GarbageDetectNet: A Framework for Multi-Class Garbage Detection using Convolutional Neural Network

The frequent generation of alarmingly high domestic solid waste adds to serious climate change and increased pollution. To separate recyclables, automated garbage classification during the early stage of disposal might be a useful approach. This study presents a method for garbage categorisation and recognition utilising a convolutional neural network, which can aid in waste resource management, prevent environmental pollution, and make garbage sorting easier. However, current research is restricted to a small number of garbage categories, like TrashN et dataset, which has 2,527 pictures and 6 types of waste. Some essential categories like clothes, batteries, and biological products are absent from this dataset, which should represent actual environmental issues. Therefore, the performance of proposed convolutional neural network model was assessed using a dataset comprising 15,596 images and 12 classes of household garbage. U sing a neural network, a garbage categorization model was created. After using a customized CNN model, a high accuracy of 96% is achieved. Compared to previous advanced models, this proposed model showed efficacy with fewer parameters and faster training due to use of Adam optimizer and ReL U activation function. This indicates the practicality of the suggested paradigm for real-life environmental concerns.