垃圾识别分类系统研究与探索

Research and Exploration on Classification System of Garbage Identification

主要研究一种图像识别系统,对图片中废弃的电子电器设备进行分类与识别,其目的是加强个人和垃圾回收机构之间高效信息的传递,从而促进智能设备在日常生活中的使用和推广。为了改进废物收集策略,个人可以拍摄废物设备图片,然后将图片上传到垃圾回收公司服务器,在该服务器中将对其进行自动识别和分类。结合当前现状,提出一种新颖的思路,利用深度卷积神经网络(Convolutional Neural Network, CNN)对电子垃圾进行分类,基于区域的快速深度卷积神经网络(Faster Region-based Convolutional Neural Network, Faster R-CNN)检测图片中废物设备的大小并分类,准确率约为90%~96.7%。从上传的图片中自动识别出垃圾的类别和大小,垃圾回收公司就可以为特定的电子垃圾分配足够的车辆和货物装载能力,精准地制定收集计划。

An image recognition system that classifies and recognizes discarded electronic and electrical equipment in pictures is mainly studied. Its main purpose is to strengthen the efficient information transfer between individuals and garbage collection agencies, thereby promoting the use and popularization of smart devices in daily life. In order to improve the waste collection strategy, individuals can take pictures of waste equipment, and then upload the pictures to the waste recycling company server, where they will be automatically identified and classified. Combined with the current situation, a novel idea is proposed, which uses deep convolutional neural network to classify electronic waste. The region-based fast deep convolutional neural network detects the size of waste equipment in the picture and classifies it, and the accuracy rate is about 90%-96.7%. By automatically identifying the type and size of garbage from the uploaded pictures, garbage recycling companies can allocate sufficient vehicles and payload capacity for specific e-waste, and accurately formulate collection plans.