InfantCare

• 

  Landing Page

• 

  AI Scan Page showing screenshot of an example

• 

  Chat Page showing screenshot of an example

• 

  Learn Page

Inspiration

Every six seconds, a child under five dies from a largely preventable cause. Reading the UN SDG 3.2 targets made that number feel urgent and personal. I wanted to build something that puts trusted health knowledge directly in the hands of caregivers. This is not to replace doctors, but to help people know when to see one.

What it does

InfantCare is an educational web app for caregivers of children under five. It offers three tools: a curated library of WHO and UNICEF health resources, a Gemini-powered AI health assistant that answers child health questions, and an AI image screening tool that analyses photos of skin conditions.. Every feature is framed as educational guidance, not medical diagnosis.

How I built it

The frontend is built with React and Tailwind CSS. The backend is a FastAPI server handling both the AI chat endpoint (via the Gemini API) and the image prediction endpoint. The skin condition classifier uses transfer learning on a pretrained ResNet50, fine-tuned on a labelled dataset of 8 skin conditions.

Accomplishments that I am proud of

Getting a working image upload on the frontend, FastAPI inference endpoint, trained PyTorch model returning top-3 predictions with confidence scores.

What's next for InfantCare

• Expanding the classifier to cover more paediatric-specific conditions with a larger, more diverse dataset
• Adding a WHO IMCI-based symptom checker for non-visual assessments

Built With

• css
• fastapi
• gemini
• html
• javacript
• python
• pytorch
• react
• tailwind
• uvicorn