PMID- 32244764
OWN - NLM
STAT- PubMed-not-MEDLINE
DCOM- 20200409
LR  - 20200502
IS  - 1424-8220 (Electronic)
IS  - 1424-8220 (Linking)
VI  - 20
IP  - 7
DP  - 2020 Apr 1
TI  - A Compact Convolutional Neural Network for Surface Defect Inspection.
LID - 10.3390/s20071974 [doi]
LID - 1974
AB  - The advent of convolutional neural networks (CNNs) has accelerated the progress 
      of computer vision from many aspects. However, the majority of the existing CNNs 
      heavily rely on expensive GPUs (graphics processing units). to support large 
      computations. Therefore, CNNs have not been widely used to inspect surface 
      defects in the manufacturing field yet. In this paper, we develop a compact 
      CNN-based model that not only achieves high performance on tiny defect inspection 
      but can be run on low-frequency CPUs (central processing units). Our model 
      consists of a light-weight (LW) bottleneck and a decoder. By a pyramid of 
      lightweight kernels, the LW bottleneck provides rich features with less 
      computational cost. The decoder is also built in a lightweight way, which 
      consists of an atrous spatial pyramid pooling (ASPP) and depthwise separable 
      convolution layers. These lightweight designs reduce the redundant weights and 
      computation greatly. We train our models on groups of surface datasets. The model 
      can successfully classify/segment surface defects with an Intel i3-4010U CPU 
      within 30 ms. Our model obtains similar accuracy with MobileNetV2 while only has 
      less than its 1/3 FLOPs (floating-point operations per second) and 1/8 weights. 
      Our experiments indicate CNNs can be compact and hardware-friendly for future 
      applications in the automated surface inspection (ASI).
FAU - Huang, Yibin
AU  - Huang Y
AUID- ORCID: 0000-0001-8517-1068
AD  - Institute of Automation, Chinese Academy of Sciences, University of Chinese 
      Academy of Sciences, Beijing 100190, China.
FAU - Qiu, Congying
AU  - Qiu C
AD  - Civil Engineering & Engineering Mechanics Department, Columbia University, New 
      York, NY 10024, USA.
FAU - Wang, Xiaonan
AU  - Wang X
AD  - Institute of Automation, Chinese Academy of Sciences, University of Chinese 
      Academy of Sciences, Beijing 100190, China.
FAU - Wang, Shijun
AU  - Wang S
AD  - Institute of Automation, Chinese Academy of Sciences, University of Chinese 
      Academy of Sciences, Beijing 100190, China.
FAU - Yuan, Kui
AU  - Yuan K
AD  - Institute of Automation, Chinese Academy of Sciences, University of Chinese 
      Academy of Sciences, Beijing 100190, China.
LA  - eng
GR  - 61421004/National Natural Science Founda- ion of China/
PT  - Journal Article
DEP - 20200401
PL  - Switzerland
TA  - Sensors (Basel)
JT  - Sensors (Basel, Switzerland)
JID - 101204366
SB  - IM
PMC - PMC7180441
OTO - NOTNLM
OT  - convolutional neural network
OT  - machine vision
OT  - surface defect inspection
COIS- The authors declare no conflict of interest.
EDAT- 2020/04/05 06:00
MHDA- 2020/04/05 06:01
PMCR- 2020/04/01
CRDT- 2020/04/05 06:00
PHST- 2020/02/09 00:00 [received]
PHST- 2020/03/29 00:00 [revised]
PHST- 2020/03/29 00:00 [accepted]
PHST- 2020/04/05 06:00 [entrez]
PHST- 2020/04/05 06:00 [pubmed]
PHST- 2020/04/05 06:01 [medline]
PHST- 2020/04/01 00:00 [pmc-release]
AID - s20071974 [pii]
AID - sensors-20-01974 [pii]
AID - 10.3390/s20071974 [doi]
PST - epublish
SO  - Sensors (Basel). 2020 Apr 1;20(7):1974. doi: 10.3390/s20071974.