基于ResNet34卷积神经网络的垃圾分类识别小程序
Garbage Classification Recognition Wechat Applets Based on ResNet34 Convolutional Neural Network
李玉信 1王嘉欣 1刘力军1
作者信息
- 1. 大连民族大学 理学院,辽宁 大连 116650
- 折叠
摘要
人类社会的生产力水平正在以指数级提升,导致垃圾数量疯涨,因此当下如何处理大量的垃圾成为一个棘手的问题.在大量堆积的垃圾中既有可以回收利用的可回收垃圾,也有能造成污染的有害垃圾,如果对其不加以区分就丢弃,对于资源是一种浪费.为了解决在垃圾分类过程中出现的错误分类的问题,构建了基于ResNet34 卷积神经网络的垃圾分类识别模型.根据垃圾分类的需求对现有的网络模型做出了相应的调整,优化模型主要参数的同时采用了迁移学习的方式训练模型使其在测试集上的准确率达到了 87%.选择与微信小程序结合,向ResNet34 模型导入数据集并训练 40 种垃圾类别,同时通过Https协议远程调用服务器上运行的模型,从而在小程序上实现对垃圾的快速精准分类.
Abstract
The productivity level of human society is increasing exponentially,resulting in a skyrocketing amount of garbage,so how to deal with a large amount of garbage has become a tricky problem.At the same time,there are recyclable garbage that can be recycled and harmful garbage that can cause pollution.If it is discarded without distinction,it is a waste of resources.In order to solve the problem of misclassification in the garbage classification process,a garbage classification recognition model based on ResNet34 convolutional neural network is constructed.According to the needs of garbage classification,the existing network model was adjusted accordingly,the main parameters of the model were optimized,and the model was trained by transfer learning,so that the accuracy of the test set reached 87%.Select and combine with Wechat applet to import datasets and train 40 garbage categories to the ResNet34 model,and remotely call the model running on the server through the Https protocol,so as to achieve fast and accurate garbage sorting on the Wechat applet.
关键词
垃圾分类/ResNet34/微信小程序Key words
garbage sorting/ResNet34/Wechat applet引用本文复制引用
出版年
2024
NETL
NSTL
维普
万方数据