Skin Lesion Segmentation

Skin Lesion Segmentation using both Deep Learning and normal Image Processing approaches.

Used the PH2 Dataset for the task.

---

The Deep Learning Approach

Used FCNet to train the model on the images and their corresponding masks.

The data required was generated by a custom script: Copy_file.py

The Image Processing Approach

Used Median filtering, Image thresolding and Canny edge detection to generate contours using hand-crafted features.

---

References:

• https://towardsdatascience.com/a-keras-pipeline-for-image-segmentation-part-1-6515a421157d

• https://www.tensorflow.org/tutorials/images/segmentation