Upload a wound photo, get an instant diabetic foot-ulcer severity read with a confidence score.
- TensorFlow
- MobileNetV2
- FastAPI
- React
- Vite
- Tailwind
THE PROBLEM
Diabetic foot ulcers are a leading path to lower-limb amputation, and clinicians misjudge wound severity a large share of the time. People in rural or under-resourced areas often cannot get a fast read on whether a wound needs urgent care.
THE BUILD
I built a web app where you upload a photo of a wound and get back a severity classification with a confidence score and a recommended clinical action. The model is a MobileNetV2 network I trained and fine-tuned on roughly 20,000 annotated foot-ulcer images. A FastAPI backend serves the model, and a React and Vite frontend handles the upload flow and the report.
THE RESULT
StepSafe won 1st place at the HMI Hackathon 2026 at Kennesaw State. The frontend and the API are both deployed and live, so anyone can run a photo through it.