Analysis of Multi-Class Convolution Neural Network on Garbage Classification System

The buildup of solid waste in metropolitan areas is a major worry that, if not effectively handled, might lead to environmental contamination and be dangerous to human health. To manage a range of waste products, it's crucial to have an advanced/intelligent waste management system. The act of separating garbage into its many components is one of the most crucial parts of waste management, and it is often carried out manually by hand-picking. To simplify this, we suggest a customized architecture for multiclass Convolutional Neural Networks (CNNs). The design includes elements that are individual-specific, enabling the model to adjust to the traits of each class. Extensive evaluation has shown that our CNN architecture, utilising a personalised approach, outperforms traditional CNN models like ResNet50 and VGG16 in multiclass classification situations. The effectiveness and resilience of our suggested architecture are demonstrated by experimental results on our dataset, opening the path for increased classification accuracy of 71.15%.