Target Recognition and Localization, Bounding Box Optimization of Disinfection Robot

Yaxin Ye; Jiasheng Wang; Fengyun Wu; Siyu Chen; Puye Ai; Xiangjun Zou; Lanyun Li

doi:10.3788/LOP202259.0415008

Laser & Optoelectronics Progress, Volume. 59, Issue 4, 0415008(2022)

Target Recognition and Localization, Bounding Box Optimization of Disinfection Robot

Yaxin Ye¹, Jiasheng Wang¹, Fengyun Wu¹, Siyu Chen¹, Puye Ai¹, Xiangjun Zou^1,2、*, and Lanyun Li²

Author Affiliations

¹College of Engineering, South China Agricultural University, Guangzhou , Guangdong 510642, China

²Foshan -Zhongke Innovation Research Institute of Intelligent Agriculture and Robotics, Foshan, Guangdong 528251, China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(17)

Fig. 1. Annotation effect of dataset samples. (a) Handle; (b) button of elevator; (c) button of light

Download full size

Fig. 2. Workflow of this research

Download full size

Fig. 3. Mask R-CNN architecture

Download full size

Fig. 4. Occlusion caused by different viewpoints of two cameras

Download full size

Fig. 5. Point cloud edge noise filtering algorithm based on depth statistics. (a) Instance segmentation edge error; (b) edge noise of point clouds; (c) denoising effect

Download full size

Fig. 6. Point cloud registration result. (a) Point cloud from perspective one; (b) point cloud from perspective two; (c) merged point cloud

Download full size

Fig. 7. Optimization effect of bounding box based on PCA. (a) Simple bounding box; (b) optimized bounding box; (c) optimization effect

Download full size

Fig. 8. Long arm disinfection robot and control debugging experiment scene. (a) Long arm disinfection robot; (b) control debugging experiment scene

Download full size

Fig. 9. Diagram of IoU calculation

Download full size

Fig. 10. Example instance segment result predicted by Mask R-CNN

Download full size

Fig. 11. Comparison of surface area of bounding box before and after optimization. (a) Handle; (b) button of elevator; (c) button of light

Download full size

Fig. 12. Comparison of bounding box volume before and after optimization. (a) Handle; (b) button of elevator; (c) button of light

Download full size

Table 1. Configuration list of network training parameters

View table

Table 1. Configuration list of network training parameters

Configuration parameter	Value
Image per GPU	1
Batch size	1
Step per epoch	1000
Image minimum dimension	800
Image maximum dimension	1024
Epoch for training network heads	40
Learning rate for network heads	0.001
Epoch for training ResNet	120
Learning rate for training ResNet	0.001
Epoch for training all layers	160
Learning rate for training all layers	0.0001

Table 2. Object detection result
View table
Table 2. Object detection result
Category Number of images AP mAP
Ground truth False positive True positive
Handle 91 3 86 0.945 0.968
Button of elevator 58 0 58 1.000
Button of light 77 5 77 0.997

Table 3. Mask R-CNN training results
View table
Table 3. Mask R-CNN training results
Category IoU Mean IoU
Handle 0.873 0.879
Button of elevator 0.813
Button of light 0.933

Table 4. Optimization rate of bounding box surface area for various disinfection targets
View table
Table 4. Optimization rate of bounding box surface area for various disinfection targets
Category Optimization rate Mean optimization rate
Handle 0.270 0.292
Button of elevator 0.241
Button of light 0.364

Table 5. Optimization rate of bounding box volume for various disinfection targets
View table
Table 5. Optimization rate of bounding box volume for various disinfection targets
Category Optimization rate Mean optimization rate
Handle 0.279 0.288
Button of elevator 0.245
Button of light 0.372

Tools

Get Citation

Copy Citation Text

Yaxin Ye, Jiasheng Wang, Fengyun Wu, Siyu Chen, Puye Ai, Xiangjun Zou, Lanyun Li. Target Recognition and Localization, Bounding Box Optimization of Disinfection Robot[J]. Laser & Optoelectronics Progress, 2022, 59(4): 0415008

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites