A Hybrid Deep Learning Web Platform for Cotton Crop Monitoring and Disease Management

Abstract

Cotton crop productivity is significantly affected by leaf diseases, which can reduce both crop quality and agricultural yield. Early and accurate disease detection is essential for effective crop monitoring and management in smart agriculture. This research presents a hybrid deep learning web platform to monitor cotton crop and disease management. The proposed system is developed to classify cotton leaf images into three categories: Bacterial Blight, Cotton Curl Leaf Disease, and Healthy Leaf. The developed platform integrates deep learning and machine learning techniques within a web-based environment to provide automated disease diagnosis and management support for farmers and agricultural experts. Several state-of-the-art models, including MobileNetV2, EfficientNetB3, EfficientNetV2S, Random Forest Classifier, and proposed hybrid architectures, were evaluated for disease classification performance. Experimental results show that the proposed HybridEfficientRForest model achieved the highest performance with 98% accuracy. Similarly, the proposed HybridEfficientMobiNet model achieved 92% classification accuracy with strong overall performance. In comparison, standalone models produced comparatively lower results. The proposed web platform underscores the significance of hybrid deep learning approaches in precision agriculture by providing a user-friendly, scalable, and intelligent platform for real-time monitoring and management of cotton crop diseases.

Keywords: Cotton Diseases, Cotton Crop Imaging, Deep Learning, Transfer Learning, EfficientNet, Disease Management.