Research on Dense Small Target Vehicle Detection Based on Thin Neck Structure —Taking LSGC-YOLOv8 as an Example
(School of Mathematics and Physics, Lanzhou Jiaotong University, Lanzhou 730070, China)
Abstract
The traditional monitoring system relies on manual identification of targets, which is inefficient and easy to misjudge. This paper improves and researches the series of YOLO algorithms. Although the YOLO algorithm has been widely used in many fields, it still faces three challenges in actual traffic scenarios. Weather changes and hardware limitations can easily lead to a decline in data quality. The detection ability of narrow area target, occlusion target and incomplete target is insufficient. In addition, in complex scenes, detail feature extraction has limitations. To this end, this paper proposes an improved model that integrates lightweight design and structural optimization. The model contains three innovations. Firstly, the backbone network introduces a lightweight architecture, which significantly reduces the computational complexity under the premise of controllable accuracy loss. Secondly, the detection head is reconstructed by VoVGSCSP module to enhance the ability of fine-grained feature capture. Thirdly, the integrated Attention Mechanism improves the quality of feature fusion. The experimental results show that the improved model achieves the balance optimization of accuracy and efficiency in complex traffic scenarios, and has application value.
Keywords
Target Detection; intelligent transportation; image recognition; Deep Learning; vehicle detection
References
- [1] Li Gang, Cai Zehao, Sun Huaxun, etc Research on Hardware Defect Detection Method Based on Improved YOLOv8 and Semantic Knowledge Fusion [J]. Journal of Graphic Science, 2024, 45 (5): 979-986
- [2] Zhou Fei, Guo Dudu, Wang Yang, etc Traffic monitoring vehicle detection algorithm based on improved YOLOv8 [J]. Computer Engineering and Applications, 2024, 60 (6): 110-120
- [3] Zhao Wenqing, Cheng Xingfu, Zhao Zhenbing, etc Insulator recognition combining attention mechanism and Faster RCNN [J]. Journal of Intelligent Systems, 2020, 15 (1): 92-98
- [4] Mao Zhihui, Zhu Jiali, Wu Xin, etc A review of research on object detection in autonomous driving based on YOLO [J]. Computer Engineering and Applications, 2022, 58 (15): 68-77
- [5] Deng Yaping, Li Yingjiang Overview of YOLO Algorithm and Its Application in Object Detection in Autonomous Driving Scenarios [J]. Computer Applications, 2024, 44 (6): 1949-1958
- [6] LIU J,WANG X W. Early Recognition of Tomato Gray Leaf Spot Disease Based on MobileNetv2-YOLOv3 Model [J].Plant Methods,2020,16(1):1-16.
- [7] HE X W,CHENG R,ZHENG Z L,et al. Small Object Detection in Traffic Scenes Based on YOLO-MXANet [J/OL].Sensors,2021,21(21):7422[2024-01-18].https://doi.org/10.3390/s21217422.
- [8] ZHU F Z,WANG Y Y,CUI J Y,et al. Target Detection for Remote Sensing Based on the Enhanced YOLOv4 with Improved BiFPN [J].The Egyptian Journal of Remote Sensing and Space Science,2023,26(2):351-360.
- [9] ZHANG S,CHE S B,LIU Z,et al. A Real-Time and Lightweight Traffic Sign Detection Method Based on Ghost-YOLO [J].Multimedia Tools and Applications,2023,82(17):26063-26087.
- [10] WANG G,CHEN Y F,AN P,et al. UAV-YOLOv8: A Small-Object-Detection Model Based on Improved YOLOv8 for UAV Aerial Photography Scenarios [J/OL].Sensors,2023,23(16):7190[2025-01-18].https://www.mdpi.com/1424-8220/23/16/7190.
- [11] ZHAO Y A,LV W Y,XU S L,et al. Detrs Beat YOLOs on Real-time Object Detection [C]//2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition.Seattle:IEEE,2024:16965-16974.
- [12] Hu Yuheng, Wu Jin Improvement of YOLO-DETR fabric surface micro damage detection method [J]. Modern Electronic Technology, 2024, 47 (13): 160-163
- [13] Wang Yaru, Feng Lilong, Song Xiaoke, etc TFD-YOLOv8: A Foreign Object Detection Method for Transmission Lines [J]. Journal of Graphic Science, 2024, 45 (5): 901-912
- [14] Zhu Qiangjun, Hu Bin, Wang Huilan, etc Detection of traffic signs based on lightweight YOLOv8s [J]. Journal of Graphic Science, 2024, 45 (3): 422-432
- [15] Qu Haicheng, Wang Meng, Chai Rui Efficient remote sensing image vehicle detection using bidirectional multi-scale feature fusion [J]. Computer Engineering and Applications, 2024, 60 (12): 346-356
- [16] LI H L,LI J,WEI H B,et al. Slim-Neck by GSConv: A Better Design Paradigm of Detector Architectures for Autonomous Vehicles [J/OL].arXiv:2206.02424 [cs.CV].[2025-01-18].https://arxiv.org/abs/2206.02424v2.
- [17] WOO S,PARK J,LEE J Y,et al. CBAM: Convolutional Block ATtention Module [C]//15th European Conference.Munich:Springer,2018:3-19.
- [18] ZHU X Z,HU H,LIN S,et al. Deformable ConvNets V2: More Deformable, Better Results [C]//2019 IEEE/CVF conference on computer vision and pattern recognition.Long Beach:IEEE,2019:9300-9308.
- [19] LI Y L,TANG X L,LIU W S,et al. An Improved Method for Detecting Crane Wheel–Rail Faults Based on YOLOv8 and the Swin Transformer [J/OL].Sensors,2024,24(13):4086[2025-01-18].https://www.mdpi.com/1424-8220/24/13/4086.
- [20] RAO Y M,ZHAO W L,TANG Y S,et al. Hornet: Efficient High-Order Spatial Interactions with Recursive Gated Convolutions [J/OL].arXiv:2207.14284 [cs.CV].[2025-01-18].https://arxiv.org/abs/2207.14284.
- [21] HU J,SHEN L,SAMUEL A,et al. Squeeze-and-Excitation Networks [J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2020:2011-2023.
- [22] LI X,ZHONG Z,WU J,et al. Expectation-Maximization Attention Networks for Semantic Segmentation [C]//2019 IEEE/CVF international conference on computer vision.Seoul:IEEE,2019:9166-9175.
- [23] GAO P,LU J S,LI H S,et al. Container: Context Aggregation Network [J/OL].arXiv:2106.01401 [cs.CV].[2025-01-18].https://arxiv.org/abs/2106.01401.
Funding
Key Research and Development Project for Provincial Ecological Civilization Construction in Gansu Province (24YFFA055)
About the authors
Zhou Zijian (1998-), male, Han ethnicity, from Ningbo, Zhejiang, is a master's student with research interests in image processing and pattern recognition; Corresponding author: Liu Haizhong (1969-), male, Han ethnicity, from Lintao, Gansu Province, associate professor, master's degree, research direction: data science, spatiotemporal prediction and decision-making.