首页|Cultivar identification of pistachio nuts in bulk mode through EfficientNet deep learning model
Cultivar identification of pistachio nuts in bulk mode through EfficientNet deep learning model
扫码查看
点击上方二维码区域，可以放大扫码查看
原文链接
NSTL
Springer Nature
Abstract Robust and readily accessible identification systems for agricultural products have received increasing attention in inspection processes. The conventional visual inspection methods rely on human experts and could not be automated efficiently. This paper presents the application of the state-of-the-art Convolutional Neural Network (CNN) model to identify the top 4 most popular Iranian pistachio cultivars, including ‘Fandoghi’, ‘Kaleh-Ghouchi’, ‘Ahmad-Aghaei’, and ‘Akbari’. An image dataset was collected by acquiring pistachio samples images in the bulk mode under the condition of placing perpendicular to the camera with a fixed distance and natural lightening. The feature extraction block of EfficientNet-B3 was combined with a custom classification block. The proposed model was fine-tuned using the pistachio image dataset and adjusted to identify the various cultivars. The pistachio identification system was implemented using Python programming and the Keras API with TensorFlow Machine Learning framework. The results show that the classification accuracy of the best model evaluated on a hold-out test dataset is 98.00%. On the other hand, the proposed CNN model using EfficientNet-B3 presents average precision, recall, and F1-score values of 96.73%, 96.70%, and 96.67%, respectively. These results confirmed that the implementation of these methods as a mobile application and/or on an automated processing line could be a useful and effective approach to develop fast and robust food processing and inspection systems.
NSTL
Springer Nature
