MUMBAI, India, Aug. 17 -- Intellectual Property India has published a patent application (202641095198 A) filed by Neethu C T; Dr. Leena Mary; Dr. Lidiya Lilly Thampi; and Indian Institute Of Information Technology Kottayam on August 06, 2026, for Morphology-Aware Dual-Stream Transformer-Based Explainable System For Automatic Identification Of Hevea Brasiliensis Clones Using Leaf Morphological Characteristics And Method Thereof.
Inventors include Neethu C T; Dr. Leena Mary; and Dr. Lidiya Lilly Thampi.
The application for the patent was published on August 14, 2026, under issue no. 33/2026.
Abstract: A morphology-aware dual-stream transformer-based explainable system (100) for automatic identification of Hevea Brasiliensis clones using leaf morphological characteristics is disclosed. The system (100) includes a dataset pre-processing and image conditioning module (102) configured to generate standardized leaf images and a Morphological feature encoding module (126) configured to extract anatomically significant leaf regions. A global contextual feature learning module (106), a local contextual feature learning module (108), and a Morphological local feature extraction module (128) generate complementary feature representations, while a bidirectional cross-attention fusion module (110) integrates global and localized morphological information. An embedding and classification module (112) identifies Hevea Brasiliensis clones, and a morphology-aware explainability module (114) generates biologically interpretable visual explanations.
Disclaimer: Curated by HT Syndication.