Deep Learning for Multipurpose Plant Disease Detection and Recognition

The objective of this work is to design a plant leaf disease detection system using deep learning models such as Mobilenet, ResNet50, and InceptionV3. This includes getting a tagged dataset of plant pictures, preprocessing the information, and then leveraging the benefits of transfer learning using the pre-trained weights of the models. The final layers of the models designs are modified considering the number of classes in the dataset. Models performance is evaluated after training by using test data. Implementation is based on Tensor Flow and Keras frameworks. Models produced have wide-variety-real applications, namely online as well as mobile platforms. The application was developed with Tensor Flow Lite and Flutter, considering the development of Android mobile applications, and it showed the flexibility of the models toward the mobile platform and transformed the study into a practical utility. This research highlights the importance of maintaining eyes on the system, which further calls for changes if necessary so that the associated performance can be maximized in the production environment. To put it shortly, work done in this paper describes some automated techniques used to identify plant diseases. Systems like these will be the effective and practical solution for early recognition and correction of diseases in the agricultural sector.