Wheat Leaf Disease Detection Using CapsNet in Agri-Tech

Various diseases that affect the growth of wheat are important threats to food security in the world. This project aims at developing a system of wheat diseases diagnosis based on the integration of CapsNet into IoT. This dataset includes the images of different types of wheat diseases to include STB, stem rust, Septoria, tannin spot and powdery mildew, that affects Tunisia. Pre-processing and data augmentation bring in conformity to methods of training and improving the non uniformity and a diversified type of the dataset. CapsNet also has the feature called affiliative invariance which helps to solve the problem of varying camera angles and increases the model's accuracy. Even deeper capabilities can be achieved in transfer learning within hyperparameter tuning of the pre-trained CapsNets models. Applications of IoT empowers a continuous monitoring of the disease that affects crops, extensive and efficient spraying of biopesticides in agriculture. Wheat grains that are transformed into sensors are able to track changes in their environment; on the other hand, cameras of specific functions capture image data. Prompts and advisories from time to time assist farmers in preventing such challenges; hence, enhancing the productions and qualities. This work is helpful to agricultural in general and more specifically it is a supplement to the process of crop health monitoring and early signs of diseases detection, thus being a boost to global food security.