FastAPI Feature Detection & Face Recognition API

This project provides a FastAPI-based web API for real-time face detection using a webcam and various computer vision feature detection techniques on uploaded images. It uses OpenCV and supports multiple algorithms like Harris, Shi-Tomasi, FAST, ORB, and Haar Cascade face detection.

Features

• 📸 Camera-based Face Detection
• 🖼️ Image Upload for Feature Detection
  • Harris Corner Detection
  • Shi-Tomasi Corner Detection
  • FAST Feature Detection
  • ORB Feature Detection
  • Haar Cascade Face Detection

Endpoints

GET /camera_face_detection/

Activates your webcam, captures a frame, detects faces using Haar Cascade, and returns an image with rectangles around detected faces.

Response:

• image/png with detected faces outlined

POST /process_image/

Uploads an image and applies selected feature detection algorithms. Returns the processed image.

Form Data Parameters:

• file: (required) The image file to upload
• use_harris: (optional) true to apply Harris Corner Detection
• use_shi_tomasi: (optional) true to apply Shi-Tomasi Corner Detection
• use_fast: (optional) true to apply FAST feature detection
• use_orb: (optional) true to apply ORB feature detection
• detect_faces: (optional) true to apply face detection

Response:

• image/png with detected features annotated

Installation

1. Clone the Repository:

   git clone https://github.com/KyleJackson6/Facial-Recognition-.git
   cd Facial-Recognition-

2. Install Dependencies:

   Create a virtual environment (optional but recommended):

   python -m venv venv
   source venv/bin/activate  # or venv\Scripts\activate on Windows

   Install packages:

   pip install fastapi uvicorn opencv-python numpy

3. Ensure Haar Cascade File is Available:

   Place haarcascade_frontalface_default.xml inside a models/ directory in the project root.

Running the API

Start the FastAPI server:

uvicorn main:app --reload

Then go to http://127.0.0.1:8000/docs to use the Swagger UI interface.

Project Structure

.
├── main.py
├── models
│   └── haarcascade_frontalface_default.xml
├── face_detection_output.png  # Output from camera face detection

Notes

• Make sure your webcam is connected and accessible for the /camera_face_detection/ endpoint.
• All image processing is done using OpenCV.
• Feature colors:
  • Harris: Red dots
  • Shi-Tomasi: Green circles
  • FAST: Yellow dots
  • ORB: Pink dots
  • Face detection: Red rectangles

License

This project is licensed under the MIT License.

Author

Kyle Jackson