Face and Emotion Recognition API

A Flask-based REST API for detecting faces in images and predicting their emotions using machine learning models.

Features

• Face detection using MTCNN (Multi-task Cascaded Convolutional Networks)
• Emotion recognition with EmotiEffLib models
• RESTful API with JSON responses
• Base64-encoded face images in responses
• CORS support for web applications
• Automatic cleanup of uploaded files

Installation

Prerequisites

• Python 3.9 or higher
• UV package manager (recommended)

Setup

1. Clone or navigate to the project directory:

   cd server

2. Install dependencies using pip:

   pip install -r requirements.txt

Usage

Running the Server

Start the Flask development server:

python app.py

The server will run on http://localhost:5000.

API Endpoint

POST /analyze

Analyzes an uploaded image for faces and predicts emotions.

Request:

• Method: POST
• Content-Type: multipart/form-data
• Body: image (file) - Image file (JPEG, PNG)

Response (200 OK):

{
  "results": [
    {
      "emotion": "happy",
      "face_image": "base64encodedstring..."
    }
  ]
}

Error Response (400 Bad Request):

{
  "error": "No image file provided"
}

Project Structure

server/
├── app.py              # Flask application
├── model.py            # Face detection and emotion recognition logic
├── pyproject.toml      # Project configuration and dependencies
├── README.md           # This file
└── uploads/            # Temporary upload directory (auto-created)

Dependencies

• Flask: Web framework
• EmotiEffLib: Emotion recognition library
• Facenet-Pytorch: Face detection
• OpenCV: Image processing
• Pillow: Image handling
• Flask-CORS: Cross-origin resource sharing