Cultivar Identification of Pistachio Nuts via Deep Learning

Asma Shams-kermani (PhD)^1*

¹ Department of Electrical Engineering, Faculty of Electrical and Computer Engineering, Sirjan University of Technology, Sirjan, Iran

Received: 14.04.2024 Accepted: 15.06.2024

Abstract

Background: This study presents a deep learning approach for identifying pistachio cultivars using sophisticated image classification techniques.

Materials and Methods: We utilized the YOLOv8 convolutional neural network, recognized for its swift training and testing capabilities, as well as its outstanding single-object classification accuracy. Unlike conventional methods that assess individual pistachio nuts, our strategy analyzes images with multiple nuts at once, greatly improving both efficiency and speed.

Results: We rigorously tested this method on a detailed dataset of images featuring five commonly grown Iranian pistachio cultivars: Badami, Fandoghi, Kalleh Ghoochi, Ahmad Aghaei, and Akbari. Our findings revealed an impressive average classification accuracy of 99.8% on the test set, highlighting the robustness and effectiveness of our approach.

Conclusion: This method marks a significant advancement over traditional techniques, providing a highly reliable, automated, and efficient solution for identifying pistachio cultivars, with extensive practical applications in agricultural sorting systems and the food processing quality control industry.

►Please cite this article as follows: