VisiCore AI is a professional-grade screening tool designed to detect Diabetic Retinopathy (DR) from retinal fundus images and provide an AI-powered clinical consultation interface.
- Automated DR Screening: Uses a Vision Transformer (ViT) to classify retinal images into stages (0-4).
- AI Health Consult: A local LLM (Qwen 2.5) that provides context-aware medical explanations with conversation memory.
- Dark/Light Mode: Premium UI with a dynamic design system.
- Privacy First: All processing happens locally on your machine.
- Optimized for Speed: Rapid inference times for clinical workflows.
- Frontend: React, TypeScript, Vite, Tailwind CSS, Lucide Icons, Framer Motion.
- Backend: Flask, Python, Llama-cpp-python (for local LLM).
- Models:
- Vision Model:
dr_model_complete.pth - Language Model:
qwen2.5-1.5b-instruct-q4_k_m.gguf
- Vision Model:
-
Clone the repository:
git clone https://github.com/codedbyasim/VisiCore-AI.git cd VisiCore-AI -
Install Backend Dependencies:
pip install -r requirements.txt
-
Install Frontend Dependencies:
cd Frontend npm install -
Prepare Models: Create a
models/directory in the root and download the following models:- Vision Model (PTH): dr_model_complete.pth
- Language Model (GGUF): Qwen2.5-1.5B-Instruct-Q4_K_M
The backend is configured to serve the pre-built frontend from the static folder.
- Start the Flask server:
python app.py
- Open your browser and navigate to
http://127.0.0.1:5000
- Start the Backend:
python app.py(Running on port 5000) - Start the Frontend Dev Server:
cd Frontend npm run dev - Open
http://localhost:5173
├── Frontend/ # React Source Code
├── static/ # Built Frontend Assets
├── models/ # AI Models (pth & gguf)
├── app.py # Flask Backend
├── requirements.txt # Python Dependencies
└── README.md # Project Documentation