Image Based Municipal Garbage Bin Surveillance System Using Machine Learning

Abstract

Garbage is a significant issue today because it contributes significantly to the contamination of our planet. With more people, it is more inevitably thrown away daily. The public generally throws it on the roadway and around the Bin. It is the primary focus of our identification system. This study aims to develop a practical technique for identifying municipal waste by employing machine learning algorithms like convolutional neural network (CNN), artificial neural network (ANN), and support vector machine (SVM). The dataset the models utilise is images of garbage and Bins, classified into two classes, filled or empty. Each image in the dataset transforms into three 2-dimensional arrays with values between 0 and 255 before being normalised within the range 0 to 1. The models are trained and validated using a training dataset, and results are analysed and evaluated using a testing dataset. Our results show that the accuracy and receiver operating characteristic (ROC) curve area of the CNN, ANN, and SVM models are high (0.83, 0.68, and 0.68, respectively) and (0.90, 0.69, and 0.76, respectively). Therefore, this research recommends the CNN model for this sort of challenge. The city's municipal authorities may utilise the model to quickly and accurately identify the source, allowing for timely and effective removal.