An Intelligent System for Waste Materials Segregation Using IoT and Deep Learning

Abstract

Abstract Disposing of Garbage is still a hurdle in today’s world and not many waste materials are disposed of correctly as per the requisite procedure. They are rather dumped in landfills and other sources like water bodies causing Pollution. The proposed system amalgamates the Internet of Things with Deep Learning and Image Processing which replaces the conventional waste management system with a smart segregation process. The proposed model consists of a Raspberry Pi camera to scan and detect the object. This detected object is then exported to the Deep learning Model as the input using the ‘Histogram of Oriented Gradients’ algorithm. This is invariable to the geometrical and photometrical changes of the object. The image classification is accomplished using the combination of Faster Regions with the CNN algorithm (RCNN) with contextual information and incremental learning which improves the detection performance under special and extreme conditions, accurately classifying the objects into bio and non-biodegradable waste. Using the Faster R-CNN algorithm, explicit information can be obtained from the different layers of the neural network of Faster R-CNN, thereby proving that the proposed system has a higher accuracy rate.