Classification of Polyethylene Terephthalate Bottles in a Recycling Plant

GODOY, DIEGO; ALBORNOZ, E. M.; SELVA, R.; IBARRA, N.; GALLARDO, C.

Cuenca

Congreso; 5th Ibero-American Congress of Smart Cities, (ICSC-CITIES); 2022

Universidad de Cuenca, Ecuador.

One of the more serious problems that the planet faces is the pollution, which exposes its integrity and the living beings that inhabit it at risk. Pollution is a constantly growing problem due to the increase in population and large companies, and it contributes to the detriment of people’s health and the environment. The advancement of technologies, and in particular of intelligent systems, allows automating and optimizing many processes that are usually carried out manually and very expensive. The artificial intelligence systems allow the redistribution of efforts and resources, increasing productivity. In this work, an approach to the detection of different types of recyclable plastic waste that arrive at a recycling plant using a convolutional neural network is presented. Several models were evaluated for this task and the YOLOv5 achieved the best results. Preliminar results show a confidence of up to 65% on images taken from the recycling plant in the city of Posadas, Misiones (Argentina).