Household Waste Classification with Convolutional Neural Networks (CNN)

Indonesia is the second largest waste-producing country in the world after Saudi Arabia, with a waste disposal rate of up to 300 kg per person yearly, which is expected to increase by 2025. Citizen participation in managing and sorting household waste is an effective way to simplify recycling. recycling and management of household waste. This research aims to build a waste classification model based on the type and method of management so that various types of waste can be grouped and can simplify the waste management process, whether for recycling or disposal at the final disposal site. The household waste classification model was built using the Convolutional Neural Network (CNN) algorithm. All research data uses images with classification results, including inorganic, B3, paper, organic, and residue waste. This research uses 5000 image data from household waste types, divided into training and testing data. The training data uses 4,000 images of 800 images of each household waste, while the test data uses 200 images of each type of household waste. The experiments in this study used 4 (four) test scenarios with variations in learning rate and dropout parameters, namely learning rate variations including 0.001 and 0.0001, while dropout variations were 0.2 and 0.5. The results of all experimental scenarios show that a learning rate of 0.001 and a dropout of 0.5 have the best accuracy value of 85%. This value proves that the CNN algorithm performs well in detecting household waste.