Research on Computer Vision-Based Waste Sorting System
Haochen Cai, Xinli Cao, Likun Huang, Lianying Zou, Shubin Yang
Abstract
Nowadays, waste sorting has become a hot topic of society in China. Many cities, such as Beijing and Shanghai, have begun to strictly implement regulations of waste sorting. However, in this process, it still exists that people are not able to distinguish them for residual waste or household food waste. This paper utilizes computer vision approach to recognize two categories of wastes. We find out typical examples from them for image recognition, and collect relevant image dataset through the Internet, with a total of 2800 images. The methods used are support vector machine based on feature extraction and transfer learning based on convolutional neural network. Our experiment shows that the latter has better performance.