Exploring Deep Learning Models for the Identification and Segmentation of Leaf Diseases

Agriculture plays a main role in our economy. Amid the worsening of agricultural problems, prompt and accurate identification of illnesses connected to plants has become a very important component of proper crop development and obtaining maximum yield. The current study deals with the practice of machine learning and deep learning methods in the Identification of diseases using the image data from the mobile and drone cameras and along with these data the environmental data that are acquired through IoT sensors. In order to make crop production sustainable, it is imperative to promptly and accurately diagnose unhealthy plants due to the rapidly expanding agricultural issues through algorithms such as Convolutional Neural Networks (CNNs), Dense Net, etc. in predicting the disease. The different data processing methods such as image enhancement, normalization, and selection of features like Principal Component Analysis (PCA) and Gated Recurrent Multi-attention neural network (GRMA-Net), which are used to optimize the productivity of disease detection. Moreover, it looks into the impact of data fusion from the images and sensors on the classification accuracy. The different classifiers such as Support Vector Machine (SVM) and Multi-Layer Perceptron (MLP) and compares with an advanced techniques like Resnet50, VGG, Efficient Net etc., on the basis of certain statistical indicators. This study draws attention to the merits of associating the information obtained from IoT to Deep learning which as a result will increase farming productivity and result in fewer chemicals used. The outcomes indicate the great possibility of the treatment of a disease in early stage, the decreasing of the use of chemicals, and the increment of the security of the global food supply at the long-term.