Reverse Vending Machine with Power Output for Mobile Devices using Vision – Based System

Abstract

Polyethylene Terephthalate (PET) bottles are containers used widely in beverages and food containers, and most of the PET bottles are not recycled. The study focuses on how to recycle PET bottles. This study used image processing to ensure that the input PET bottles are in the state of recycling. It used the YOLOv3 framework algorithm for the training of images. This algorithm is advantageous when it comes to its fast speed while maintaining its high accuracy. For image detection, it used a camera to take an image. Then, the data will be sent to raspberry pi to classify the input bottle. If the input bottle satisfies the conditions, a signal will be sent to the DC motor to open the PET bottle platform. The corresponding type of bottle, including the PET bottle’s equivalent time, will be displayed on the LCD screen. The research conducted was successful. It has an accuracy rate of 90 % and 10% error rate. The device’s time delay is 10.966 seconds. As a result, a reverse vending machine is one of the alternatives for recycling.