Fast and Efficient Non-Contact Ball Detector for Picking Robots

Object detectors based on deep learning requires high-performance computing and large run-time memory footprint to maintain good detection performance. They bring high computation overhead and power consumption to on-board embedded devices of non-contact ball picking robot. Furthermore, it is diffic...

Full description

Bibliographic Details
Main Authors: Qi-Chao Mao, Hong-Mei Sun, Yan-Bo Liu, Rui-Sheng Jia
Format: Article
Language:English
Published: IEEE 2019-01-01
Series:IEEE Access
Subjects:
Online Access:https://ieeexplore.ieee.org/document/8911435/
_version_ 1818618462215340032
author Qi-Chao Mao
Hong-Mei Sun
Yan-Bo Liu
Rui-Sheng Jia
author_facet Qi-Chao Mao
Hong-Mei Sun
Yan-Bo Liu
Rui-Sheng Jia
author_sort Qi-Chao Mao
collection DOAJ
description Object detectors based on deep learning requires high-performance computing and large run-time memory footprint to maintain good detection performance. They bring high computation overhead and power consumption to on-board embedded devices of non-contact ball picking robot. Furthermore, it is difficult to deploy on the machine because the model size is so big. The accuracy of the existing simplified detectors deployed on embedded devices cannot meet the requirements of practical applications. Therefore, how to reduce floating point operations (FLOPs) and the size of model without notably sacrificing detection precision becomes an urgent problem to be solved. To solve this problem, a shuttle residual block which is more efficient network unit based on depthwise separable convolution was proposed. And we designed a non-contact ball object detector for picking robots, which is shallower than YOLOv3 and has narrower structure. We evaluate the proposed method on non-contact Ball dataset and compelling results are achieved by the proposed method. Compared with YOLOv3, the proposed method reduces FLOPs by 86.2%, declines parameter size by 89.5%. Overall, the proposed method achieves comparable detection accuracy than YOLOv3, and its speed is 2.2 times faster than YOLOv3.
first_indexed 2024-12-16T17:21:58Z
format Article
id doaj.art-c7da34cf048f459dbfa035dd74e607d7
institution Directory Open Access Journal
issn 2169-3536
language English
last_indexed 2024-12-16T17:21:58Z
publishDate 2019-01-01
publisher IEEE
record_format Article
series IEEE Access
spelling doaj.art-c7da34cf048f459dbfa035dd74e607d72022-12-21T22:23:09ZengIEEEIEEE Access2169-35362019-01-01717548717549810.1109/ACCESS.2019.29558348911435Fast and Efficient Non-Contact Ball Detector for Picking RobotsQi-Chao Mao0https://orcid.org/0000-0003-1986-4581Hong-Mei Sun1https://orcid.org/0000-0002-4830-3364Yan-Bo Liu2https://orcid.org/0000-0001-7374-675XRui-Sheng Jia3https://orcid.org/0000-0003-1612-4764College of Computer Science and Engineering, Shandong University of Science and Technology, Qingdao, ChinaCollege of Computer Science and Engineering, Shandong University of Science and Technology, Qingdao, ChinaCollege of Computer Science and Engineering, Shandong University of Science and Technology, Qingdao, ChinaCollege of Computer Science and Engineering, Shandong University of Science and Technology, Qingdao, ChinaObject detectors based on deep learning requires high-performance computing and large run-time memory footprint to maintain good detection performance. They bring high computation overhead and power consumption to on-board embedded devices of non-contact ball picking robot. Furthermore, it is difficult to deploy on the machine because the model size is so big. The accuracy of the existing simplified detectors deployed on embedded devices cannot meet the requirements of practical applications. Therefore, how to reduce floating point operations (FLOPs) and the size of model without notably sacrificing detection precision becomes an urgent problem to be solved. To solve this problem, a shuttle residual block which is more efficient network unit based on depthwise separable convolution was proposed. And we designed a non-contact ball object detector for picking robots, which is shallower than YOLOv3 and has narrower structure. We evaluate the proposed method on non-contact Ball dataset and compelling results are achieved by the proposed method. Compared with YOLOv3, the proposed method reduces FLOPs by 86.2%, declines parameter size by 89.5%. Overall, the proposed method achieves comparable detection accuracy than YOLOv3, and its speed is 2.2 times faster than YOLOv3.https://ieeexplore.ieee.org/document/8911435/Depthwise separable convolutionpicking robotsembedded devicedeep learning
spellingShingle Qi-Chao Mao
Hong-Mei Sun
Yan-Bo Liu
Rui-Sheng Jia
Fast and Efficient Non-Contact Ball Detector for Picking Robots
IEEE Access
Depthwise separable convolution
picking robots
embedded device
deep learning
title Fast and Efficient Non-Contact Ball Detector for Picking Robots
title_full Fast and Efficient Non-Contact Ball Detector for Picking Robots
title_fullStr Fast and Efficient Non-Contact Ball Detector for Picking Robots
title_full_unstemmed Fast and Efficient Non-Contact Ball Detector for Picking Robots
title_short Fast and Efficient Non-Contact Ball Detector for Picking Robots
title_sort fast and efficient non contact ball detector for picking robots
topic Depthwise separable convolution
picking robots
embedded device
deep learning
url https://ieeexplore.ieee.org/document/8911435/
work_keys_str_mv AT qichaomao fastandefficientnoncontactballdetectorforpickingrobots
AT hongmeisun fastandefficientnoncontactballdetectorforpickingrobots
AT yanboliu fastandefficientnoncontactballdetectorforpickingrobots
AT ruishengjia fastandefficientnoncontactballdetectorforpickingrobots