Automated Waste Segregation Using ResNet50 for Biodegradable and Non-Biodegradable Classification

Aiming to separate waste into biodegradable and non-biodegradable categories, this work proposes the construction of an automated waste classification system utilizing the ResNet50 deep learning model. Over 20 epochs, the model was trained and assessed to have a final training accuracy of 98.24 % and a validation accuracy of 96.23%. Indicating the model's great learning ability and generalization to unseen data, the training loss dropped dramatically from 0.5799 to 0.0575 while the validation loss dropped from 0.5782 to 0.2936. These results highlight how well the ResNet50 architecture handles challenging picture classification tasks, therefore providing a strong instrument for practical use in automated trash segregation systems. The high accuracy and low loss values demonstrate the model's potential to improve waste management practices by enhancing recycling efficiency and reducing environmental impact. This research highlights the critical role of AI in environmental sustainability and sets the stage for future advancements in waste classification technologies, with the potential for integration into comprehensive waste management systems to address global waste challenges.