A Comparative Analysis of Garbage Classification Using ResNet50, Xception and DenseNet121 Architecture: A Review

The volume of waste is fast increasing daily due to the rise in economy and the rise in living standards of people. If garbage is not gathered and controlled appropriately, it will pollute and impact waterways and environments for thousands of years. Garbage classification has significance in managing waste and ensuring the sustainability of the environment. Deep learning-based convolutional neural networks are used for classifying the garbage. The paper uses the dataset containing 2527 images of six distinct classes: cardboard, glass, metal, paper, plastic and trash. This comparative analysis uses the ResNet50, Xception and DenseNet121 models to classify the waste. After comparing the performance of these three models, it is observed that ResNet50 achieved the best accuracy of 87.50%, while Xception and DenseNet121 attained 86.33% and 85.43%, respectively.