摘要
垃圾分类的实施过程中存在配套设施不完善、居民的分类意识浅薄和分类的正确性较低的问题。为了提高垃圾分类的准确度,降低二次分拣成本,本文基于深度学习中的图像识别技术设计和实现了一款移动端垃圾分类软件。该软件分为前端与后端,前端向用户提供图像识别、垃圾检索、模型优化等服务,后端响应用户的请求并进行模型的训练。通过对60种垃圾图像的识别结果,识别准确度达54%,说明该软件可以帮助居民提高对生活垃圾分类的准确度。
During the implementation of waste sorting,there are problems such as imperfect supporting facilities,shallow awareness of residents and low correctness of classification.In order to improve the accuracy of Waste sorting and reduce the cost of secondary sorting,this paper designs and implements a mobile Waste sorting software based on image recognition technology in deep learning.The software is divided into a front-end and a back-end.The front-end provides users with services such as image recognition,garbage retrieval,and model optimization,while the back-end provides response to user requests and model training.Through the recognition results of 60 kinds of garbage images,the recognition accuracy reaches 54%,which indicates that the software can help residents improve the accuracy of Waste sorting.
作者
陆公正
江思源
LU Gongzheng;JIANG Siyuan(School of Computing Science and Artificial Intelligence,Suzhou City University,Suzhou,China,215104;School of Computer Engineering,Suzhou Vocational University,Suzhou,China,215104)
出处
《福建电脑》
2023年第7期102-105,共4页
Journal of Fujian Computer
基金
江苏高校哲学社会科学研究项目(No.2021SJA1461)
江苏省高等学校大学生创新创业训练计划项目(No.202111054005Y)资助。
