Detection and Tracking of Low-Altitude Unmanned Aerial Vehicles Based on Optimized YOLOv4 Algorithm

Yuemeng Zhao; Huigang Liu

doi:10.3788/LOP202259.1215017

Laser & Optoelectronics Progress, Volume. 59, Issue 12, 1215017(2022)

Detection and Tracking of Low-Altitude Unmanned Aerial Vehicles Based on Optimized YOLOv4 Algorithm

Yuemeng Zhao^1,2 and Huigang Liu^1,2、*

¹Engineering Research Center of Thin Film Optoelectronics Technology, Ministry of Education, College of Electronic Information and Optical Engineering, Nankai University, Tianjin 300350, China

²Tianjin Key Laboratory of Optoelectronic Sensor and Sensing Network Technology, Tianjin 300350, China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(12)

Fig. 1. Schematic diagram of YOLOv4 structure

Download full size

Fig. 2. Comparsion between Darknet-53 and CSPDarknet-53. (a) Darknet-53; (b) CSPDarknet-53

Download full size

Fig. 3. Cascade matching process

Download full size

Fig. 4. K-means clustering analysis result

Download full size

Fig. 5. Optimized YOLOv4 network structure

Download full size

Fig. 6. Detection and tracking model based on a combination of optimized YOLOv4 and DeepSORT

Download full size

Fig. 7. Loss function curve

Download full size

Fig. 8. Test results. (a) YOLOv4; (b) optimized YOLOv4

Download full size

Fig. 9. Visual detection effects of small targets. (a) YOLOv3; (b) YOLOv4; (c) optimized YOLOv4

Download full size

Fig. 10. Visual tracking performance of optimized YOLOv4-DeepSORT method

Download full size

Table 1. Value of anchor in LARotorcraft dataset

View table

Table 1. Value of anchor in LARotorcraft dataset

Feature Level	Size of feature map	Value of anchor
Feature Level 1	19×19	（485，493），（495，279）
Feature Level 2	38×38	（128，110），（241，186）
Feature Level 3	76×76	（36，75），（76，55）
Feature Level 4	152×152	（12，16），（19，36）

Table 2. Performance comparison of different detection algorithms
View table
Table 2. Performance comparison of different detection algorithms
Detection algorithm AP /% AP50 /% AP75 /% Detection speed /（frame·s^-1）
YOLOv3 69 77.1 41.5 8.9
YOLOv4 75.6 93.2 86.5 12.7
Optimized YOLOv4 77.2 97.1 90.1 10.8

Tools

Get Citation

Copy Citation Text

Yuemeng Zhao, Huigang Liu. Detection and Tracking of Low-Altitude Unmanned Aerial Vehicles Based on Optimized YOLOv4 Algorithm[J]. Laser & Optoelectronics Progress, 2022, 59(12): 1215017

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites