Pistachio Species Classification: Leveraging Machine Learning for Quality Assurance and Market Optimization

Leveraging a large dataset of 2,148 high-resolution photos, this work proposes a fresh method for classifying pistachio species using a MobileNetV2 deep learning model. The suggested system seeks to automate the classification process, hence improving accuracy and efficiency over conventional hand sorting methods. The dataset consists of pictures of several pistachio varieties, so guaranteeing strong representation for good model training. After data preparation and augmentation, transfer learning was used to fine-tune the MobileNetV2 architecture producing a quite good classification model. The model had an overall accuracy of 94%; specific class measures show excellent performance. For several classes, specifically, precision and recall scores peaked around 99%; F1-scores across most species often exceeded 0.90. The classification report shows the success of the suggested model by displaying F1-scores demonstrating balanced performance in different species and accuracy values ranging from 0.81 to 0.99, so separating between various species. This study shows the possibilities of deep learning in agricultural uses, especially in optimizing pistachio processing and guaranteeing the quality of goods on the market. The results imply that the MobileNetV2 model not only simplifies classification problems but also provides a scalable solution fit for like agricultural difficulties.