README.md

metadata

title: Detect4Enhance Backend
emoji: 🤖
colorFrom: blue
colorTo: green
sdk: docker
sdk_version: latest
app_file: app.py
pinned: false

Engagement Detection API

A Flask-based API for detecting engagement levels from facial expressions using TensorFlow Lite.

API Endpoints

• GET /: Returns API information
• POST /predict: Accepts an image file and returns engagement predictions

Deployment

1. Clone this repository
2. Build Docker image: docker build -t engagement-api .
3. Run container: docker run -p 7860:7860 engagement-api

Example Usage

import requests

url = "https://your-huggingface-space-url.hf.space/predict"
files = {'image': open('test.jpg', 'rb')}
response = requests.post(url, files=files)
print(response.json())

Model Details

• Model: Custom TensorFlow Lite model (engagement_model_89.tflite)
• Input: 224x224 RGB face image
• Output: Probabilities for 4 engagement states:
  • Engaged
  • Frustrated
  • Bored
  • Confused