Conference paper

Selection of the Most Suitable Parameters for Classification of Hazelnut Fruit Using ShuffleNet Deep Learning Algorithm

Özgür Tomak, Mehmet Cem Dikbaş, Sercan Külcü

11th International Azerbaijan Congress on Life, Engineering, Mathematical, and Applied Sciences, Congress Proceedings Book, Vol. I, pp. 64–73 · BZT Turan Publishing House · ISBN 978-9952-8573-6-8

Abstract

It may be possible to increase efficiency or improve quality control processes by using artificial intelligence to classify hazelnut fruits. Our aim in this study is to select the appropriate parameters in the classification of hazelnut fruits using the Shufflenet deep learning algorithm. ShuffleNet is a convolutional neural network designed for devices with limited processing power and uses fewer resources by reducing the number of parameters and computational load. In the data used, there is one hazelnut in each image. These hazelnuts are divided into three classes as good and bad hazelnuts and inner hazelnuts. A data set was created using 519 bad hazelnuts, 535 good hazelnuts and 523 inner hazelnuts. A total of 15770 images were obtained with 10 images taken from different angles from each hazelnut. As a result of the analysis, ShuffleNet with SGDM reached 99.79% accuracy in training and 99.94% in the test phase. ShuffleNet with ADAM reached 99.89% accuracy in training and 99.96% accuracy in the test phase. ShuffleNet with RMSprop reached 99.89% accuracy in training and 99.92% accuracy in the test phase. According to the complexity matrices, ShuffleNet with SGDM has 0.9997 precision, 0.9994 specificity, and 0.9997 F1-score. ShuffleNet with ADAM has 0.9997 precision, 1 specificity, and 0.9998 F1-score. With RMSprop, ShuffleNet has a sensitivity of 0.9997, a specificity of 0.9998, and an F1-score of 0.9997. Although the overall results are close, it has been observed that ShuffleNet algorithm with ADAM gives better performance.

Keywords: Hazelnut, Classification, Deep Learning, Shufflenet

How to cite

Tomak, Ö., Dikbaş, M. C., & Külcü, S. (2025). Selection of the Most Suitable Parameters for Classification of Hazelnut Fruit Using ShuffleNet Deep Learning Algorithm. In 11th International Azerbaijan Congress on Life, Engineering, Mathematical, and Applied Sciences, Congress Proceedings Book, Vol. I, pp. 64–73. BZT Turan Publishing House.

© 2025 Sercan Külcü. All rights reserved.