Implementation of Support Vector Machine for Recycled Plastic Bottles Detection

Societal and economic developments impact the level of consumption, which causes the addition of plastic waste. This study used the Support Vector Machine (SVM) method with Oriented Fast and Rotated Brief (ORB) feature extraction to detect recycled plastic bottles. The experiment was carried out using preprocessed datasets, and the results showed that the Support Vector Machine model with a polynomial kernel gave the best performance in classifying with an accuracy of 74%. In addition, the variation experiments on split data and kernel types show that the SVM model with polynomial kernels provides the best accuracy on the division of different training and test data. We recommend carefully selecting parameters according to the characteristics of the data. In future work, we will consider developing a more complex system by integrating recycled plastic bottle detection with Internet of Things (IoT) technology and developing better feature extraction to improve the accuracy of recycled plastic bottle detection.