Laser & Optoelectronics Progress, Volume. 59, Issue 12, 1215015(2022)

Recognition of Wild Animals Using Infrared Camera Images Based on YOLOv5

Minglun Yang1,2, Xu Zhang1,2, Ying Guo1,2、*, Xinwen Yu1,2, Yanan Hou1,2, and Jiajun Gao1,2
Author Affiliations
  • 1Research Institute of Forest Resource Information Techniques, Chinese Academy of Forestry, Beijing 100091, China
  • 2Key Laboratory of Forestry Remote Sensing and Information Technology, State Forestry and Grassland Administration, Beijing 100091, China
  • show less
    Figures & Tables(11)
    Network structure diagram of YOLOv5
    Complexity of image background
    Trend of total loss of YOLOv5
    Detection effects of YOLOv3-tiny and YOLOv5m. (a) YOLOv3-tiny; (b) YOLOv5m
    Examples of detection results in complex background
    • Table 1. Depth comparison of four network structures

      View table

      Table 1. Depth comparison of four network structures

      Residual componentmodelYOLOv5sYOLOv5mYOLOv5lYOLOv5x
      CSP1_1_A1234
      CSP1_3_B36912
      CSP1_3_C36912
      CSP2_1_A1234
      CSP2_1_B1234
      CSP2_1_C1234
      CSP2_1_D1234
      CSP2_1_E1234
    • Table 2. Width comparison of four network structures

      View table

      Table 2. Width comparison of four network structures

      Filter modelYOLOv5sYOLOv5mYOLOv5lYOLOv5x
      Focus32486480
      CBL_A6496128160
      CBL_B128192256320
      CBL_C256384512640
      CBL_D51276810241280
    • Table 3. Number of images in each class

      View table

      Table 3. Number of images in each class

      ClassTrainValTestTotal
      Sika deer6507393816
      Tufted deer5385463655
      Wild boar5717472717
      Impala5677169707
      Tragopan temminckii4013240473
      Total27273043373368
    • Table 4. Comparison of experimental results of each model on the test set

      View table

      Table 4. Comparison of experimental results of each model on the test set

      ModelF1-scoremAPModel size /MBInference time /ms
      YOLOv5s0.9170.965149.5
      YOLOv5m0.910.9614212.9
      YOLOv5l0.9080.969317.9
      YOLOv5x0.9180.96517021.7
    • Table 5. Statistics of number of false and missed detections

      View table

      Table 5. Statistics of number of false and missed detections

      ClassYOLOv5sYOLOv5mYOLOv5lYOLOv5x
      Sika deer3622
      Tufted deer1291210
      Boar0002
      Impala12443
      Tragopan temminckii1131
      Empty prediction box0002
      Total28202121
    • Table 6. Comparison of YOLOv5m and YOLOv3-tiny

      View table

      Table 6. Comparison of YOLOv5m and YOLOv3-tiny

      ModelF1-scoremAPWeight /MBInference time /ms
      YOLOv5m0.910.9614212.9
      YOLOv3-tiny0.8420.94316.73.7
    Tools

    Get Citation

    Copy Citation Text

    Minglun Yang, Xu Zhang, Ying Guo, Xinwen Yu, Yanan Hou, Jiajun Gao. Recognition of Wild Animals Using Infrared Camera Images Based on YOLOv5[J]. Laser & Optoelectronics Progress, 2022, 59(12): 1215015

    Download Citation

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

    Category: Machine Vision

    Received: Jul. 28, 2021

    Accepted: Aug. 31, 2021

    Published Online: May. 23, 2022

    The Author Email: Ying Guo (guoying@ifrit.ac.cn)

    DOI:10.3788/LOP202259.1215015

    Topics