DeepDate: A deep fusion model based on whale optimization and artificial neural network for Arabian date...

Area of Science:

• Agricultural Technology
• Computer Vision
• Artificial Intelligence

Background:

• Increasing scale of date fruit production necessitates efficient classification methods.
• Traditional classification methods struggle with large volumes and diverse species.
• Image recognition offers a potential solution for automated date fruit identification.

Purpose of the Study:

• To develop and evaluate a deep learning model for classifying Arabian date fruit species.
• To improve the accuracy and efficiency of date fruit classification.
• To address the challenges posed by increased yields in the agricultural sector.

Main Methods:

• Proposed a deep fusion model (DeepDate) combining whale optimization and artificial neural networks.
• Utilized a dataset of five Arabian date fruit image classes: Barhi, Khalas, Meneifi, Naboot Saif, and Sullaj.
• Employed a three-phase process: feature extraction, feature selection, and model training/testing.

Main Results:

• The DeepDate model achieved a highest test accuracy of 95.9%.
• DeepDate demonstrated a superior balance between classification accuracy and time consumption compared to established models.
• Outperformed several deep transfer learning models including Alexnet, Squeezenet, Googlenet, VGG-19, NasNet, and Inception-V3.

Conclusions:

• DeepDate significantly improves accuracy and efficiency in date fruit classification.
• The model offers a promising AI-driven solution for the agricultural industry.
• Recommendations include technology transfer and collaborations to enhance sustainability and productivity in date fruit production.