Hybrid Deep Learning Model for Date Palm Disease Classification: A Fusion of HybridConv Mixer and Vision Transformer

Date palms sustain agricultural operations in dry areas, yet they encounter two serious diseases: brown spots and white scale, leading to harvest degradation and inferior production. The current practice of manual detection shows both inefficient processing along with substantial human error, thus requiring automated disease classification systems. The proposed research develops a disease identification system for date palms by merging the capabilities of Hybrid Convolutional Mixer (HybridConv) and Vision Transformer (ViT). The HybridConv Mixer focuses on detecting local disease patterns alongside ViT, enhancing global feature analysis, thus resulting in better disease classification. The trained model operated on brown spots and white scale, and healthy date palm leaf images from a dataset that received data augmentation for increased model reliability. The ensemble model demonstrates outstanding performance by reaching 99.89% accuracy, which outperforms single Convolutional Neural Networks (CNN) models in terms of precision, recall, and F1-score, thus providing promising technology for date palm cultivation disease detection automation.