Analysis Performance of Optimization Method in ResNet50 for Cotton Leaf Disease Classification

Cotton plants are a large commodity. The cotton crop is one of the leading crops grown in several countries around the world, so in this case, crop loss leads to yield loss. There are various reasons for crop loss, but one of them is leaf disease. Therefore, disease classification was considered in this study. We analyze the performance of four optimization techniques in a simple transfer learning with the ResNet50 architecture. The investigated techniques are the Stochastic Gradient Descent (SGD), Root Mean Square Propagation (RMSprop), Adaptive Moment Estimation (Adam), and Adaptive moment estimation Extension based on infinity norm (Adamax). We train the model and calculate the accuracy rate for each optimizer using publicly available cotton leaf disease datasets. The results obtained from the experiment show that Adamax achieved better performance across the other three optimization techniques. Accuracy, accuracy, recall, and f1-score obtained using the Adamax optimizer are 100%, respectively.