Deep-RIC: Plastic Waste Classification using Deep Learning and Resin Identification Codes (RIC)

Abstract

In this study, the authors designed an algorithm based on deep learning that can automatically classify plastic waste according to Resin Identification Codes (RIC). The proposed algorithm is built through several stages as follows. In the first stage, image acquisition of plastic waste is carried out, which is the input of the designed algorithm. The acquired plastic waste image must display the resin code of the plastic waste to be classified. Furthermore, the acquired image is divided into two sets, namely training and testing sets. The training set contains images of plastic waste used in the training phase of the deep learning architecture DenseNet-121 to identify the resin code of each plastic waste image and classify it into the appropriate class. The training phase is run for 100 epochs, and at each epoch, the cross-entropy loss function is calculated, which expresses the performance of the deep learning architectures in classifying plastic waste images. In the next stage, a trained deep learning architecture is used to classify the plastic waste images from the test set. Classification performance in the test set is also expressed as the cross-entropy loss function value. In addition, the accuracy value has also been calculated, which shows the percentage of the number of plastic waste images successfully classified correctly to the total number of plastic waste images in the test set, which the best accuracy is equal to 85%.