• Infrared and Laser Engineering
  • Vol. 51, Issue 4, 20210290 (2022)
Xiaofeng Zhao, Yebin Xu, Fei Wu, Jiahui Niu, Wei Cai, and Zhili Zhang
Author Affiliations
  • Armament Launch Theory and Technology Key Discipline Laboratory of China, Rocket Force University of Engineering, Xi'an 710025, China
  • show less
    DOI: 10.3788/IRLA20210290 Cite this Article
    Xiaofeng Zhao, Yebin Xu, Fei Wu, Jiahui Niu, Wei Cai, Zhili Zhang. Ground infrared target detection method based on a parallel attention mechanism (Invited)[J]. Infrared and Laser Engineering, 2022, 51(4): 20210290 Copy Citation Text show less
    Downsampling channel conversion process
    Fig. 1. Downsampling channel conversion process
    SE attention mechanism
    Fig. 2. SE attention mechanism
    Convolution module based on parallel attention mechanism
    Fig. 3. Convolution module based on parallel attention mechanism
    PAM-YOLO target detection model
    Fig. 4. PAM-YOLO target detection model
    Loss curve
    Fig. 5. Loss curve
    Comparison chart of target detection results. (a) Original image; (b)YOLOv3; (c) PAM-YOLO
    Fig. 6. Comparison chart of target detection results. (a) Original image; (b)YOLOv3; (c) PAM-YOLO
    ModelGFLOPSFPS/frame·s-1mAP@0.5
    Faster R-CNN224.56.577.1%
    SSD5001373872.6%
    YOLOv3155.148.275.3%
    PAM-YOLO50.64682.2%
    Table 1. Experimental results of different target detection models based on the Infrared-VOC dataset
    YOLOv3 CIoU loss and focal loss PAConv PAM-YOLO GFLOPS Training time/h mAP@0.5AP@0.5
    CarBus
    155.12.1275.3%72.9%77.6%
    155.12.1077.1%78.2%76.0%
    151.31.4878.4%80.1%76.8%
    151.31.4880.4%80.7%80.1%
    50.60.92782.2%81.4%82.9%
    Table 2. Ablation experiment based on Infrared-VOC dataset
    YOLOv3 with lossSE modulePAM moduleGFLOPSTraining time/hParameters/MmAP@0.5
    155.12.1059.6677.1%
    206.33.1459.6778.0%
    151.31.4856.8680.4%
    Table 3. Comparison test of SE module and PAM module
    Xiaofeng Zhao, Yebin Xu, Fei Wu, Jiahui Niu, Wei Cai, Zhili Zhang. Ground infrared target detection method based on a parallel attention mechanism (Invited)[J]. Infrared and Laser Engineering, 2022, 51(4): 20210290
    Download Citation