In this project we classified melanoma pictures into Low Risk or Melanoma.Melanoma classification is finished utilizing Convolutional Neural Network (CNN), and build a small web interface to upload an image and get the results.

Melanoma is a type of cancer that begins in the pigment cells (melanocytes) of the skin. Melanoma disease appears on the skin as pigmented moles or marks. It can also spread to other body organs. Melanoma can be caused by the excess exposure to ultraviolet radiation from the sun. The fact that melanoma body marks can be confused with normal pigments of the skin makes it hard to classify the skin pigment into benign or malignant. Having an automated algorithm to classify melanoma images and breast cancer tumors will support early diagnosis and help improve cancer detection performance. For an accurate diagnosis, medical experience is fundamental in the diagnostic test analysis, and especially, to determine the cancer stage. The identification of cancers depends on the physicians interpretation from information obtained from the patients through examinations, and correct diagnosis in a premature state of the cancer can aid in decision making, action planning and treatment efficiency.

You may download the dataset form the above challenge or Kaggle hosts it too.

The model was trained in Google Colab (recommended). And the webpage is built using:

• Flask
• HTML
• CSS

Save the model, and run

python3 upload.py