CNN and Edge Detection Based Novel Leaf Disease Identification from Bush with Irrespective of Plant Types and Drone Footage-Based Real-Time Application

This study offers a Convolution Neural Network (CNN) based novel method for identifying leaf disease by using edge detection as a preliminary step. Existing literature focuses on the identification of leaf diseases irrespective of plant type. However, these models are specified for specific leaf types. Whereas, the proposed unique technique can efficiently identify leaf diseases for different types of plants by using a CNN without using transfer learning. Additionally, it is also capable of identifying defective leaves from bushes by tracking the leaf contour, which is rare in existing studies. This methodology holds significant potential for enhancing different types of crop health monitoring and reducing the impact of plant diseases on yield. To evaluate this a drone-based field survey is conducted where the captured image is utilized for real-time leaf disease detection. This algorithm can detect diseases with 93% accuracy, which shows the effectiveness of this algorithm in real-time applications.)