How aayushakumar/skinguardian works
Project Type and Purpose
• SkinGuardian is a desktop application with a local web interface (Flask). It packages a fine‑tuned BEIT model (ONNX) for on‑device skin cancer detection.
• There is no external database or cloud dependency—everything runs locally via a Flask server and ONNX Runtime.
File Structure
• Top‑level:
• MobileApp/ – (future mobile version scaffold)
• SkinGuardian/ – core Python app + templates + model + packaging files
• extras/ – scripts and notebooks for downloading and fine‑tuning the model
• SkinGuardian/ contains:
• app.py – Flask application (controller)
• beit_finetuned_model.onnx – ML model (inference engine)
• /templates – HTML views (index, result, about, contact)
• PyInstaller artifacts (app.spec, dist/app.exe)
• requirements.txt, environment.yml, runtime.txt – environment/config
• extras/ holds development tools (download_model.py, finetune_model.ipynb) and test images.
README for Additional Insights
• Confirms a Flask‐based UI on port 5000.
• Explains on‑device inference via ONNX Runtime, packaged as a single EXE using PyInstaller.
• No database; input is user‐uploaded images, output is classification result.