A Novel Fusion Study on Disease Detection in Cotton Plants Using Embedded Approaches of Neural Networks

Abstract

If not detected and managed correctly, Cotton diseases can lead to substantial economic losses for farmers. By implementing an automated disease detection system, farmers can save costs by reducing the reliance on manual scouting and labor-intensive inspections. It enables more efficient use of resources by focusing treatments only on affected plants instead of the entire field. Image segmentation methods may need help to generalize well across different datasets or disease types. The performance of segmentation algorithms can be affected by variations in leaf shape, disease symptoms, or even different camera setups. Training and validating the segmentation models on diverse datasets representing various disease severities and environmental conditions can help mitigate this limitation. When trained on large and diverse datasets, deep learning models can generalize well to unseen data. Once trained on a representative dataset, the models can generalize to new cotton leaf images and detect diseases even in different environments or with variations in lighting conditions, leaf shapes, or disease severities. Deep learning approaches promise precision farming. Cotton leaf disease detection in precision farming enables optimized resource management. This paper studies the different approaches for disease detection in cotton plants. Out of the many approaches, most optimization results are obtained by the deep learning approaches.