Recyclable Waste Categorization with Transfer Learning
Abdelrazzaq Ahmad Abuhejleh, Moath Zaid Alafeshat, Natheer Almtireen, Hisham ElMoaqet, Mutaz Ryalat, Mohammad M. AlAjlouni · 2024
Environmental sustainability has become a critical concern globally, with waste management being a significant challenge. Traditional methods of waste segregation often rely heavily on manual labor, leading to inefficiencies, inaccuracies, and increased costs. As the volume of waste continues to grow, there is an urgent need to adopt innovative technologies to enhance waste management practices. This study presents an AI algorithm designed to improve the efficiency and accuracy of waste segregation, thereby enhancing sustain ability efforts. The algorithm utilizes advanced Convolutional Neural Network (CNN) techniques, specifically utilizing a pre-trained DenseNet169 model and employing transfer learning to classify waste materials into specific categories: paper, plastic, and metal. By employing CNNs, the system achieves high precision in identifying various types of waste based on visual data, thereby reducing human error and labor costs associated with manual sorting.