Waste Segregation using Deep Learning-based Convolutional Neural Network Model

Waste segregation is a major issue faced by recycling systems in big cities across the country. Annually, about 62 million tons of trash are generated in India, with plastic materials accounting for about 5.6 million metric tons of this waste. Out of the 43 million tonnes of waste that is produced, approximately 60% is recycled, which is approximately 11.9 million tons. Waste segregation is critical in the recycling industry, and its positive effects cannot be overstated. However, in India, garbage is not segregated during the time of its collection due to the requirement of a significant amount of staff and effort to separate it. Furthermore, workers in this field are at risk of various illnesses due to the presence of hazardous compounds in the trash. Therefore, the goal is to increase productivity by reducing social interaction during the waste segregation process. To achieve this, the proposed model has been developed using a convolutional neural network-based image classifier capable of identifying objects and indicating the type of waste they contain. The CNN model retrieves information from images to generate predictions and distinguish one type of waste from another in the same category. In the proposed work, the results have shown that CNN classifies the waste with an accuracy of 93% resulting in an efficient and effective model.