A disease monitoring system using multi-class capsule network for agricultural enhancement in muskmelon.

For any agricultural society, the well-being of the plants is crucial to achieve a greater yield. The health and vigor of plants play a pivotal role in shaping the ultimate outcome of crop production. There are, too many infections affecting the plants generate harm to diverse economies and communities. It can also result in significant environmental losses. To prevent such losses, it is easier to diagnose diseases correctly and promptly at an early stage of plant life. This research mainly focuses on Muskmelon leaf diseases. Muskmelon is a remunerative crop with a short life span of around 65 days. Any disease attack in this duration will affect the crop entirely which in turn leads to yield loss. Hence, there needs an early prediction system for predicting diseases. The primary goal of this research is to develop a prediction model based of Multi Class Capsule Network for early detection of disease and pest in plants. The performance indicators examined for classification of leaf diseases are Accuracy, Precision, Recall, F1 score and, Loss function. The performance of Multi – Class Capsule Network [MCCN] is compared with existing pre-trained models such as, AlexNet, ResNet, VGG16, VGG19, GoogleNet, and CapsuleNet. Experimental results indicated that the MCCN model performs with an accuracy of 97.30% which is better than the accuracy of other models under considerations. [ABSTRACT FROM AUTHOR]