A Hybrid U-Net with Active Contours for Plant Leaf Disease Segmentation and Classification

• Published in: 2024 International Conference on Advancement in Renewable Energy and Intelligent Systems (AREIS) pp. 1 - 6
• Main Authors: Muthukumar, P., Eswaramoorthy, Muthuvel, K., Boopathi, R., Suganyadevi, M.V., V. K., Reshma
• Format: Conference Proceeding
• Language: English
• Published: IEEE 05.12.2024
• Subjects: Accuracy; Active contours; Adaptive Bilateral Filtering; Filtering; Hybrid U-Net with Active Contours; Image Processing; Image segmentation; Neural networks; Plant Disease; Plant diseases; Real-time systems; Renewable energy sources; Soil; Weather forecasting
• DOI: 10.1109/AREIS62559.2024.10893656

The economic growth of a nation entirely depends upon the agriculture and agricultural products. In developing countries like India, agriculture is the primary source of income and its contributing 17% to the total GDP. There are plenty of factors lead to plant disease which impacts the quality and yield of plants. Though manual method of detection is time consuming and it may have chance for errors. This method is not enough to identify and limit the spread of plant disease. To establish an automated plant disease detection in farms will reduce the risk of plant disease and promotes real-time monitoring of crops in a daily basis. Artificial Intelligence (AI) took part in supporting farmers to get instant solution in selection of fertilizers, classifying the quality of agricultural products, weather prediction and soil nutrient level detection. In this study, we proposes a novel segmentation approach namely Hybrid U-Net with active contours to segment the disease affected portion on leaf which support farmers to identify disease at early stage. This study provides a comprehensive analysis of plant disease segmentation by proposed method with conventional approaches. The publically available dataset is chosen for this analysis and performance of conventional studies was compared. This study presented current trends of plant disease segmentation and several other image classification techniques. Experimental results evaluating that the proposed study improved segmentation better than conventional methods.