Skip to content

VSRohith/Skin-Cancer-Detection-using-Attention-Based-Deep-Learning

Repository files navigation

Skin Cancer Detection using Attention-Based Deep Learning

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.


Project Overview

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.


Live Demo

Streamlit Deployment: https://dermai-pro.streamlit.app

Demo Access

The application is currently running in demo mode for testing purposes.

  • Username: doctor1
  • Password: password123

Features

  • 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

Technologies Used

  • Python
  • PyTorch
  • Streamlit
  • OpenCV
  • Matplotlib
  • Pandas
  • Scikit-learn

Project Structure

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.md

How to Run

1. Clone Repository

git clone https://github.com/VSRohith/Skin-Cancer-Detection-using-Attention-Based-Deep-Learning.git
cd Skin_Cancer_Attention_Project

2. Install Dependencies

pip install -r requirements.txt

3. Run Application

streamlit run app.py

Explainable AI Output

The project generates Grad-CAM heatmaps showing the regions focused on by the AI model during prediction.


Current Status

This project is currently under active development and will be further expanded as a final year project with additional features and improved model performance.


Future Enhancements

  • Improved model accuracy
  • Real-time image prediction
  • Cloud deployment
  • Patient history management
  • Enhanced medical report generation
  • Multi-model comparison

Author

Developed by Rohith

About

Attention-based deep learning system for skin cancer detection with Grad-CAM explainability and Streamlit deployment.The application is currently running in demo mode for testing purposes. Username: doctor1 ,Password: password123

Topics

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages