Reverse Vending Machine Item Verification Module using Classification and Detection Model of CNN

Abstract: Reverse vending machine (RVM) is an interactive platform that can boost recycling activities by rewarding users that return the recycle items to the machine. To accomplish that, the RVM should be outfitted with material identification module to recognize different sort of recyclable materials, so the user can be rewarded accordingly. Since utilizing combination of sensors for such a task is tedious, a vision-based detection framework is proposed to identify three types of recyclable material which are aluminum can, PET bottle and tetra-pak. Initially, a self-collected of 5898 samples were fed into classification and detection framework which were divided into the ratio of 85:15 of training and validation samples. For the classification model, three pre-trained models of AlexNet, VGG16 and Resnet50 were used, while for the detection model YOLOv5 architecture is employed. As for the dataset, it was gathered by capturing the recycle material picture from various point and information expansion of flipping and pivoting the pictures. A progression of thorough hyper parameters tuning were conducted to determine an optimal structure that is able to produce high accuracy. From series of experiments it can be concluded that, the detection model shows promising outcome compare to the classification module for accomplishing the recycle item verification task of the RVM.

Razali Tomari, Nur Syahirah Razali, Nurul Farhana Santosa, Aeslina Abdul Kadir and Mohd Fahrul Hassan. “Reverse Vending Machine Item Verification Module using Classification and Detection Model of CNN”. International Journal of Advanced Computer Science and Applications (IJACSA) 12.10 (2021). http://dx.doi.org/10.14569/IJACSA.2021.0121044

@article{Tomari2021,
title = {Reverse Vending Machine Item Verification Module using Classification and Detection Model of CNN},
journal = {International Journal of Advanced Computer Science and Applications},
doi = {10.14569/IJACSA.2021.0121044},
url = {http://dx.doi.org/10.14569/IJACSA.2021.0121044},
year = {2021},
publisher = {The Science and Information Organization},
volume = {12},
number = {10},
author = {Razali Tomari and Nur Syahirah Razali and Nurul Farhana Santosa and Aeslina Abdul Kadir and Mohd Fahrul Hassan}
}