期刊
ARTIFICIAL INTELLIGENCE REVIEW
卷 -, 期 -, 页码 -出版社
SPRINGER
DOI: 10.1007/s10462-023-10475-7
关键词
Deep learning; Computer vision; Surface defect inspection; Real-time inspection
In recent years, deep learning methods have been widely used in various industrial scenarios, promoting industrial intelligence. Real-time surface defect inspection of industrial products is one of the research focuses in industry. Surface defect inspection methods based on deep learning show great advantages and make it possible to detect defects in real time with high accuracy.
In recent years, deep learning methods have been widely used in various industrial scenarios, promoting industrial intelligence. Real-time surface defect inspection of industrial products is one of the research focuses in industry. Surface defect inspection methods based on deep learning show great advantages and make it possible to detect defects in real time with high accuracy. From the perspective of real-time inspection, according to different types of surfaces in industry, this paper reviews the latest deep learning-based surface defect inspection methods from three levels: defect classification, defect detection and defect segmentation. After that, this paper introduces commonly used metrics for evaluating the performance of surface defect inspection models and public surface defect datasets. Then, this paper discusses the challenges faced by deep learning-based real-time surface defect inspection methods, including the acquisition of surface defect datasets, balancing the accuracy and speed of inspection models, and the application in industrial environments. Finally, this paper provides an outlook on the future development trend of surface defect inspection.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据