Scholarly record

THE OPTICAL METHOD FOR THE PLASTIC WASTE RECOGNITION AND SORTING IN A REVERSE VENDING MACHINE

Andrey N. Kokoulin

Abstract

Reverse vending machines (RVM) are a key part of used plastic containers utilization system in Europe and the United States. Waste recognition and sorting in RVM machines can be performed by any of the following procedures: by determining the container material (e.g. using the IR-spectrometer), by recognition of the container type by its shape, or by the barcode identification.[1] These three basic control-procedures make any attempt of the fraud completely impossible. But at the same time, it makes the RVM too expensive. With the modern computer vision technologies, we can design another kind of efficient and non-expensive RVM having the same functionality using energy-efficient IoT MCUs. In this paper an efficient approach of computer vision and image processing application in automatic recognition of empty recyclable containers is considered. The RVM construction was optimized to be as small as possible due to binocular optical system and the human-machine interaction of RVM was reduced due two levels of data processing (distributed functionality) making it possible to embed the module of RVM into the vending machine. The list of the available object recognition methods and frameworks was revised because IoT controllers and tiny single-board computers usually have memory and computational restrictions. CNN training takes into account that recycled cans or bottles could be twisted or jammed. Also, we analyze the performance of image recognition procedures in Python and C ++ languages.

Publication details

Back to publication list