Surface defect detection of rollers based on improved DeepLabv3+ and migration learning

Ruixin Dong; Shigang Xiao

doi:10.1117/12.3039512

4 September 2024 Surface defect detection of rollers based on improved DeepLabv3+ and migration learning

Ruixin Dong, Shigang Xiao

Author Affiliations +

Proceedings Volume 13259, International Conference on Automation Control, Algorithm, and Intelligent Bionics (ACAIB 2024); 132593E (2024) https://doi.org/10.1117/12.3039512
Event: Fourth International Conference on Automation Control, Algorithm, and Intelligent Bionics (ICAIB 2024), 2024, Yinchuan, China

Abstract

Aiming at the status quo that the defect detection method on the surface of the rollers is less efficient and unable to detect abnormal faults quickly in real time, a defect detection algorithm on the surface of the rollers based on improved DeepLabv3+ and migration learning is proposed. First, a lightweight MobileNetv2 network is used to replace the original backbone network Xception of DeepLabv3+ algorithm in order to reduce the number of model parameters and improve the inference speed; then the strategy of combining heterogeneous sensory field fusion and void depth separable convolution is used to improve the void space pyramid pooling structure to improve the information utilization and training efficiency of the model; finally, the attention mechanism is introduced to improve the performance of the model by using data augmentation and migration learning to improve the network training effect and enhance the accuracy of defect recognition on the surface of the rollers. The experimental results show that the improved DeepLabv3+ model has fewer iteration steps and faster training speed than the original model, the MIoU value has been improved by 3.46%, the number of model parameters has been compressed by 69.8% of the original model, and the segmentation effect on the defect edge has been improved significantly, which can complete the defect detection on the surface of the rollers in a fast and realtime manner, and has certain reference value and application significance.

(2024) Published by SPIE. Downloading of the abstract is permitted for personal use only.

Citation Download Citation

Ruixin Dong and Shigang Xiao "Surface defect detection of rollers based on improved DeepLabv3+ and migration learning", Proc. SPIE 13259, International Conference on Automation Control, Algorithm, and Intelligent Bionics (ACAIB 2024), 132593E (4 September 2024); https://doi.org/10.1117/12.3039512

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

;

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available

Members: $17.00

Non-members: $21.00 ADD TO CART

PROCEEDINGS
6 PAGES

DOWNLOAD PAPER SAVE TO MY LIBRARY

GET CITATION

RIGHTS & PERMISSIONS

Get copyright permission Get copyright permission on Copyright Marketplace

KEYWORDS

Performance modeling

Defect detection

Education and training

Data modeling

Image segmentation

Convolution

Detection and tracking algorithms

Show All Keywords

Keywords/Phrases

Search In:

Publication Years