Potato Plant Leaf Disease Detection Distinctive Deep Attention Convoluted Network (DACN) Mechanism

Plant diseases, a leading cause of economic loss in the agricultural field, greatly impact crop productivity and quality. Detecting plant diseases at an early stage is very important for controlling such sicknesses effectively. It helps to protect crops in the best way and keep high levels of agriculture production. Taking action as soon as possible can reduce how fast diseases spread, lessen the need for chemical treatments and encourage farming methods that are sustainable—all this helps protect food supplies and keep financial stability within agriculture sector intact. The existing ways to diagnose plant leaf diseases greatly depend on people who are highly skilled examining samples manually. This method takes a lot of time, requires much work and can have errors. These factors make it difficult for quick and precise disease identification. Therefore, finding an automated system that is accurate and efficient in detecting and diagnosing plant leaf diseases early becomes crucial for enhancing both crop productivity as well as quality of harvests. For these difficulties, the solution is to create a new system for detecting the diseases in plant leaves. This method, known as Deep Attention Convoluted Network (DACN), uses computer vision and deep learning methods. Here, a CLAHE based preprocessing is applied to generate the contrast enhanced images with low noise artifacts. Then, a lightweight Attention Gate ResNet (AGRNet) architecture is implemented to segment the preprocessed plant leaf image with increased accuracy. Moreover, the Convoluted Deep Xception Net (CDXNet) based classification algorithm is used identify the infected plant leaf images with low training complexity and time consumption. During classification, the activation function of the CDXNet is optimally computed with the use of Q-Parameter updated Mountain Gazelle Optimization (QMoGO) algorithm. To verify performance, a range of parameter types are used to compare and validate the classifier’s prediction outcomes. The results from the tests were that the system made an outstanding precision of 99.2, 99.1, 99.15, and 99.2%. These results showcase the capability of DACN to identify the diseased leaves of the plants with almost 100% accuracy while using the inexpensive training and classification method. These excellent figures of accuracy are the confirmation of the reliability of the process and its possible application in a more widespread manner in detecting diseases in farm automation.