基于ResNet50的列车常见垃圾自动分类算法

Automatic Sorting Method for Common Garbage in Train based on ResNet50

针对现今的智能垃圾桶使用方法大都无法应用在列车的问题,提出了一种基于迁移学习方法的智能垃圾分类方法.首先建立了12种列车常见垃圾数据集,并应用迁移学习ResNet50模型进行模型设计与参数设计,给出了模型的准确率与模型收敛速度曲线,并与Inception-V3模型进行了对比实验,实验结果表明ResNet50模型有更好的识别效果.最后通过PyQt5编写GUI仿真系统,实现了基本的拍照与分类识别功能.

For the most of the current methods of using intelligent garbage which cannot be applied to the train,an intelligent garbage classification method is presented based on migration learning method.12 kinds of common garbage data sets of trains are established,and the model and parameter design are carried out by u-sing the resnet50 model of migration learning.The accuracy and convergence speed curve of the model are given,and the comparison experiment with the concept-v3 model is conducted.The experiment results show that the resnet50 model has better recognition effect.Finally,the GUI simulation system is programmed by pyqt5 to realize the basic functions of photographing and classification.