Development of an Intelligent System for the Automatic Classification of Plastic Bottles by Size at the Company INPLASET SAC

León León R.A.; López Sanchez M.A.; Reyes Cáceres R.C. · Memorias de la Conferencia Iberoamericana de Complejidad, Informatica y Cibernetica, CICIC (2025) · DOI: 10.54808/cicic2025.01.322

Resumen

This study aims to develop an intelligent system for the automatic classification of plastic bottles by size at INPLASET SAC, with a minimum expected accuracy of 90%, aiming to improve production efficiency and reduce error margins. The methodology included the collection and processing of a dataset of 2,649 images representing three bottle sizes. The images were labeled and preprocessed using Roboset to normalize their quality and resolution, facilitating proper model training. The YOLOv8 architecture was used in Google Colab, optimizing its accuracy and detection speed in identifying plastic bottles. After training, the system was implemented in PyCharm to perform real-time detection, evaluating the height and width of the bottles using computer vision techniques, including segmentation and edge detection, to classify the sizes. The results showed that the YOLOv8 model achieved 93% accuracy in bottle classification, displaying an optimal precision curve with high confidence levels.in the final printed copies of the Proceedings. Please keep in mind that the paper you prepare will be printed as it is received. Copyright 2025. © por el International Institute of Informatics and Systemics.

Instituciones: Universidad Privada del Norte