ISIC 2019 Leaderboards
| Rank 16 total | Team 16 unique teams | Approach | Manuscript | Used External Data 6 yes | Primary Metric Value Balanced Accuracy | ||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 |
DAISYLab
Hamburg University of Technology/University Medical Center Hamburg-Eppendorf
|
Ensemble of Multi-Res EfficientNets with Meta FC-NN (best) | description |
public
Yes
|
0.634 | ||||||||||||||
|
|||||||||||||||||||
| 2 |
Torus Actions
Torus Actions
|
Voting several models | description |
public
Yes
|
0.597 | ||||||||||||||
|
|||||||||||||||||||
| 3 |
DermaCode
|
Best 3 models + hierarchical approach to select outliers + meta-data | description |
public_off
No
|
0.560 | ||||||||||||||
|
|||||||||||||||||||
| 4 |
BGU_hackers
Ben Gurion University
|
AAAR Appoach with meta data | description |
public
Yes
|
0.541 | ||||||||||||||
|
|||||||||||||||||||
| 5 |
BITDeeper
Beijing Institute of Technology
|
MelaNet: A Deep Dense Attention Network for Melanoma Detection in Dermoscopy Images and metadata | description |
public_off
No
|
0.534 | ||||||||||||||
|
|||||||||||||||||||
| 6 |
offer_show
iSee-SYSU
|
Ensemble two model with BCE loss and Cross Entropy by using meta-data | description |
public_off
No
|
0.532 | ||||||||||||||
|
|||||||||||||||||||
| 7 |
Tencent Medical AI Lab
|
meta-data attention for CNN ensemble with threshold for UNK category | description |
public_off
No
|
0.527 | ||||||||||||||
|
|||||||||||||||||||
| 8 |
VisinVis
ETRI
|
Ensembled Transfer Neural Networks by using Lesion Correlation Learning with meta-data | description |
public_off
No
|
0.517 | ||||||||||||||
|
|||||||||||||||||||
| 9 |
MGI
National Institutes of Biotechnology Malaysia
|
2-stage CNN-Tabular neural nets | description |
public_off
No
|
0.500 | ||||||||||||||
|
|||||||||||||||||||
| 10 |
Le-Health
Lenovo Research
|
ensemble strategy 2 | description |
public_off
No
|
0.488 | ||||||||||||||
|
|||||||||||||||||||
| 11 |
MMU-VCLab
Manchester Metropolitan University
|
Ensemble Method with Metadata | description |
public
Yes
|
0.481 | ||||||||||||||
|
|||||||||||||||||||
| 12 |
SY1
Beihang university
|
ResNet101+Feature Fusion+Threshold 5 | description |
public_off
No
|
0.470 | ||||||||||||||
|
|||||||||||||||||||
| 13 |
IML group - DFKI
Interactive Machine Learning (IML) - German Research Center for Artificial Intelligence (DFKI)
|
One-class SVM pre-filter + VGG16 CNN + metadata on flat conv out | description |
public_off
No
|
0.445 | ||||||||||||||
|
|||||||||||||||||||
| 14 |
Panetta's Vision and Sensing Systems Lab
Tufts University
|
Approach 1: XGBoost Ensemble 1 | description |
public
Yes
|
0.431 | ||||||||||||||
|
|||||||||||||||||||
| 15 |
KDIS
University of Cordoba & Maimonides Biomedical Research Institute of Cordoba
|
Multi-view convolutional architecture - Margin sampling & metadata version | description |
public_off
No
|
0.417 | ||||||||||||||
|
|||||||||||||||||||
| 16 |
mvlab-skin
Indian Institute of Technology Roorkee
|
A Network-Surgery Based Deep-Ensembled Framework for Skin Lesion Analysis | description |
public
Yes
|
0.324 | ||||||||||||||
|
|||||||||||||||||||