README.md

---
title: Engagement Detection API
emoji: 🎥
colorFrom: blue
colorTo: green
sdk: docker
pinned: false
---

# Engagement Detection API

A Flask API that uses **MARLIN** (Video Foundation Model) + **LSTM** to predict student engagement from video clips.

## Endpoints

| Method | Route | Description |
|--------|-------|-------------|
| POST | `/upload` | Upload a video chunk (`user_id`, `meeting_id`, `video`) |
| GET | `/process/<user_id>` | Trigger processing for a user |
| GET | `/health` | Health check |

## Example Upload Request

```bash
curl -X POST https://your-space.hf.space/upload \
  -F "user_id=student1" \
  -F "meeting_id=meeting42" \
  -F "video=@clip.mp4"
```

## Models Used
- [ControlNet/MARLIN](https://huggingface.co/ControlNet/MARLIN) — Video feature extractor
- Custom LSTM classifier trained on engagement data