Computer vision detection method for product defect detection

The invention relates to the technical field of computer vision, and particularly discloses a computer vision detection method for product defect detection, and the method comprises the steps: S1, multi-modal data collection, S2, feature extraction, S3, multi-modal fusion, S4, self-supervised learni...

Full description

Saved in:
Bibliographic Details
Main Authors ZUO PENG, HU YUTONG, WANG PENG, WANG ZHUORAN, DU ZEJIAN
Format Patent
LanguageChinese
English
Published 18.06.2024
Subjects
CALCULATING
COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS
COMPUTING
COUNTING
IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
PHYSICS
Online AccessGet full text

Cover

More Information
Summary:The invention relates to the technical field of computer vision, and particularly discloses a computer vision detection method for product defect detection, and the method comprises the steps: S1, multi-modal data collection, S2, feature extraction, S3, multi-modal fusion, S4, self-supervised learning, S5, generative adversarial network, S6, model evaluation and optimization, and S7, deployment and maintenance. The system is allowed to acquire abundant information from different sensors or modals through multi-modal fusion, so that the model can more comprehensively capture and understand the characteristics of a product, visible light and infrared images are combined, defects can be recognized under different illumination conditions, the defect images are generated and synthesized by using the generative adversarial network, and the defect recognition efficiency is improved. The model can learn the feature representation of some defect types which are rare or difficult to obtain a large number of samples, wh
Bibliography:Application Number: CN202410444246