IoT-Integrated Deep Learning Model and SmartBin System for Real-Time Solid Waste Management

The increasing number of people living in metropolitan areas increases the risk that garbage will be disposed of in an unsustainable manner. Because of the high volume of people frequenting city halls and other government facilities, many urban areas now incur astronomical costs for garbage disposal. Waste collection and sorting is the most important part of any waste management system. Smart trash management is recommended in this research by the use of electronic smart sorting through the Internet of Things. The system's two primary functions-trash collection and waste classification-are controlled by a Raspberry Pi 4b microprocessor and three modules. In the past, these two primary features have been implemented independently; however, in this study, features are merged to provide a more complete smart bin waste disposal system. Overflow alarms using ultrasonic and tracker sensors initiate garbage pickup. To effectively separate biodegradable from non-biodegradable solid wastes, two methods have been used. The first method incorporates a Convolutional Neural Network (CNN) and Long Short Term Memory (LSTM) with the IoT, whereas the second method takes the first method's model and adds more sensors. Three different approaches of data collection are used with CNN+LSTM-based IoT. Images from Kaggle is the first approach, while using search engines like Google and Bing is the second, and direct capture in a studio is the third. It has been shown that the second method is superior, with an accuracy of 99%.