Deep learning for skin lesion segmentation

Jun 1, 2018 · 1 min read
Top: skin images. Middle: Ground Truth. Bottom: Predictions.

Project Supervisor: Professor Jagath Rajapakse

Proposed a novel algorithm for the segmentation of skin lesions. Developed an image pre-processing pipeline, a modified deep learning architecture, and a post-processing method that gave state of the art results and showed an improvement of 7% compared to training the network without pre-processing, using the Keras library with a Tensorflow backend.

Joshua Peter Ebenezer
Authors
Staff Research Engineer at Samsung Research
My research interests include computational photography, image and video quality assessment, and deep learning.