基于深度学习的木材缺陷智能检测的研究进展与展望  

Research Progress and Prospect of Intelligent Detection of Wood Defects Based on Deep Learning

作　　者：王明涛 项晓扬 崔文燕 院霖享 多化琼[1] WANG Ming-tao;XIANG Xiao-yang;CUI Wen-yan;YUAN Lin-xiang;DUO Hua-qiong(College of Materials Science and Art Design,Inner Mongolia Agricultural University,Huhhot 010018,Inner Mongolia,P.R.China)

机构地区：[1]内蒙古农业大学材料科学与艺术设计学院,内蒙古呼和浩特010018

出　　处：《林产工业》2024年第3期38-44,共7页China Forest Products Industry

基　　金：内蒙古自治区重点研发和成果转化计划项目(2022YFDZ0031)。

摘　　要：木材作为天然生物材料很容易受到内外界影响从而产生不符合人们生产需求的缺陷,人们为了准确高效的识别木材缺陷进行了大量的研究。本文对近年来基于深度学习的木材缺陷检测技术进行梳理,根据使用方法的侧重点不同将其分类,并针对典型方法加以细分归类和对比分析,总结了每种方法的优缺点及其应用面。此外,提出了基于深度学习的木材缺陷检测技术目前所存在的难点与所陷困境。As a natural biological material,wood is easily affected by internal and external influences,resulting in defects that do not meet people's production needs.People have conducted a lot of research in order to accurately and efficiently identify wood defects.In recent years,the wood defect detection technology based on deep learning has been studied and sorted out,and it has been classified according to the different emphasis of the use method,and the typical methods have been subdivided,classified and compared.The advantages and disadvantages of each method and its application were summarized.In addition,the difficulties and dilemmas of wood defect detection technology based on deep learning were put forward.

关 键 词：木材缺陷 单阶段目标检测 双阶段目标检测 神经网络 深度学习 

分 类 号：TS6[轻工技术与工程] TS396