Anthracnose Blight Phytophthora Cucumber leaf Prediction through Six Convolutional ADAM Optimized Deep Convolutional Neural Network

Infections that harm the leaves of cucumber plants are referred as cucumber leaf diseases. These ailments can reduce cucumber yields significantly and have a detrimental effect on cucumber quality. Anthracnose, Blight and Phytophthora are a few typical cucumber leaf ailments. Rapid recognition and appropriate intervention of cucumber leaf disease detection are essential for effective control, but this can be challenging for farmers without access to special knowledge or equipment. This research proposes Six Convolutional ADAM DCNN to predict the cucumber leaf disease showing high efficiency. The proposed method utilizes the Cucumber leaf dataset from KAGGLE. The Cucumber leaf dataset having category of disease as Anthracnose, Blight and Phytophthora and Healthy with 4400 cucumber leaflets samples and applied for preprocessing. Single input, outcome, dense, and average pooling were used to create the six convolutional ADAM DCNN (6A-DCNN) that was suggested. Six convolutional layers and a maximum pooling layer associated with every convolution were used in the construction of the suggested 6A-DCNN model, which was then applied for optimization using ADAM optimizer. The cucumber leaf dataset was divided into 4000 training photos, 200 validation and testing images. For the purpose of evaluating performance indicators, the Cucumber leaf training dataset was applied to the proposed 6A-DCNN model as well as to the VGG, Resnet, MobileNetV2, DenseNet, Inception, Xception, and NASNet Large models. Using a Nvidia Geforce gtx V100 GPU server, 40 training epochs and a sample size of 96 were used while running Python. When compared to other CNN techniques, experimental findings reveal that the proposed 6A-DCNN exhibits the highest accuracy of 98.55%, precision of 98.22%, recall of 97.65%, and FScore of 97.67%.