基于GPU深度学习的家用智能垃圾桶设计  

Design Of Household Smart Trash Can Based On GPU Deep Learning

作　　者：李美锟 杨明夏 凌滨[1] Li Meikun;Yang Mingxia;Ling Bin(Northeast Forestry University,Harbin Heilongjiang,150040)

机构地区：[1]东北林业大学,黑龙江哈尔滨150040

出　　处：《电子测试》2021年第1期8-11,共4页Electronic Test

基　　金：东北林业大学消极“大学生创新创业”项目。

摘　　要：垃圾分类成为日常所需,为解决垃圾分类的问题,使投放垃圾变得更加环保高效,本文设计了基于GPU深度学习的智能垃圾桶。应用Tensorflow深度学习框架,使用GPU加速训练,并结合硬件树莓派处理垃圾的图像信息,通过单片机控制步进电机实现垃圾分类。本设计具有分类准确度高,使用简单便捷,可实现垃圾分类的清洁化、高效化、智能化。Garbage classification has become a daily necessity.In order to solve the problem of ga rbage classification and make garbage disposal more environmentally friendly and efficient,this p aper designs a smart garbage can based on GPU deep learning.Apply Tensorflow deep learning framew ork,use GPU to accelerate training,and combine the hardware raspberry pie to process garbage ima ge information,and realize garbage classification through the stepper motor controlled by single-chipmicrocomputer.The design has high classification accuracy,is simple and convenient to use,and c an realize the clean,efficient and intelligent garbage classification.

关 键 词：智能分类垃圾桶 神经网络 GPU加速 单片机控制 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程] X705[自动化与计算机技术—控制科学与工程]