The skin is the biggest organ of the human body which protects our inner vital parts and organs from the outside environment. However, there are a number of diseases that affect the human skin such as, fungus, bacteria, allergies, enzyme, and viruses that have caused to skin abnormalities and need to be treated at earlier stages to avoid it from spreading. Identifying the disease type based on manual feature extractions or the symptoms is time consuming and requires extensive knowledge for perfect identification. The main objective of this study was developing human skin disease detection and classification model using deep convolutional neural network. The images of human diseased skin Tinea capitis, Tinea pedis Tinea corporis and Healthy skin have been collected from Tibebe Giyon Specialized Hospital, Felege Hiywot Comprehensive Referral Hospital, Gamby General Teaching Hospital using Techno CAMONX smartphones camera 14 mega pixel and DermNet.com image repository in jpg file format. In this study, a total of 4 classes and a total of 2,226 images are used. After collected the dataset, image augmentation, image preprocessing, thresholding segmentation, combined (CNN and BRISK) had been used as feature extraction techniques and SVM and SoftMax for classifier. And also, the researcher has applied Principal Component Analysis (PCA) to reduce the dimension of the combined features. The researcher has used MATLAB R2019a programming tools for overall coding mechanisms. From the experiment, the model achieved the testing accuracy of 88.9% and training accuracy of 98.44%. Key words: CNN, SVM, Local Feature Descriptor, Feature Extraction, PCA