Deep-PDSC: A Deep Learning-Based Model for a Stage-Wise Classification of Parlatoria Date Scale Disease

Precision agriculture is one of the solutions that has garnered significant attention from specialists recently, as it enables automatic detection and classification of plant and crop diseases, thereby increasing crop quality and yield, and it is also viewed as a tool for taking proactive measures to mitigate natural disasters, whether related to climate change or damage caused by certain insects. In this regard, our current study intends to develop a model for the automatic identification and classification of date palm disease caused by the Paralatoria Date Scale insect using deep learning techniques. A novel approach based on the combination of two deep learning algorithms is proposed. The first is a VGG16-based backbone, which was used as a feature extractor, and the second is an improved Fully Connected Neural Network, which was used as a classifier. The proposed model was trained using date palm leaflet RGB images of four categories, three of which are damaged to varying degrees, and one of which is healthy. A comparative study was conducted and demonstrated that the proposed model outperformed with an average accuracy of 98.06%, outpacing existing solution and some state-of-the-art methods. This study is meant to help the development and improvement of oasis agriculture, which relies heavily on palms and dates. It can help farmers and farm owners take preventative measures quickly and thus ensure a high-quality crop.