MUMBAI, India, May 1 -- Intellectual Property India has published a patent application (202641049892 A) filed by Easwari Engineering College; and Srm Institute Of Science And Technology, Ramapuram Campus, Chennai, Tamil Nadu, on April 20, for 'multi-task deep learning architecture for crop health diagnosis and dosage prediction.'
Inventor(s) include Dr. K P K Devan; Anandavalli; Dhana Anjana S; Sakthi Dharan M A; Sivakumar Balaji; Sree Simhan; and Vishnu Varthini M.
The application for the patent was published on May 1, under issue no. 18/2026.
According to the abstract released by the Intellectual Property India: "The invention discloses a multi-task deep learning architecture designed for comprehensive crop health diagnosis and dosage prediction. The system processes RGB images through a shared backbone and Bidirectional Feature Pyramid Network (BiFPN), enabling multi-scale feature extraction. From these features, specialized heads perform pest/disease classification, fine-grained disease identification, infection segmentation, severity estimation, and dosage regression. The architecture uniquely integrates visual diagnosis with agronomic prescription, producing outputs such as crop condition labels, infection masks, severity scores, and pesticide dosage recommendations. Training is conducted jointly on IP102 and PlantVillage datasets, supplemented with custom mask generation and severity labels, ensuring robust performance across diverse agricultural conditions. By unifying classification, segmentation, severity scoring, and dosage prediction into a single pipeline, the invention reduces pesticide misuse, improves yield, and provides farmers with actionable, data-driven decision support for sustainable agricultural management."
Disclaimer: Curated by HT Syndication.
