Laser & Optoelectronics Progress, Volume. 57, Issue 10, 101507(2020)
Improved Real-Time Vehicle Detection Method Based on YOLOV3
Figures & Tables(8)
Fig. 2. Feature maps of different sizes in the last three layers of network. (a) 52×52; (b) 26×26; (c) 13×13
Fig. 5. Model detection results in different scenarios. (a) Original images; (b) detection results of YOLOV3; (c) detection results of improved model
Table 1. Comparison of network layers and sizes of different models
View tableTable 1. Comparison of network layers and sizes of different models
Model Layer Parameter Size /MB SSD 88 27188676 103 YOLOV2 75 50983561 194 YOLOV3 256 61587112 235 Proposed 223 22314120 85.6
Table 2. Influence of different improvement strategies on mAP
View tableTable 2. Influence of different improvement strategies on mAP
Improvement strategy Inverted residuals GN SoftNMS Focal-loss Change of mAP /% -3.17 1.15 1.39 1.78
Table 3. Comparison of test results of different models
View tableTable 3. Comparison of test results of different models
Model mAP /% Time /ms SSD 89.88 48.8 YOLOV2 89.60 30.2 YOLOV3 91.91 42.3 Proposed 93.06 28.5
Tools
Get Citation
Copy Citation Text
Hanbing Li, Chunyang Xu, Chaochao Hu. Improved Real-Time Vehicle Detection Method Based on YOLOV3[J]. Laser & Optoelectronics Progress, 2020, 57(10): 101507
Paper Information
Category: Machine Vision
Received: Aug. 5, 2019
Accepted: Oct. 22, 2019
Published Online: May. 8, 2020
The Author Email: Hanbing Li (1340733996@qq.com)
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20