Vision-Based Plant Disease Identification for Autonomous Crop Management Systems

Le résumé fourni par la source

The integration of robotics into agriculture is transforming traditional farming practices by enabling intelligent systems to address challenges in plant health monitoring and disease detection. These advancements are essential for precision farming, as they improve productivity, reduce resource wastage, and mitigate crop losses caused by diseases. In this context, autonomous systems equipped with robust disease identification capabilities are crucial for scalable and sustainable solutions. This work proposes a novel approach for plant disease identification leveraging the ConvNeXt deep learning model to extract and classify visual features of plants across different species and disease types. Our experiments utilize two well-established image datasets, the Plant Village and the Plant Pathology 2020, which encompass a diverse range of images representing healthy and diseased plants. The proposed method demonstrates high accuracy in identifying plant diseases, achieving 99.47% of accuracy in the Plant Village dataset and 93.83% of accuracy in the Plant Pathology 2020 dataset, outperforming comparative approaches. Notably, the results highlight the robustness of our approach across datasets with varying features, making it suitable for scalable deployment in autonomous robotic systems. This contribution emphasizes the potential of deep learning-driven solutions to advance agricultural robotics, promoting sustainable and efficient farming practices.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.