Implementation of a Computer Vision System for Cucumber Classification Based on Quality Standards in the Agro-Food Industry

León León R.A.; Concepcion Vazquez A.E.; Valqui Moya D.J.; Bailon Diaz M.M.; Huaman Mosqueira F.J.; Salcedo Marreros J.C. · Memorias de la Conferencia Iberoamericana de Complejidad, Informatica y Cibernetica, CICIC (2025) · DOI: 10.54808/cicic2025.01.244

Resumen

The objective of this work was to implement a computer vision system based on convolutional neural networks (CNN) and the YOLOv8 model to classify cucumbers according to quality standards in the agrofood industry. A total of 869 cucumber images were collected in real production conditions, which were labeled and processed using the COCOdataset platform. Using Google Colab and PyTorch, the YOLOv8 model was trained, with performance optimized through image preprocessing techniques and data augmentation. The results showed a 95% accuracy in cucumber classification. This high accuracy supports YOLOv8's capability to detect and classify objects in real-time, proving to be an effective and precise solution for automating the selection process in the agro-food industry. Copyright 2025. © por el International Institute of Informatics and Systemics.

Instituciones: Universidad Privada del Norte