基于深度學(xué)習(xí)的可回收垃圾自動(dòng)分揀系統(tǒng)

打開(kāi)文本圖片集

[摘    要]目前，市場(chǎng)對(duì)垃圾后端處理仍處于萌芽狀態(tài)，而深度學(xué)習(xí)技術(shù)作為近些年的熱點(diǎn)，能夠配合硬件系統(tǒng)，完成可回收垃圾的分揀工作。系統(tǒng)根據(jù)視覺(jué)識(shí)別結(jié)果控制相應(yīng)的機(jī)械爪對(duì)目標(biāo)垃圾進(jìn)行抓取，最終在傳送帶末端打包。實(shí)驗(yàn)結(jié)果表明，所提出的方法可以較好地實(shí)現(xiàn)預(yù)期分揀目標(biāo)。

[關(guān)鍵詞]垃圾分類;深度學(xué)習(xí);卷積神經(jīng);視覺(jué)系統(tǒng)

[中圖分類號(hào)]TP18 [文獻(xiàn)標(biāo)志碼]A [文章編號(hào)]2095–6487（2022）08–0–03

Automatic Sorting System for Recyclable Garbage Based on Deep Learning

Xu Li，Liu Yi-zhi，Wang Ze-long，Guo Jin-shuai，Ding Jia-qi

[Abstract]At present， the market for waste back-end treatment is still in its infancy. As a hot spot in recent years， deep learning technology can cooperate with hardware systems to complete the sorting of recyclable garbage. According to the visual recognition result， the system controls the corresponding mechanical claw to grab the target garbage， and finally pack it at the end of the conveyor belt. The experimental results show that the proposed method can achieve the expected sorting goal well.

[Keywords]garbage classification; deep learning; convolutional neural; visual system

1 研制背景及意義

垃圾圍城是我國(guó)當(dāng)前城市管理的一大難題，僅2020年新增可回收垃圾總量便超過(guò)30億t。（剩余4166字）