A Web-Based Dataset for Garbage Classification Based on Shanghai’s Rule

Abstract

In 2019, Shanghai has published a new regulation of trash management, which obtains a series of achievements in managing trash. In order to implement this regulation further and help citizens understand clearer about trash classification, we decided to develop a deep learning dataset and models to classify waste automatically. The object of this study is to take an image as input and identify the category of trash. We write a web crawler to capture images. After pre-processing, we gain about 14,000 images in total. We compare models including CNN, a ResNet50 model, and a VGG16 model on the dataset. Our experiments show that the ResNet50 model perform better than the others.

Topics & Concepts