摘要
随着生活水平的不断提高,我国生活垃圾总量不断增长,但生活垃圾的分类处理能力及技术相对有限和落后。现有的垃圾分类方式主要依靠人工分辨,很多居民对于垃圾分类的意识不足,往往无法做到正确有效的分类。本系统由主控板、机器视觉模块及云服务器后台组成。利用Tensorflow框架下的MobileNet v2网络模型进行识别分类,借助STM32F103ZET6主控核心对识别结果进行处理,并通过WiFi模块上传至云服务器实现数据存储。云服务器使用腾讯云部署的EMQ服务器,作为本系统后台以供后续数据分析使用。实验结果表明,在网络模型文件仅占697KB的情况下,训练集的准确率达到93%,可以实现可回收垃圾较为准确的智能分类。
With the continuous improvement of living standards,the total amount of domestic garbage in China is increasing,but the sorting and processing capacity and technology of domestic garbage are relatively limited and backward.The existing gar⁃bage classification methods mainly rely on manual classification.Many residents have insufficient awareness of garbage classifica⁃tion,often unable to achieve correct and effective classification.The system consists of machine main control board,vision module and cloud server background.The MobileNet v2 network model under Tensorflow framework is used for identification and classifi⁃cation.The STM32F103ZET6 main control core is used to process the identification results and upload them to the cloud server through WiFi module.The cloud server uses the EMQ server deployed in Tencent Cloud as the background of the system for subse⁃quent analysis.The experimental results show that when the network model file occupies only 697KB,the accuracy rate of the train⁃ing set reaches 93%,which can achieve more accurate intelligent classification of recyclable garbage.
作者
张兆屹
李和福
王明红
Zhang Zhaoyi;Li Hefu;Wang Minghong(School of Physics Science and Information Technology,Liaocheng University,Liaocheng 252000)
出处
《现代计算机》
2022年第14期96-101,116,共7页
Modern Computer
基金
山东省自然科学基金面上项目(ZR2021MF097)。
