An AI-powered skin cancer detection system developed using Deep Learning and Explainable AI techniques. This project began as a mini project and is being continuously enhanced toward a final year major project.
This system detects different categories of skin cancer lesions using an attention-based deep learning model. The application also includes Grad-CAM heatmap visualization to improve model interpretability and explain the AI's predictions visually.
Streamlit Deployment: https://dermai-pro.streamlit.app
The application is currently running in demo mode for testing purposes.
- Username:
doctor1 - Password:
password123
- Skin lesion image classification
- Attention-based deep learning architecture
- Grad-CAM heatmap visualization
- Interactive Streamlit web application
- Explainable AI (XAI) support
- GPU acceleration with CUDA support
- Python
- PyTorch
- Streamlit
- OpenCV
- Matplotlib
- Pandas
- Scikit-learn
Skin_Cancer_Attention_Project/
│
├── app.py
├── train_final.py
├── test_visual.py
├── model_attention.py
├── attention_blocks.py
├── explainability.py
├── data_loader.py
├── skin_attention_model.pth
├── requirements.txt
└── README.mdgit clone https://github.com/VSRohith/Skin-Cancer-Detection-using-Attention-Based-Deep-Learning.git
cd Skin_Cancer_Attention_Projectpip install -r requirements.txtstreamlit run app.pyThe project generates Grad-CAM heatmaps showing the regions focused on by the AI model during prediction.
This project is currently under active development and will be further expanded as a final year project with additional features and improved model performance.
- Improved model accuracy
- Real-time image prediction
- Cloud deployment
- Patient history management
- Enhanced medical report generation
- Multi-model comparison
Developed by Rohith