07_Face Detection with OpenCV YuNet

This project demonstrates real-time face detection using YuNet, a high-performance deep learning model from the OpenCV Model Zoo.

Overview

YuNet is a convolutional neural network (CNN) specifically designed for face detection. It is extremely fast and capable of detecting faces even with small sizes or side views.

Key technical highlights:

Model format: ONNX (Open Neural Network Exchange).
Framework: OpenCV’s dnn module.
Features: Detects face bounding boxes and 5 facial landmarks (eyes, nose, mouth corners).

Implementation Details

The implementation uses the cv2.FaceDetectorYN class provided by OpenCV.

1. Model Initialization

The detector is initialized with specific parameters to balance speed and accuracy.

import cv2

# Configuration
model_path = "face_detection_yunet_2023mar.onnx"
input_size = (width, height) # Size of the input image
score_threshold = 0.9
nms_threshold = 0.3
top_k = 5000

# Create the detector
detector = cv2.FaceDetectorYN.create(
    model_path,
    "",
    input_size,
    score_threshold,
    nms_threshold,
    top_k
)

2. Face Detection

Running the detection process is a simple method call.

# Detect faces
_, faces = detector.detect(image)

if faces is not None:
    for face in faces:
        # face contains [x, y, w, h, x1, y1, x2, y2, x3, y3, x4, y4, x5, y5, score]
        bbox = face[0:4].astype(int)
        landmarks = face[4:14].reshape((5, 2)).astype(int)
        score = face[14]

Example Output

Below is an example of the model detecting a face in a high-resolution image.

Face Detection Example

Conclusion

YuNet provides a modern, robust alternative to traditional Haar Cascades or HOG-based detectors, making it ideal for edge devices and real-time applications.

TODO

Implement video stream detection
Add face recognition layer

Home

Last Updated: 2026-01-27