Household Waste Image Classification Using Deep Learning Model

Household waste management poses a significant challenge in maintaining environmental cleanliness, particularly in the classification process, which is still largely carried out manually. This research aims to develop an automatic classification system for inorganic household waste using a Deep Learning model based on Convolutional Neural Networks (CNN). The waste types classified include glass, plastic, metal, and paper. The system development process involves several stages, dataset collection, image preprocessing, CNN model design using the MobileNetV2 architecture, model training and evaluation, and integration into a web-based application using Flask. Testing results show that the developed system is capable of classifying waste images with an accuracy exceeding 80%. The system offers advantages in classification speed, ease of use, and an intuitive user interface featuring image uploads and real-time classification via the device camera. However, limitations such as dataset diversity and image capture quality still affect prediction accuracy. This research demonstrates that the application of CNN in waste management provides a significant contribution to the efficiency and effectiveness of waste classification systems.