基于视觉识别的生活垃圾分类装置设计  

Design of Domestic Garbage Classification Device Based on Visual Recognition

作　　者：崔传坤 王舜 张乐乐 冯麟皓 江亚峰[1] 申燚[1] CUI Chuankun;WANG Shun;ZHANG Lele;FENG Linhao;JIANG Yafeng;SHEN Yi(College of Mechanical Electrical and Power Engineering,Jiangsu University of Science and Technology,Zhangjiagang 215600,China)

机构地区：[1]江苏科技大学机电与动力工程学院,江苏张家港215600

出　　处：《机械制造与自动化》2022年第6期193-197,共5页Machine Building & Automation

基　　金：江苏省大学生创新创业训练计划项目(202013991019Y)。

摘　　要：为了减少人工垃圾分类工作强度,提高垃圾分类效率和准确率,设计一种基于视觉识别的新型生活垃圾分类装置。进行分类装置的机械结构设计,包括投放识别机构、分类储运机构等部分;进行控制系统的设计,完成关键元器件的选型及各单元电路设计;基于MobileNet可分离卷积神经网络建立垃圾识别模型,构建数据集训练样本;完成分类装置的实验测试。测试结果表明:该装置能够完成可回收、有害、厨余和其他共4类垃圾的自动识别及分类,平均正确识别率可达93.33%,运行可靠。In order to reduce the work intensity of manual garbage classification and improve the efficiency and accuracy of garbage classification,a new type of domestic garbage classification device based on visual recognition is designed.Its mechanical structure including the placement identification mechanism,the classification storage and transportation mechanism,etc.is constructed.The control system is designed to complete the selection of key components and the circuit design of each unit.The garbage recognition model is established based on the MobileNet separable convolutional neural network,and the data set training samples are constructed.The experimental test of the classification device is conducted,with the results showing that the designed domestic waste sorting device can complete the automatic identification and classification of four types of recyclable,hazardous,food waste and other garbage,and the average correct recognition rate is 93.33%,which verifies the effectiveness and reliability of the device.

关 键 词：生活垃圾 分类装置 视觉识别 神经网络 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]