Skip to main navigation Skip to main content
  • E-Submission

JKSPE : Journal of the Korean Society for Precision Engineering

OPEN ACCESS
ABOUT
BROWSE ARTICLES
EDITORIAL POLICIES
FOR CONTRIBUTORS
REGULAR

영상기반 파지 동작 제어를 위한 딥러닝 기반 물체검출과 파지물체 선정

Deep Learning-Based Object Detection and Target Selection for Image-Based Grasping Motion Control

Journal of the Korean Society for Precision Engineering 2020;37(5):389-394.
Published online: May 1, 2020

1 한국기계연구원 의료기계연구실

2 경북대학교 전자공학부

1 Medical Device Laboratory, Korea Institute of Machinery & Materials

2 School of Electronics Engineering, Kyungpook National University

#E-mail: minyoung.kim2@gmail.com, TEL: +82-53-950-7233, E-mail: jhseo@kimm.re.kr, TEL: +82-53-670-9103
• Received: December 5, 2019   • Revised: February 13, 2020   • Accepted: February 25, 2020

Copyright © The Korean Society for Precision Engineering

This is an Open-Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/3.0) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.

  • 7 Views
  • 0 Download
  • 1 Crossref
  • 1 Scopus
prev next

Citations

Citations to this article as recorded by  Crossref logo
  • A Study on Defect Detection Model of Bone Plates Using Multiple Filter CNN of Parallel Structure
    Song Yeon Lee, Yong Jeong Huh
    Journal of the Korean Society for Precision Engineering.2023; 40(9): 677.     CrossRef

Download Citation

Download a citation file in RIS format that can be imported by all major citation management software, including EndNote, ProCite, RefWorks, and Reference Manager.

Format:

Include:

Deep Learning-Based Object Detection and Target Selection for Image-Based Grasping Motion Control
J. Korean Soc. Precis. Eng.. 2020;37(5):389-394.   Published online May 1, 2020
Download Citation

Download a citation file in RIS format that can be imported by all major citation management software, including EndNote, ProCite, RefWorks, and Reference Manager.

Format:
Include:
Deep Learning-Based Object Detection and Target Selection for Image-Based Grasping Motion Control
J. Korean Soc. Precis. Eng.. 2020;37(5):389-394.   Published online May 1, 2020
Close

Figure

  • 0
  • 1
  • 2
  • 3
  • 4
  • 5
  • 6
Deep Learning-Based Object Detection and Target Selection for Image-Based Grasping Motion Control
Image Image Image Image Image Image Image
Fig. 1 Schematic diagram of grasping motion control of robotic prosthetic hand
Fig. 2 Prediction process of YOLO algorithm
Fig. 3 Two grasping types and representative objects
Fig. 4 Training images of two classes grasping
Fig. 5 Results of the object detection test
Fig. 6 Object priority selection method for grasping control
Fig. 7 Selection test of the grasping target
Deep Learning-Based Object Detection and Target Selection for Image-Based Grasping Motion Control

Accuracy and operating speed for object detection algorithm

Algorithm mAP [%] Speed [ms]
Inception SSD 52 40
Mobilenet SSD 53 29
YOLOv3 80 21

mAP (mean Average Precision)

False positive (FP) and true positive (TP) for object detection algorithm

Algorithm FP TP Sum
Inception
SSD
Grasp: 47
Pinch: 2
Sum: 49
Grasp: 12
Pinch: 39
Sum: 51
Grasp: 59
Pinch: 41
Total: 100
Mobilenet
SSD
Grasp: 49
Pinch: 0
Sum: 49
Grasp: 12
Pinch: 39
Sum: 51
Grasp: 61
Pinch: 39
Total: 100
YOLOv3 Grasp: 21
Pinch: 2
Sum: 23
Grasp: 38
Pinch: 39
Sum: 77
Grasp: 59
Pinch: 41
Total: 100

Probability of detected objects in case of Figs. 7(a) and 7(b)

Object
number
Percentage
[%]
Object
number
Percentage
[%]
a-1 0.4 b-1 6.5
a-2 0.2 b-2 8.7
a-3 12.1 b-3 0.1
a-4 0.2 b-4 0.3
a-5 1.0
Table 1 Accuracy and operating speed for object detection algorithm

mAP (mean Average Precision)

Table 2 False positive (FP) and true positive (TP) for object detection algorithm
Table 3 Probability of detected objects in case of Figs. 7(a) and 7(b)