README.md

---
title: NeuroFocus Backend
emoji: 🧠
colorFrom: indigo
colorTo: purple
sdk: docker
app_port: 7860
pinned: false
---

# NeuroFocus Backend

FastAPI object detection service powering the NeuroFocus visual attention simulator.
Loads `facebook/detr-resnet-50` and exposes detection results over a REST API.

> NeuroFocus is a neuro-inspired **simulation**. It does not measure or diagnose
> attention, ADHD, fatigue, or any neurological condition.

## Endpoints

| Method | Path             | Description                                   |
| ------ | ---------------- | --------------------------------------------- |
| GET    | `/`              | Service info                                  |
| GET    | `/health`        | Status + device                               |
| POST   | `/detect`        | Image → `{image_size, objects}`               |
| POST   | `/analyze-focus` | Image → `{image_size, objects, scores, note}` |

`/detect` and `/analyze-focus` accept `multipart/form-data` with a `file` field.

## Run locally

```bash
pip install -r requirements.txt
uvicorn app:app --port 7860 --reload
```

Then open http://localhost:7860/docs to try uploads interactively.

## Deploy to Hugging Face

Create a **Docker** Space (CPU Basic is free), then push the contents of this
`backend/` directory. The first boot downloads the DETR model (~160 MB); CPU
inference takes a few seconds per image, which is fine for a demo.