Smart Waste Classification System using Deep Learning for Automated Recycling
Aljawharah Alshammari, Shahad Alkhabbaz, Rawan F. AL-Mutairi, Haya Alhabad, Sadan Alhamdan, Ilyes Boulkaibet · 2025
Efficient waste management and recycling are crucial in addressing global environmental sustainability challenges. Traditional manual sorting methods are slow, errorprone, and inefficient. This paper proposes an AI-driven waste classification system that leverages deep learning to automate waste segregation. A custom image dataset with five waste categories was created to train and improve classification accuracy. The system employs an Enhanced CNN-based MobileNetV3 model, achieving an average accuracy of 94.25%. The model is integrated into a Raspberry Pi module, which controls the waste sorting mechanism using an infrared (IR) sensor for waste detection and a USB camera for image acquisition. Once classified, actuators direct waste into the appropriate bins. The results demonstrate the effectiveness of AIpowered waste management in enhancing recycling efficiency and promoting environmental sustainability.