---
title: JAI
emoji: 🚀
colorFrom: indigo
colorTo: gray
sdk: docker
pinned: false
---

Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

# 🎥 Multi-Face Attention Detector (YOLOv7, Streamlit, GPU)

This app detects attention in multiple faces using YOLOv7 + mediapipe + a custom ML model.

## 🚀 Setup on Hugging Face

1. Create a Space → choose Blank Docker template.
2. Upload:
   - app.py
   - requirements.txt
   - Dockerfile
   - model.pkl (your ML model)
   - example_input.jpg (optional test image)

3. Enable GPU hardware.

4. Deploy!

## ✅ Features

- YOLOv7 GPU face + phone detection
- Mediapipe fallback (via mediapipe-nightly)
- OpenCV video processing
- Streamlit dashboard with charts, logs, CSV download

## 💡 Notes

- Base image: `nvidia/cuda:11.8.0-cudnn8-runtime-ubuntu22.04`
- Compatible with `torch==2.1.2+cu118`
- Includes placeholder model.pkl for demo