Skip to main navigation menu Skip to main content Skip to site footer

Skin Lesion Segmentation using Deep Hypercolumn Descriptors


We present a image segmentation method based on deep hypercolumn
descriptors which produces state-of-the-art results for the
segmentation of several classes of benign and malignant skin lesions.
We achieve a Jaccard index of 0.792 on the 2017 ISIC Skin
Lesion Segmentation Challenge dataset.