Quantum Deep Learning: Reliability of deep learning empowered garbage sorting and detection by Visual Context for Aerial Images

Abstract

Environmental pollution by garbage is the biggest problem in most developing countries; garbage waste processing management and recycling are significant for ecological and economic reasons. Computer vision techniques are very advanced in many applications for object detection and classification, an extensive study on the use of artificial intelligence for garbage processing has been done and it is lagging because of dataset availability which has the top view images of garbage. A new dataset ‘KACHARA’ which has 4715 images of seven classes has been created. Classification is performed by the transfer learning by popular Deep learning model MobileNetv3 large with fine tuning the top layers achieving the classification accuracy of 94.37. Our proposed method yields an accuracy of 94.37% with 7 classes, the maximum in garbage where it was 5000 images, and 90% accuracy. CNN1 Model accuracy is 94% with only 2 classes of 6000 images. With 94.37% accuracy, our model classifies objects significantly. Deep learning and the principles of quantum computing have been used for garbage sorting and detection by visual context for aerial images.