Lightweight deep learning: An overview. IEEE consumer electronics magazine, 2022, 13(4): 51-64. [3]Wang C H, Huang K Y, Yao Y, et al. Lightweight deep learning: An overview. IEEE consumer electronics magazine, 2022, 13(4): 51-64. [4]Howard A G, Zhu M, Chen B, et al. Mobilenets: Efficient convolutional neural networks for mobile vision applications. arXiv preprint arXiv:1704.04861, 2017.DOI: 10.1109/CVPR.2017.730. [5]Iandola F N, Han S, Moskewicz M W, et al. SqueezeNet: AlexNet-level accuracy with 50x fewer parameters and< 0.5 MB model size. arXiv preprint arXiv:1602.07360, 2016. [6]Zhu L, Wang X, Ke Z, et al. Biformer: Vision transformer with bi-level routing attention//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2023: 10323-10333. [7]Han K, Wang Y, Tian Q, et al. Ghostnet: More features from cheap operations//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2020: 1580-1589. [8]Litjens G, Kooi T, Bejnordi B E, et al. A survey on deep learning in medical image analysis. Medical image analysis, 2017, 42: 60-88. [9]Redmon J, Farhadi A. Yolov3: An incremental improvement. arXiv preprint arXiv:1804.02767, 2018. [10]Khanam R, Hussain M. Yolov11: An overview of the key architectural enhancements. arXiv preprint arXiv:2410.17725, 2024. [11]Wang C Y, Yeh I H, Mark Liao H Y. Yolov9: Learning what you want to learn using programmable gradient information//European conference on computer vision. Cham: Springer Nature Switzerland, 2024: 1-21. [12]Li J, Wen Y, He L. Scconv: Spatial and channel reconstruction convolution for feature redundancy//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2023: 6153-6162. [13]Xu Y, Lu J, Wang C. YOLO-SOD: Improved YOLO Small Object Detection//Pacific Rim International Conference on Artificial Intelligence. Singapore: Springer Nature Singapore, 2024: 164-176. [14]Zhu L, Wang X, Ke Z, et al. Biformer: Vision transformer with bi-level routing attention//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2023: 10323-10333. [15]Zhang J, Chen Z, Yan G, et al. Faster and lightweight: an improved YOLOv5 object detector for remote sensing images. Remote Sensing, 2023, 15(20): 497.