An Efficient Model on AIoT Devices for Trash Classification Applications

Waste segregation is crucial for protecting the environment and promoting sustainable development. In developing countries like Vietnam, there is a lack of public awareness and action regarding garbage separation at its origin. By leveraging the Internet of Things (IoT) and artificial intelligence (AI), we have developed an IoT-based smart device integrated with an AI-based model, the so-called AIoT device, to classify waste via a camera. In contrast to current smart trash cans, which can automatically categorize and dispose of waste, our AIoT device offers users information regarding the specific type of waste, facilitating appropriate disposal practices. This feature enhances users’ understanding of garbage classification at its origin. Significantly, we propose a CNN-based model called BEGNet that employs RegNetY120 as its backbone, adds two additional layers, and adopts a revised activation function. The experimental results indicate that the proposed model demonstrates high efficiency in terms of accuracy compared to other approaches on both the Trashnet dataset and our built dataset.