DADDNet: A Desert-Aware Deep Learning Approach for Potato Leaf Disease Detection

This paper explores the growing application of artificial intelligence within the agricultural sector, specifically addressing the challenge of accurate plant disease detection in desert environments. Conventional deep learning techniques often fail to meet the rigorous demands of arid regions with extreme conditions, necessitating novel, domain-focused solutions. Accordingly, a Desert-Aware Disease Detection Network (DADDNet) is proposed, combining transfer learning, spatial attention, and data augmentation tailored to harsh desert conditions. A carefully curated dataset of potato leaf images, capturing healthy, early, and late infections, underpins the study. Experimental results show an impressive accuracy rate of 91% in training and 88% in validation, supported by consistently low loss values, indicating balanced generalization. These findings highlight the system’s potential to enhance early disease identification, mitigate crop losses, and improve sustainability in resource-limited agricultural settings.

• potato leaf disease detection
• plant disease classification
• desert agriculture
• arid regions
• transfer learning
• spatial attention
• data augmentation
• early disease detection

• 2 — Faim « zéro »
• 9 — Industrie, innovation et infrastructure

• Smart Agriculture and AI
• Advanced Neural Network Applications
• Industrial Vision Systems and Defect Detection
• potato leaf disease detection
• plant disease classification
• desert agriculture
• arid regions
• transfer learning
• spatial attention
• data augmentation
• early disease detection
• Plant Science
• Computer Vision and Pattern Recognition
• Industrial and Manufacturing Engineering
• Agricultural and Biological Sciences
• Computer Science
• Engineering