Many people face a lot of problems with their skin. Many are suffering from various skin diseases, which have always been a common complication in humans. Traditional treatments for diagnosing skin problems require a lot of tests and the diagnosis is seen to be time-consuming and requires an extensive understanding of the domain. Visual assessment in combination with clinical information can be helpful for the diagnosis. The approach involves the development using Convolutional Neural Networks (CNN) and an ensemble model using VGG16, DenseNet and Inception. Specific skin diseases namely Actinic Keratoses, Basal Cell Carcinoma, Benign Keratosis, Dermatofibroma, Melanoma, Melanocytic Nevi and Vascular Lesions are considered. Results show that using CNN, the accuracy ranges from 71%-75%, VGG16 has attained an accuracy of 80.3%, DenseNet with 82.3%, Inception with 80.4% and an ensemble of VGG16, DenseNet and Inception has achieved an accuracy of 83%-85%.