Hidden defect detection based on metric learning

Ankang Liu; Lan Cheng; Yingchun Lv

doi:10.1117/12.2659344

22 November 2022 Hidden defect detection based on metric learning

Ankang Liu, Lan Cheng, Yingchun Lv

Proceedings Volume 12475, Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022); 124750M (2022) https://doi.org/10.1117/12.2659344
Event: Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 2022, Hulun Buir, China

Abstract

Aiming at the problems that hidden defects inside objects are difficult to be visually recognized and the defect samples obtained from inspection are few, a small-sample learning detection model using ultrasonic flaw detection to extend machine vision is proposed. The model introduces an attention mechanism into the deep nearest neighbor network to adjust the image features, so that the model pays more attention to the useful defect area features, increases the amount of defect-related information, and makes full use of key defect features to detect image targets. Experiments show that the proposed method has the best performance compared with the baseline model on the self-made hidden defect dataset, and the average correct rate is up to 83.85% under 10-shot; the model is tested with noisy images, and the results show that the model detection under noisy conditions has an accuracy rate of about 76%, and it has a certain anti-noise interference ability.

Citation Download Citation

Ankang Liu, Lan Cheng, and Yingchun Lv "Hidden defect detection based on metric learning", Proc. SPIE 12475, Second International Conference on Advanced Algorithms and Signal Image Processing (AASIP 2022), 124750M (22 November 2022); https://doi.org/10.1117/12.2659344

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