基于人工智能的广西都安山羊皮革瑕疵自动检测分级研究  

Research on Automatic Detection and Grading of Imperfections in Du'an Goat Leather from Guangxi Province,based on Artificial Intelligence

作　　者：莫佳慧 MO Jiahui(Nanning Vocational and Technical University,Nanning 530008,China)

机构地区：[1]南宁职业技术大学,广西南宁530008

出　　处：《西部皮革》2024年第14期58-60,共3页West Leather

基　　金：南宁职业技术大学2023年“三教”改革项目“互联网创业实务创新实践示范课程建设”(2023JG152);南宁职业技术大学2023年“三教”改革项目“互联网创业实务创业高阶选修课程示范建设”(2023JG167)。

摘　　要：通过文献研究和案例分析,文章探究了基于人工智能的皮革瑕疵自动检测分级技术。文章重点探究如何解决广西都安山羊板皮经脱毛鞣制后,由于其本身的皮质特性使得针对其进行人工检测的人员因受到天气条件、光线变化、视力健康等因素干扰而导致其在标记瑕疵时经常出现错检或漏检等出错率较高的问题。研究认为,基于深度学习(DL)中的全卷积网络(FCN)和机器学习(ML)中的深度神经网络(DNN)等人工智能技术能够对广西都安山羊皮革各类瑕疵进行有效地自动检测分级,最大程度上使广西都安山羊皮革能够更好地对标符合当前中国-东盟RCEP(区域全面经济伙伴关系协定)中对于成员国皮革进出口贸易规范(质量标准)的要求。Through literature review and case analysis,the paper explores the technology of AI-based automatic defect detection and grading in leather.It focuses on addressing the challenges faced by manual inspectors of Du'an goat leather,where factors such as weather conditions,variations in lighting,and visual impairments often lead to high error rates in defect marking.The study argues that artificial intelligence techniques such as Fully Convolutional Networks(FCN)in deep learning(DL)and Deep Neural Networks(DNN)in machine learning(ML)can effectively automate the detection and grading of various defects in Du'an goat leather.This approach maximizes compliance with the quality standards required under the China-ASEAN RCEP(Regional Comprehensive Economic Partnership)for leather imports and exports among member countries.

关 键 词：人工智能 皮革瑕疵 都安山羊 机器学习 深度学习 深度神经网络 全卷积网络 检测分级 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程] TS56[自动化与计算机技术—控制科学与工程]