Classification of Pistachio Nut Using Convolutional Neural Network

Abstract

The application of innovative technologies in the agricultural industry has the potential to boost yield productivity and affect the well-being of farmers. Pistachio nuts are widely considered among the most precious things agriculture produces. The kirmizi and sirt are the two distinct varieties of pistachio nuts that are available. It is essential to categorize the different types of pistachio nuts to keep the product's quality and worth at a high level. This paper proposes a classified pistachio variety of kirmizi and siirt based on Convolutional Neural Network (CNN) models Inception V3 and ResNet50. The dataset used in this research is 2148 samples of pistachio images. The sample images are divided into 80% training data, 10% testing data, and 10% validation data. First, we pre-process and normalize by wrapping and cropping the images. The next, Inception-V3 and ResNet50 architectures, were trained and tested on the sample datasets. The experimental results show that the accuracy of both models is 96% and 86%, respectively. This can be concluded that the performance of the CNN model using Inception-V3 architecture outperforms ResNet50 architecture.