C# Face recognition app with Emgu CV library

Emgu Cv - Face recognition app

This app is recognizing faces in your webcam.

Emgu Cv install

You can install Emgu Cv library in NuGet Packages.

After that you must call libraries. You can do it with this code:

using Emgu.CV;
using Emgu.Util;
using Emgu.CV.Structure;
using Emgu.CV.CvEnum;

Now you can use Emgu cv functions.
Here is Emgu CV documantation

İnput Format - Machine Learning

When the application is started, it captures the image on the camera. And this image made by frames.

Image<Bgr, byte> binaryGoruntu = webcam.QueryFrame().ToImage<Bgr, Byte>()

in this row we are converting frames to Image<Bgr, byte> Array.

After that for the process this converted image , we need to convert image to grayscale.

 Image<Gray, byte> greyScaleGoruntu = binaryGoruntu.Convert<Gray, byte>();

This type of facial recognition, object detection apps generally use XML based HAAR CASCADEfiles.
These files are containing training data for machine learning. You can find these files in the Internet.

The images are searching in the Haar file and the found faces are return. And the binary image we captured at the beginning is overwritten.

Rectangle[] rectangles = haar.DetectMultiScale(greyScaleGoruntu, 1.4, 1, new Size(100, 100), new Size(800, 800));
                    foreach (var face in rectangles)
                    {
                        binaryGoruntu.Draw(face, new Bgr(0, double.MaxValue, 0), 3);
                    }
                    pictureBox1.Image = binaryGoruntu.ToBitmap();

Complexity Analysis

When you examine the Emgu Cv library, detectMultiScale method is the method of machine learning.

In the worst case Complexity is O(n).

The complexity of the test phase is O(n) because foreach loop will return once.

Test

In this app we have two buttons:

• Open Cam
• Close Cam

My Github profile
Projects Github Link

---

Posted on Utopian.io - Rewarding Open Source Contributors

---