Application of convolutional neural networks for the classification of Camu Camu (Myrciaria dubia) according to its state of maturity

León León R.A.; Garcia Coronel E.A.; Vásquez Zevallos Y.R. · Memorias de la Conferencia Iberoamericana de Complejidad, Informatica y Cibernetica, CICIC (2025) · DOI: 10.54808/cicic2025.01.268

Resumen

The classification of Camu Camu (Myrciaria dubia) by maturity stage is essential in the agribusiness industry, as it influences the fruit's quality and commercial value. This project implements an automated system using convolutional neural networks (YOLOv8) to classify Camu Camu into three maturity stages: immature, ripe, and overripe, optimizing the harvesting and post-harvest process. With a dataset of 785 labeled images and high-performance GPU training, the model achieved an average accuracy of 99 % across maturity classes. The results show that the model can classify fruits accurately and reliably, outperforming manual methods. Implementation in a test environment confirms its applicability in real-world scenarios, enhancing operational efficiency and advancing computer vision in the agrifood industry. Copyright 2025. © por el International Institute of Informatics and Systemics.

Instituciones: Universidad Privada del Norte