ISIC2016_challenge

ISIC2016challenge

Motivation

These are just my trial in ISIC 2016 to surpass the top of the table.

Using Focalloss, the accuracy is higher than top of the table (2016). NTSNet is effective to this task.

Name		Name	Last commit message	Last commit date
Latest commit History 93 Commits
NTSNet		NTSNet
results		results
.DS_Store		.DS_Store
CAM.jpg		CAM.jpg
README.md		README.md
clustering.py		clustering.py
cutout.py		cutout.py
focalloss.py		focalloss.py
pytorch_CAM.py		pytorch_CAM.py
train_ISIC2016.py		train_ISIC2016.py