Melanoma Skin Cancer Classification Dataset

Table of Contents

• Overview
• Dataset Structure
• Reference
• License

Overview

The Melanoma Skin Cancer Classification dataset is a curated collection of high-resolution dermatoscopic images for binary classification of skin lesions as Benign or Malignant (Melanoma). It is designed for research and education in medical image analysis, data augmentation, and deep learning for dermatology.

The goal is to classify each skin lesion image into one of two categories: Benign or Malignant (Melanoma).

Property	Value
Total samples	1,400
Training samples	978
Test samples	422
Modality	Dermatoscopic images
Classes	0 Benign, 1 Melanoma

Dataset Structure

Clone the repository:

git clone [email protected]:AIOZAI/Melanoma_Skin_Cancer_Data.git

Or download directly: Melanoma Skin Cancer Classification Dataset

Files

File	Samples	Description
Train/Melanoma_skin_cancer_phase1_release1.zip	978 images	Training data with labels in ground_truth.csv
Test/Melanoma_skin_cancer_phase1_release1.zip	422 images	Test data for submission (ground truth hidden)

Ground Truth File Format

ground_truth.csv contains the ground truth labels for the training images.

Field	Type	Description
id	string	Image filename
label	integer	0 = Benign, 1 = Melanoma

Example (ground_truth.csv):

id,label
9c9a204a57f34a71b9454fef633f7038.jpg,0
cec04296a9bf4b969a8770f3ee2b439f.jpg,1

Reference

For more information about the dataset, see the Kaggle page.

License

This dataset is released under CC0 1.0 Universal (Public Domain Dedication).