An Efficient CNN Architecture for Classifying Skin Cancer on an Imbalanced Dataset

✨ Key Features

🖼️ Dermatology Image Preprocessing – Normalization, augmentation, and resizing

🤖 Deep Learning Models – CNNs, Transfer Learning (ResNet, VGG, EfficientNet, Inception)

📊 Performance Evaluation – Accuracy, Precision, Recall, F1-score, ROC curve, and Confusion Matrix

🔍 Multi-Class Classification – Different types of skin lesions (benign vs malignant)

📈 Visualization Tools – Training curves, misclassified samples, Grad-CAM heatmaps

⚡ Scalable Deployment (Optional) – Streamlit/Flask app for real-time image classification

🧰 Tech Stack

Programming: Python 🐍

Libraries & Frameworks: TensorFlow / Keras, PyTorch, OpenCV, scikit-learn, NumPy, Pandas, Matplotlib, Seaborn

Environment: Jupyter Notebook / Google Colab

Deployment (Optional): Flask / FastAPI / Streamlit / Django

📁 Project Structure 📁 data/ # Skin lesion datasets (train/test/validation) 📁 notebooks/ # Jupyter notebooks for model experiments 📁 models/ # Trained CNN & transfer learning models 📁 src/ # Scripts for preprocessing, training, evaluation 📁 results/ # Metrics, graphs, confusion matrices 📁 app/ # Web app or API for deployment

🚀 Getting Started git clone https://github.com/yourusername/Skin-Cancer-Classification.git cd Skin-Cancer-Classification pip install -r requirements.txt jupyter notebook

📌 Use Cases

🏥 Medical Diagnosis Assistance – Helps dermatologists with skin lesion classification

📊 Research & Healthcare AI – Academic projects in computer vision & medical AI

🎓 Educational Resource – Learn deep learning applications in healthcare

🌍 Telemedicine Solutions – Deploy as a web/mobile app for remote skin screening

🤝 Contributing

Contributions are welcome! Add datasets, improve CNN architectures, or extend deployment by submitting a PR.

📜 License

MIT License – Free for research, education, and healthcare innovation.

⭐ Support

If this project inspires you, please star ⭐ the repo and share it with the AI + Healthcare community!

📚 This repository contains the original implementation of SkinNet-8: An Efficient CNN Architecture for Classifying Skin Cancer on an Imbalanced Dataset.

📧 Contact: For inquiries, you can reach out to us at rhslion34@gmail.com.

Dataset Availability

This project uses the following dataset for Skin Cancer Images:

Code

You can find the implementation of SkinNet-8 in the following file:

• SkinNet-8 Skin Cancer Classification.ipynb

Paper Framework

🖼️ PIPELINE:

Copying

This code is shared for research use only. If you encounter any issues or find inappropriate content in this code, please feel free to contact us.