Illness directly affecting the skin is the fourth most frequent cause of all human disease, and is seeking the attention of researchers. In this research work, one such effort is made by proposing a mobile-enabled expert system named “i-Rash” for the classification of inflammatory skin lesions. i-Rash can classify the skin image into one of the four non-overlapping classes, i.e. healthy, acne, eczema, and psoriasis. The classification model for i-Rash is trained using deep learning model SqueezeNet. The pre-trained SqueezeNet is re-trained on the skin image dataset using transfer learning approach. The i-Rash classification model is trained and tested on 1856 images. The trained model is only of 3MB size and is capable of classifying an unseen image in a fraction of seconds with an accuracy, sensitivity, and specificity of 97.21%, 94.42% and 98.14% respectively. i-Rash is based on a client-server architecture and can serve in initial classification of skin lesions, hence, can play a very important role in minimising the global burden caused by skin diseases.
|Number of pages||6|
|Publication status||Published - 22 Aug 2019|
|Event||2019 International Conference on Computing, Electronics & Communications Engineering (iCCECE) - London, United Kingdom|
Duration: 22 Aug 2019 → 23 Aug 2019
|Conference||2019 International Conference on Computing, Electronics & Communications Engineering (iCCECE)|
|Period||22/08/19 → 23/08/19|