Skin Disease classification using deep learning

• The algorithm will classify the type of skin disease based on the patient's pictures
  • Data set : 877 pictures from DermNet NZ
  • Data Split : Training: 80%, Validation: 10%, Test: 10%
  • Framework : Keras and Tensorflow
  • Neural Network Architecture : Convolutional Neural Networks
  • Accuracy : 62% in raw data and 75% after data augmentation
  • Impact : Automatically classify skin diseases through the input pictures and give a probability of each potential disease