Identifying Melanoma Images using EfficientNet Ensemble: Winning Solution to the SIIM-ISIC Melanoma Classification Challenge

TL;DR

This paper describes a winning ensemble approach using diverse CNN models with metadata integration for melanoma image classification, achieving high AUC scores in the SIIM-ISIC challenge.

Contribution

It introduces a robust ensemble method combining various CNN backbones and metadata, with a focus on validation stability and model diversity.

Findings

Achieved 0.9600 AUC on cross validation

Achieved 0.9490 AUC on private leaderboard

Demonstrated effectiveness of diverse model ensembling

Abstract

We present our winning solution to the SIIM-ISIC Melanoma Classification Challenge. It is an ensemble of convolutions neural network (CNN) models with different backbones and input sizes, most of which are image-only models while a few of them used image-level and patient-level metadata. The keys to our winning are: (1) stable validation scheme (2) good choice of model target (3) carefully tuned pipeline and (4) ensembling with very diverse models. The winning submission scored 0.9600 AUC on cross validation and 0.9490 AUC on private leaderboard.