Design and Analysis of AI based Autonomous Waste Segregator using Deep Learning

Although the existing manual waste segregator method is simple, it is not effective in terms of accurate waste classification. This necessitates the need to automate the process of waste classification in order to maintain a clean environment. The proposed model is an efficient system suitable for satisfying the emerging societal and environmental requirements. The proposed system utilizes the supervised machine learning algorithms to classify the waste into the different categories such as cardboard, glass, metal, garbage, paper and plastic. Initially, the data is collected and the enhancement process will be performed. The proposed algorithm converts a set of color images available in multiple folders to grayscale by converting them into a 2D matrix. The images will be then transformed and stored in a 1-Dimensional array, which is then used for labeling the data during testing process. By using CNN, the input images are then sampled and integrated to identify the human faces present in the image. Pooling process if done iteratively to reduce the image size. The proposed system presents a simple and efficient method to perform garbage classification by using machine learning algorithms by completely eliminating the manual human intervention in the waste classification step. The testing process has achieved an 80% efficiency.