Agrosagenet: Self-Adaptive Intelligence to Address Global Food Demand and Crop Disease Detection and Diagnosis for Sustainable Precision Agriculture

Abstract

The escalation of global food demand will require a complete rethinking of the detection and management of crop diseases. In this paper, we introduce AgroSageNet, a new self-adaptive intelligence adaptive learning framework to support automated detection and diagnosis of crop diseases in precision agriculture. Unlike traditional machine learning and deep learning models, AgroSageNet uses Adaptive Self-Attentive Morphological Encoding (ASAME) and Causal Temporal Disease Progression Estimation (CTDPE), to interpret crop health dynamically and contextually. AgroSageNet provides the ability to autonomously identify known and new diseases using multi-spectral images and data recognition of phenotypic patterns without the need for pre-labeling or manual feature engineering. Results from assessments on curated real-world datasets, representing five major crop types, indicates a diagnostic accuracy of 97.2%, significantly outperforming typical convolutional and transformer-based models while maintaining real-time inference capabilities for edge deployment. AgroSageNet illustrates a shift toward intelligent agricultural disease monitoring that is resilient in field conditions, ultimately creating a pathway to support sustainable high precision farming practices.