Intellectual Property India Publishes Patent Application for Yolo-Based Feature Learning With Rf–xgb Ensemble For Enhanced Citrus Leaf Disease Detection

MUMBAI, India, July 7 -- Intellectual Property India has published a patent application (202641081100 A) filed by Sasi Institute Of Technology & Engineering; V. Ajay Kumar; P. Vijay Kumar; Mvss Nagendranath; and Ev. Sandeep on July 01, 2026, for Yolo-Based Feature Learning With Rf–xgb Ensemble For Enhanced Citrus Leaf Disease Detection.

Inventors include Sasi Institute Of Technology & Engineering; V. Ajay Kumar; P. Vijay Kumar; Mvss Nagendranath; and Ev. Sandeep.

The application for the patent was published on July 03, 2026, under issue no. 27/2026.

Abstract: Citrus leaf diseases significantly affect fruit production worldwide, leading to reduced yields and substantial economic losses. Early detection and accurate assessment of disease severity are essential for effective crop management and treatment. This study presents an automated citrus leaf disease detection and severity classification system using a combination of deep learning and machine learning ensemble techniques. The proposed framework employs the YOLO algorithm to detect and localize diseased regions on citrus leaves in real time. After disease identification, a hybrid ensemble classifier integrating Random Forest (RF) and Extreme Gradient Boosting (XGBoost) is utilized to evaluate the severity level of the detected disease. The model was trained and tested on a dataset containing citrus leaf images representing multiple disease categories and varying severity levels. Experimental results demonstrate that YOLO effectively identifies diseased regions with high detection performance, while the RF–XGBoost ensemble achieves superior severity classification by efficiently handling complex feature relationships and noisy data. Performance evaluation using accuracy, precision, recall, and F1-score confirms the robustness of the proposed approach. The ensemble classifier achieved a maximum accuracy of 96%, outperforming traditional classifiers such as Logistic Regression (LR) and Decision Tree (DT) models. The proposed system provides farmers and agricultural experts with a reliable tool for rapid disease monitoring and severity assessment in citrus plantations. Furthermore, the framework supports the advancement of smart agriculture through automated crop health monitoring, mobile-based diagnosis, and large-scale plant disease surveillance.

Disclaimer: Curated by HT Syndication.