Smart waste segmentation deep learning based approach

Abstract

The volume of solid waste in urban areas is becoming an alarming threat that causes environmental deterioration and a threat to human health. To manage a variety of wastes, it is important to have an efficient and intelligent waste management system. Separating waste into several components is one of the most important steps in waste management, and this process is usually done manually by sorting. To simplify the process, we proposed two trained convolutional neural network-based models (Unet and Mask RCNN). We used them to segment the waste and then facilitate the process of sorting the waste into different groups/types such as glass, metal, paper, plastic, cardboard and garbage. The proposed system is tested on Mju-Waste dataset, developed by Tao Wang from. Our Unet model achieved 98% accuracy after 40 training epochs and an average accuracy that exceeds 90% for our Mask RCNN model after 50 training epochs. Our proposed garbage segmentation system makes the garbage separation process faster, smarter and optimal.