Development of an Image-Based Reverse Vending Machine Using Raspberry Pi

An increasing population and urbanization influence waste from plastic bottled and canned drinks products. Studies have recognized the impact of both on environmental issues; hence, an efficient waste management system is necessary. Waste segregation, recycling, and reduction have been prioritized in the circular economy to reduce environmental impacts. Intertwining technology that collects specific objects in exchange for rewards, such as through the reverse vending machine (RVM), effectively involves individuals in waste management. This paper uses Raspberry Pi to develop an image-based RVM for plastic bottles and aluminum cans. The system utilizes YOLOv5 to recognize the specified trash items accurately. A load cell is integrated to ensure these items are either empty or at least lightweight. The detection accuracy of the system yields a mean average recall of 96%, precision of 97%, and F-measure of 96%. Importantly, the image-detection feature allows the system to be easily scaled up without the need to restructure the whole hardware system, providing reassurance about its future-proof nature.