Neural Network-Based Early Detection of Wheat Stripe Rust Disease for Enhanced Crop Management

This work focuses on the deep learning model intended to categorise wheat leaf photos depending on biotic and abiotic stress situations, namely nitrogen shortage and leaf rust, together with healthy leaf images.The main goal was to develop a strong and accurate model to improve precision farming methods by means of consistent and timely evaluations of crop condition.High-quality photographs were obtained with a Sony IMX363 RGB camera from a dataset gathered during the rabi season of 2019-20 from the Indian Agricultural Research Institute (IARI).The dataset included healthy leaves, leaves impacted by leaf rust, and nitrogen-deficient leaves, therefore guaranteeing a complete depiction of stress markers.To improve visibility of stress features, many preprocessing methods were used including Otsu-based background segmentation, contrast stretching, and Contrast Limited Adaptive Histogram Equalisation (CLAHE).To increase model resilience, rotation and scaling were used among data augmentation techniques.With hyperparameters painstakingly calibrated to maximise classification accuracy, the model architecture combined advanced ideas such residual and squeeze-excite blocks.Training, validation, and test sets-70:15:15-made up a balanced dataset split for the model.Accuracy measures were used in performance assessment to show a noteworthy capacity to separate stressed from healthy leaves.High classification accuracy of CropStressNet was shown, therefore enabling accurate identification of the stress conditions in wheat crops.This method helps to create more environmentally friendly farming methods in addition to provide understanding of crop health monitoring.The results highlight how deeply learning methods might be used to solve problems in precision farming.