Scholarlink

Abstract <p>Wet waste has always been a special type of waste. Its disposal usually requires an effective monitoring mechanism. In order to realize the intelligent wet waste disposal monitoring system, we proposed a joint scheme based on key target identification and target tracking. First, a waste bag and human hand recognition model based on a convolutional neural network is designed. Then, according to the recognition results, we use Kalman filtering and Hungarian algorithm to track the movement of waste bags and human hands. Finally, compare the target motion trajectory with the calibrated position of the trash can to determine the correctness of the delivery. In the process, a new data set was formed based on more than 300 public data sets and more than 20 000 actual scene image data. On this basis, we conduct field experiments to verify the effectiveness of this method. The experimental results show that the method in this article can effectively identify key targets, track and discriminate movement trajectories, and efficiently monitor wet waste disposal.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A Deep Learning-Based Wet Waste Management System

  • 期刊论文

  • 发表日期：

    2025年12月1日

查看全文
  • Wenxiu Wang,
  • Yifei Wu,
  • Wei Zhang

摘要

Abstract

Wet waste has always been a special type of waste. Its disposal usually requires an effective monitoring mechanism. In order to realize the intelligent wet waste disposal monitoring system, we proposed a joint scheme based on key target identification and target tracking. First, a waste bag and human hand recognition model based on a convolutional neural network is designed. Then, according to the recognition results, we use Kalman filtering and Hungarian algorithm to track the movement of waste bags and human hands. Finally, compare the target motion trajectory with the calibrated position of the trash can to determine the correctness of the delivery. In the process, a new data set was formed based on more than 300 public data sets and more than 20 000 actual scene image data. On this basis, we conduct field experiments to verify the effectiveness of this method. The experimental results show that the method in this article can effectively identify key targets, track and discriminate movement trajectories, and efficiently monitor wet waste disposal.