Optimization of Several Deep CNN Models for Waste Classification

With urbanization, population, and consumption on the rise, urban waste generation is steadily increasing. Consequently, management systems have become integral to city life, playing a critical role in resource efficiency environmental protection. Inadequate can adversely affect environment, human health, economy. Accurate rapid automatic classification poses significant challenge recycling. Deep learning models achieved successful image various fields recently. However, optimal determination of many hyperparameters crucial these models. In this study, we developed deep model that achieves best performance by optimizing depth, width, other hyperparameters. Our six-layer Convolutional Neural Network (CNN) with lowest depth width produced result an accuracy value 89% F1 score 88%. Moreover, several state-of-the-art CNN such as VGG19, DenseNet169, ResNet101, Xception, InceptionV3, RegnetX008, RegnetY008, EfficientNetV2S trained transfer fine-tuning. Extensive experimental work has been done find GridSearch. most comprehensive DenseNet169 model, which fine-tuning, provided 96.42% 96%. These be successfully used variety automation.