Skip to main navigation Skip to search Skip to main content

Fast and High Precision Surface Defect Detection Method Based on New Label Assignment

  • Ruitao Li*
  • , Xiaojun Wu
  • , Jinghui Zhou
  • , Jiarui Zheng
  • *Corresponding author for this work
  • Harbin Institute of Technology Shenzhen
  • School of Information Technology and Engineering
  • Ltd

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Current object detection networks suffer from low accuracy and slow speed for industrial defect detection tasks. Industrial defect detection tasks are characterized by small area and large aspect ratio of the detected objects, as well as high speed requirements. We provide a label assignment strategy for defect shape characteristics to improve the training efficiency of a one-stage target detection network for defect detection scenarios. Also, label assignment distillation learning is used to obtain a model that takes into account the speed of detection. In this paper, experiments are conducted on several industrial defect datasets, and metrics such as mAP (mean average precision) values and inference speed are calculated. Compared with other models, the label assignment algorithm results in a 3% improvement in detection accuracy and a 58% acceleration in model inference speed after lightweighting.

Original languageEnglish
Title of host publicationProceedings of the 2023 IEEE International Conference on Real-Time Computing and Robotics, RCAR 2023
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages257-262
Number of pages6
ISBN (Electronic)9798350327182
DOIs
StatePublished - 2023
Externally publishedYes
Event2023 IEEE International Conference on Real-Time Computing and Robotics, RCAR 2023 - Datong, China
Duration: 17 Jul 202320 Jul 2023

Publication series

NameProceedings of the 2023 IEEE International Conference on Real-Time Computing and Robotics, RCAR 2023

Conference

Conference2023 IEEE International Conference on Real-Time Computing and Robotics, RCAR 2023
Country/TerritoryChina
CityDatong
Period17/07/2320/07/23

Keywords

  • deep learning
  • defect
  • detection
  • distillation learning
  • label assignment

Fingerprint

Dive into the research topics of 'Fast and High Precision Surface Defect Detection Method Based on New Label Assignment'. Together they form a unique fingerprint.

Cite this