Development of an artificial vision algorithm using image processing techniques to assess the characteristics of export-grade bananas

The main objective of this study is to develop a computer vision algorithm capable of identifying export-quality bananas, simplifying the quality control process. Unlike previous research, such as that of Figueroa and Roa (2016), which focused on different fruits using segmentation and clustering techniques, this work integrates a controlled lighting system and advanced neural networks trained with TensorFlow. These innovative approaches ensure an average accuracy of 97.22%, surpassing previous standards in the field and demonstrating their applicability in industrial settings.