Deep Learning Based Face Recognition

Using deep learning based face recognition, you can perceive and understand faces within your application. This document is a guide on how to register a new face, unregister a face that is registered with a given label, and recognize a given face.

Prerequisites

To enable your application for using the deep learning based face recognition functionality, create DeepLearningFaceRecognizer instance:

                
C#
                

                    
Copy
                    

                
            
var deepLearningFaceRecognizer = new DeepLearningFaceRecognizer();

Register a new face

To register a new face image with a given label, follow these steps:

1. Prepare an input source to register. In this example, we use ImageUtil APIs to get the image buffer:

   
                   
   C#
                   
   
                       
   Copy
                       
   
                   
               
    MediaVisionSource inputSource = null;
   
    using (JpegDecoder jpegDecoder = new JpegDecoder())
    {
        jpegDecoder.SetColorSpace(ColorSpace.Rgb888);
   
        var frame = (await jpegDecoder.DecodeAsync("path")).FirstOrDefault();
   
        inputSource = new MediaVisionSource(frame.Buffer, (uint)frame.Size.Width, (uint)frame.Size.Height, ColorSpace.Rgb888);
    }
   

   Note

   The input source for DeepLearningFaceRecognizer should be RGB data.

2. Now we can register a new face. This is a process where the face recognition framework trains and generates an internal model file with the given face data and its label. You could repeat this step to train the internal model file with more faces. For best accuracy, we recommend training at least three images per face:

   
                   
   C#
                   
   
                       
   Copy
                       
   
                   
               
   deepLearningFaceRecognizer.RegisterFace(inputSource, "FaceName1");
   inputSource.Dispose();
   

   You can unregister the face from DeepLearningFaceRecognizer internal model:

   
                   
   C#
                   
   
                       
   Copy
                       
   
                   
               
   deepLearningFaceRecognizer.UnregisterFace("FaceName1");
   

Recognize face

To recognize a given face image, follow these steps:

1. Prepare a target image source. In this example, we use ImageUtil APIs to get the image buffer:

   
                   
   C#
                   
   
                       
   Copy
                       
   
                   
               
   MediaVisionSource targetSource = null;
   
   using (JpegDecoder jpegDecoder = new JpegDecoder())
   {
       jpegDecoder.SetColorSpace(ColorSpace.Rgb888);
   
       var frame = (await jpegDecoder.DecodeAsync("path")).FirstOrDefault();
   
       targetSource = new MediaVisionSource(frame.Buffer, (uint)frame.Size.Width, (uint)frame.Size.Height, ColorSpace.Rgb888);
   }
   

2. Recognize a given face image:

   
                   
   C#
                   
   
                       
   Copy
                       
   
                   
               
   var result = deepLearningFaceRecognizer.Recognize(targetSource);
   
   // The result have Label property for recognized face.
   Log.Info("Vision", result.Label);
   
   targetSource.Dispose();
   

Related information

• Dependencies
  • Tizen 7.0 and Higher