Σύστημα πρόβλεψης για την ταξινόμηση διαφορετικών ειδών και ποικιλιών σπόρων δημητριακών
Prediction system for the classification of different species and varieties of cereal seeds
View/
Keywords
Συνελικτικά νευρωνικά δίκτυα ; Βαθιά μάθηση ; Μεταφορά γνώσης ; Ταξινόμηση σπόρων ; Μοντέλο VGG16 ; Μοντέλο ResNet50 ; Μοντέλο XceptionAbstract
Cereals play a crucial role in food supply globally. The quality of cereal seeds is of major importance within the grain chain industry, particularly in seed testing and certification procedures. Currently, this recognition and verification process relies on manual visual analysis of grains. However, an automated approach leveraging computer vision and machine learning classification has emerged as a faster and more efficient alternative method. Nevertheless, classifying cereals from different species accurately remains a complex and demanding task, especially at the varietal level. This study introduces a deep learning-based method designed to achieve precise classification for cereals of different species and varieties. Specifically, Convolutional Neural Networks (CNNs) were employed to classify wheat grain images into six different varieties (Krithari Triptolemos, Dimitra, Vromi Flega, Vromi Pigasos, Skliro Sitari Sellas and Levante). Additionally, three variations of commonly used CNN architectures were employed, namely VGG16, ResNet50 and Xception. To evaluate the effectiveness of the proposed models, a dataset comprising of 12.720 individual grain seed images, captured using a digital camera, was utilized. The results demonstrated test accuracies ranging from 65% to 97% for classification. The highest test accuracies were achieved using the ResNet50 and Xception architectures standing at 97%. Consequently, the outcomes of the proposed approach are both accurate and dependable, encouraging its practical implementation.
Postgraduate Studies Programme
Κυβερνοασφάλεια και Επιστήμη ΔεδομένωνDepartment
Σχολή Τεχνολογιών Πληροφορικής και Επικοινωνιών. Τμήμα ΠληροφορικήςNumber of pages
54Language
GreekCollections
Except where otherwise noted, this item's license is described as
Αναφορά Δημιουργού-Μη Εμπορική Χρήση-Όχι Παράγωγα Έργα 3.0 Ελλάδα
Αναφορά Δημιουργού-Μη Εμπορική Χρήση-Όχι Παράγωγα Έργα 3.0 Ελλάδα