Intellectual Property India Publishes Patent Application for A Multi-Task Learning Framework For Tomato Leaf Disease And Severity Prediction

MUMBAI, India, Aug. 24 -- Intellectual Property India has published a patent application (202621081464 A) filed by Symbiosis International Deemed University on July 01, 2026, for A Multi-Task Learning Framework For Tomato Leaf Disease And Severity Prediction.

Inventors include Anusri Kadam; Parnika Jain; Srishti Tripathi; Tushar Garg; Shruti Sunnad; Dr. Aniket K. Shahade; Dr. Priyanka V. Deshmukh; and Dr. Ketan Kotecha.

The application for the patent was published on August 21, 2026, under issue no. 34/2026.

Abstract: ABSTRACT A MULTI-TASK LEARNING FRAMEWORK FOR TOMATO LEAF DISEASE AND SEVERITY PREDICTION The present invention discloses a multi-task learning framework (100) for simultaneous tomato leaf disease classification and severity prediction. The framework (100) employs a shared ResNet50 backbone (122) with CBAM attention (124), a disease classification branch (130), and a severity prediction branch (140) with a cross-task attention mechanism (144) that incorporates disease-specific features into severity prediction through learnable projections with gradient detachment. A multi-task loss module (150) with a masking strategy handles partially labeled data. The framework achieves 97.85 percent disease classification accuracy and significantly improves severity estimation from 0.32 to 0.75 F1-score compared to single-task approaches, supporting precision agriculture applications. [

Disclaimer: Curated by HT Syndication.