IoT incorporated deep learning model combined with SmartBin technology for real‐time solid waste management. (3rd December 2021)

• Record Type: Journal Article
• Authors: Sivakumar, Muthuramalingam; Renuka, Perumal; Chitra, Pandian; Karthikeyan, Sundararajan
• Is Part Of: Computational intelligence. Volume 38:Number 2(2022)
• Page Start: 323
• Page End: 344
• Publication Date: 2021-12-03
• Subjects: biodegradable -- convolutional neural network -- deep learning -- Internet of Things -- non‐biodegradable; Artificial intelligence -- Periodicals; Computational linguistics -- Periodicals; 006.3
• DOI: 10.1111/coin.12495
• Languages: English
• ISSNs: 0824-7935

Abstract

Abstract: With a view of the massive human resources and time requirements, the need for an automated, more accurate, and quicker method for handling the classification of solid wastes is felt more than ever worldwide. In this work, an attempt has been made to develop a model named SmartBin. Two different approaches have been followed to classify solid wastes as biodegradable and non‐biodegradable efficiently. The first approach is based on convolutional neural network (CNN) and Internet of Things (IoT), while the second approach adds several sensors to the model developed using the first approach. CNN‐based IoT is applied on datasets collected using three methods. The first one is Images from Kaggle; the second method adopted searches through Google and Bing, whereas the third one involved captured manually under a controlled environment. It is observed that the second approach has proved to be better, with an accuracy level of 98.57, which is a significantly improved performance over the first approach with an accuracy of 95.24%.