Intelligent Waste Sorting Bin for Recyclable Municipal Solid Waste

This paper describes about a sorting bin of Recyclable Municipal Solid Waste. This bin is described as intelligent due to automatic source separation method has been implemented in this bin. A method called Convolutional Neural Network (CNN) has been implemented in waste sorting process. CNN is one kind of deep learning process. CNN is very fast, efficient and reliable method for object classification. For this reason, this method is implemented for waste sorting. There are many algorithms have already been developed for object detection. Among these algorithms the Inception-ResNet V2 is used. Inception-ResNet V2 is simply the combination of the Inception and the Residual Network structure. This model is trained by images of PET and LDPE about 300 of each sample. PET bottles and chips packet has been considered as sample. Training has been done by Google Collaboratory platform. Accuracy of detection is measured on the basis of 100 observations and accuracy is 84 CSS for manual feeding of image and decision making. This technique may increase recyclable MSW (PET LDPE) collection rate, reduce environment pollution, can help recycling plants economically.