A machine learning pipeline for classifying skin lesions as cancerous or non-cancerous using dermoscopic images.
This project implements a skin cancer detection system using traditional machine learning techniques. It includes:
- Preprocessing: Removes hair and glare from images.
- Feature Extraction: Extracts 273 features (color, texture, shape).
- Modeling: Trains an ensemble of SVM and Random Forest models.
- Evaluation: Provides metrics like accuracy, ROC-AUC, and visualizations.
The project uses the Skin Cancer: Malignant vs. Benign dataset. Results are detailed in docs/paper.pdf
.
- Clone the repository:
git clone https://github.com/ahmed-226/Skin-Cancer-Detection.git cd Skin-Cancer-Detection
- Install dependencies:
pip install -r requirements.txt
- Use:
or
jupyter notebook notebooks/main.ipynb
python src/main.py