Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Video

Updated: Dec 25, 2025

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications
03:31

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications

Published on: December 15, 2023

934

A Compact Convolutional Neural Network for Surface Defect Inspection.

Yibin Huang1, Congying Qiu2, Xiaonan Wang1

  • 1Institute of Automation, Chinese Academy of Sciences, University of Chinese Academy of Sciences, Beijing 100190, China.

Sensors (Basel, Switzerland)
|April 5, 2020
PubMed
Summary

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Zoledronic acid in the treatment of severe scoliosis associated with neurofibromatosis type 1 (NF1): a case report.

Annals of medicine and surgery (2012)·2026
Same author

An updated inventory of rock glaciers in the Eastern Himalaya.

Scientific data·2026
Same author

Battle of the wall for aromatic "orifice-opening" agents: defending and breaching the blood-brain barrier.

Chinese journal of natural medicines·2026
Same author

Genomic mapping reveals cisplatin disruption of protein phosphorylation signalling genome-wide.

Metallomics : integrated biometal science·2026
Same author

Harnessing Gluconobacter oxydans DSM2003 for enantioselective diol oxidation: Substrate scope, stereochemical control, and bioreactor application.

Enzyme and microbial technology·2026
Same author

Cytokines in cerebrospinal fluid combined with machine learning improve the diagnostic accuracy and predict the progression of neurosyphilis.

Frontiers in immunology·2026

We developed a compact convolutional neural network (CNN) model for surface defect inspection that runs efficiently on central processing units (CPUs). This hardware-friendly approach makes automated surface inspection more accessible in manufacturing.

Area of Science:

  • Computer Vision
  • Machine Learning
  • Manufacturing Technology

Background:

  • Convolutional Neural Networks (CNNs) have advanced computer vision but require powerful GPUs, limiting their use in manufacturing for surface defect inspection.
  • Existing CNNs are computationally intensive, hindering their deployment on less powerful hardware commonly found in industrial settings.

Purpose of the Study:

  • To develop a compact and efficient CNN-based model for tiny surface defect inspection.
  • To enable CNNs to run effectively on low-frequency CPUs, making automated surface inspection more accessible.

Main Methods:

  • Designed a lightweight (LW) bottleneck with a pyramid of lightweight kernels for rich feature extraction at reduced computational cost.
  • Developed a lightweight decoder using atrous spatial pyramid pooling (ASPP) and depthwise separable convolution layers.
Keywords:
convolutional neural networkmachine visionsurface defect inspection

More Related Videos

Combining Reflectance Confocal Microscopy with Optical Coherence Tomography for Noninvasive Diagnosis of Skin Cancers via Image Acquisition
09:37

Combining Reflectance Confocal Microscopy with Optical Coherence Tomography for Noninvasive Diagnosis of Skin Cancers via Image Acquisition

Published on: August 18, 2022

2.8K

Related Experiment Videos

Last Updated: Dec 25, 2025

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications
03:31

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications

Published on: December 15, 2023

934
Combining Reflectance Confocal Microscopy with Optical Coherence Tomography for Noninvasive Diagnosis of Skin Cancers via Image Acquisition
09:37

Combining Reflectance Confocal Microscopy with Optical Coherence Tomography for Noninvasive Diagnosis of Skin Cancers via Image Acquisition

Published on: August 18, 2022

2.8K
  • Reduced redundant weights and computation through lightweight architectural designs.

    • Main Results:

    • The compact CNN model achieved high performance on tiny defect inspection tasks.
    • The model successfully classified/segmented surface defects on an Intel i3-4010U CPU within 30 ms.
    • The model demonstrated comparable accuracy to MobileNetV2 with significantly fewer floating-point operations per second (FLOPs) and weights (1/3 FLOPs, 1/8 weights).

    Conclusions:

    • CNNs can be designed to be compact and hardware-friendly for automated surface inspection (ASI).
    • The developed model offers a practical solution for defect detection on resource-constrained hardware.
    • This research paves the way for wider adoption of AI in manufacturing quality control.