Researching | Vehicle detection based on FVOIRGAN-Detection

Optics and Precision Engineering, Volume. 30, Issue 12, 1478(2022)

Vehicle detection based on FVOIRGAN-Detection

Hao ZHANG^1,2,3,4, Jianhua YANG^1,2,3,4, and Haiyang HUA^1,2、*

Author Affiliations

¹Key Laboratory of Opto-Electronic Information Processing，Chinese Academy of Sciences， Shenyang006， China

²Shenyang Institute of Automation， Chinese Academy of Sciences， Shenyang110016， China

³Institutes for Robotics and Intelligent Manufacturing， Chinese Academy of Sciences， Shenyang110169， China

⁴University of Chinese Academy of Sciences， Beijing10009， China

show less

Abstract Get PDF(in Chinese)

Figures&Tables (9)

References (26)

Paper Information

Figures & Tables(9)

Fig. 1. Network architecture of CrossGAN-Detection

Download full size

View in Article

Fig. 2. Network architecture of point cloud processing based on FVOI

Download full size

View in Article

Fig. 3. Schematic diagram of working principle of relative probability

Download full size

View in Article

Fig. 4. Relationship between AP and IOU on KITTI validation set

Download full size

View in Article

Fig. 5. Comparison example of target detection results in challenging scenes

Download full size

View in Article

Table 1. Ablation analysis of 2D Car detection performance on KITTI verification set： average accuracy
View table
View in Article
Table 1. Ablation analysis of 2D Car detection performance on KITTI verification set： average accuracy
Method FVOI Relative Easy（ $I O U_{0.7}$ ） Moderate（ $I O U_{0.7}$ ） Hard（ $I O U_{0.7}$ ）
Baseline 96.66% 87.15% 78.46%
+FVOI $\sqrt$ 97.04% 87.21% 78.52%
+Relative $\sqrt$ 97.64% 87.83% 79.01%
Ours $\sqrt$ $\sqrt$ 97.67% 87.86% 79.03%

Table 2. Ablation analysis of 2D Car detection performance on KITTI challenging scenarios： average accuracy
View table
View in Article
Table 2. Ablation analysis of 2D Car detection performance on KITTI challenging scenarios： average accuracy
Method AP（ $I O U_{0.5}$ ）
val promote
Cross fusion 86.12% 0
+FVOI 88.04% 1.92%
+relative 83.55% -2.57%
Ours 88.49% 2.37%

Table 3. Comparison of texture information extraction performance： correlation eigenvalue （COR）
View table
View in Article
Table 3. Comparison of texture information extraction performance： correlation eigenvalue （COR）
Method COR
0° 45° 90° 135° Mean
RGB 0.074 0 0.071 9 0.073 1 0.072 0 0.072 7
CrossGAN-Detection 0.069 1 0.067 9 0.068 6 0.067 9 0.068 4
Ours 0.070 7 0.069 6 0.070 3 0.069 5 0.070 0

Table 4. Comparison of 2D Car detection performance on KITTI verification set： average accuracy

View in Article

Table 4. Comparison of 2D Car detection performance on KITTI verification set： average accuracy

Approach	Easy （ $I O U_{0.7}$ ）	Moderate （ $I O U_{0.7}$ ）	Hard （ $I O U_{0.7}$ ）
Mono3D^［24］	93.89%	88.67%	79.68%
3DOP^［25］	93.08%	88.07%	79.39%
M3D-RPN^［26］	90.24%	83.67%	67.69%
MV3D^［10］	95.01%	87.59%	79.90%
Proposed Method	97.67%	87.86%	79.03%

Tools

Get Citation

Copy Citation Text

Hao ZHANG, Jianhua YANG, Haiyang HUA. Vehicle detection based on FVOIRGAN-Detection[J]. Optics and Precision Engineering, 2022, 30(12): 1478

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites

Paper Information

Category: Information Sciences

Received: Dec. 14, 2021

Accepted: --

Published Online: Jul. 5, 2022

The Author Email: HUA Haiyang (c3i11@sia.cn)

DOI:10.37188/OPE.20223012.1478

Topics

laser devices and laser physics

Lasers and Laser Optics

laser manufacturing

Instrumentation, Measurement and Metrology