Laser & Optoelectronics Progress, Volume. 62, Issue 2, 0215005(2025)

Electric Tricycle Detection Based on Improved YOLOv5s Model

Xiaofang Ou*, Fengchun Han, Jing Tian, Jijie Tang, and Zhengtao Yang
Author Affiliations
  • School of Traffic Management, People's Public Security University of China, Beijing 100038, China
  • show less
    Figures & Tables(14)
    YOLOv5s model structural diagram
    Mosaic data augmentation
    SPPF module structure diagram
    PANet fusion structure diagram
    Improved YOLOv5s structure diagram
    CTR 3 encoder
    Efficient additive attention mechanism
    Dataset images. (a) Crossroad; (b) T-shaped intersection; (c) main road; (d) expressway
    Data enhanced images
    Comparison chart between loss and mAP@0.5 change. (a) Loss; (b) mAP@0.5
    FPS-mAP@0.5 coordinate diagram
    Comparison of detection results diagram. (a) Original model; (b) proposed model
    • Table 1. Ablation experiments

      View table

      Table 1. Ablation experiments

      ModulePrecision /%Recall /%mAP@0.5 /%FPSSize
      YOLOv5s90.5482.6784.686814.36
      YOLOv5s+New head89.8783.7986.97(+2.29)6615.75
      YOLOv5s+Transformer88.7483.5985.64(+0.96)7015.79
      YOLOv5s+EAA88.3183.2485.59(+0.91)9614.27
      YOLOv5s+New head+Transformer+EAA91.1584.8889.58(+4.90)9216.14
    • Table 2. Comparative experiment

      View table

      Table 2. Comparative experiment

      ModelPrecision /%Recall /%mAP@0.5/%FPS /(frame/s)Size /Mbit
      SSD83.5682.3984.532856.74
      Faster-RCNN85.4983.4586.491341.87
      YOLOv3-tiny79.5868.7476.7212116.59
      YOLOv8s91.3285.7390.437421.44
      YOLOv5s90.5482.6784.686814.36
      Proposed91.1584.8889.589216.14
    Tools

    Get Citation

    Copy Citation Text

    Xiaofang Ou, Fengchun Han, Jing Tian, Jijie Tang, Zhengtao Yang. Electric Tricycle Detection Based on Improved YOLOv5s Model[J]. Laser & Optoelectronics Progress, 2025, 62(2): 0215005

    Download Citation

    EndNote(RIS)BibTexPlain Text
    Save article for my favorites
    Paper Information

    Category: Machine Vision

    Received: Apr. 9, 2024

    Accepted: Jun. 3, 2024

    Published Online: Jan. 6, 2025

    The Author Email:

    DOI:10.3788/LOP241065

    CSTR:32186.14.LOP241065

    Topics