Deep Learning based Multiple Skin Disease Classification in Indian Territory

Skin problems can range from completely painless to Iifethreatening, and from minor to life-changing. The patient has trouble in understanding the complexity and uncertainty associated with the problem the doctor is treating. As a result if left untreated or ignored, it can be fatal. To reduce their prevalence and severity, there must be a reliable method of early diagnosis for skin diseases. In this paper, the effectiveness of the DenseNet Convolutional Neural Network (CNN model is analyzed for multiple skin disease detection. Compare the similarities and differences between two CNN models (DenseNetl21 and DenseNet201). Introduce the dataset Skin_Disease_v1 for this experiment. Based on the results, Densenet201 performs exceptionally well when used for multiple skin disease classifications. In the Fold3 dataset the Densenet201 model achieves a best-case accuracy of 79.26%. The outcomes of 10 randomly split datasets are also accomplished. DenseNetl21 is an average accuracy of 69.47 ± 8.78, whereas DenseNet201 average accuracy is 73.52 ± 7.88.