Garbage classification method based on deep learning

Abstract

The invention relates to a garbage classification method based on deep learning. The method comprises the steps of firstly, using a crawler for crawling and constructing a deep learning training set,and a data set of the version is called as an initial data set; training a model on the basis of the initial data set by using different network structures to obtain a plurality of models; performingautomatic crawling, cleaning and marking on the data by using the models in combination with crawlers to complete automatic expansion of the initial data set and construct a final version data set; and finally, selecting the network structure with the best performance as a backbone network, training the model on the finally constructed data set, and optimizing by using various optimization strategies to improve the accuracy and robustness of the model.

Description

technical field [0001] The invention relates to the fields of pattern recognition and computer vision, in particular to a garbage classification method based on deep learning. Background technique [0002] With the improvement of people's living standards, more and more domestic waste is produced in cities. For the disposal of massive waste, the cost to be paid is very high. Whether it is incineration or landfill, it needs to pay huge manpower and material resources, and also at the cost of polluting the environment. Sorting domestic waste, maximizing the utilization of resources, reducing the burden of waste disposal, and achieving sustainable development have gradually gained people's attention. However, since the domestic waste classification policy was implemented in all provinces and cities across the country, the status quo is not too optimistic. The implementation of waste sorting in many pilot communities is not good, and some provinces and cities even hire social...

AI Technical Summary

Problems solved by technology

[0003] The existing garbage classification data sets have the characteristics of few categories and small data volume, which cannot well represent the wide variety of garbage in reality.

Existing garbage classification algorithms are all aimed at relatively small data sets. Such algorithms have the characteristics of poor generalization ability and are difficult to be widely applied to daily life scenarios.

Images