The main objective is to devise an efficient and computationally optimum approach by improving the existing classification model in the field of the Medical World more precisely Dermatology Department for improved diagnosis followed by better medical assistance. Skin diseases are a common problem affecting a lot of people, and lack of proper assistance can lead to chronic problems like acne, vascular lesions, Eczema, etc. With evolving computational performance and algorithms, classification of the disease can be automated and deployed with the live stream using Computer Vision techniques. This project aims toward the classification of 7-types of common pigmented skin lesions using a Neural Network architecture comprising of CNN, Dropout, MaxPooling, and Dense layers and precisely tuned hyper-parameters like Learning Rate, Early stopping, Loss function, etc.
Dataset Images:
Project Structure and Pipeline:
