Face Recognition on Jetson TX1 with NVIDIA DIGITS


According to

It is said that Facenet-120 can be used as single class facial-recognition detector.

I run the example, and it is possible to detect a face on an image, as long as the face is well aligned.
If the faces are not aligned in the image, it cannot detect them.

Is it possible to retrain this model to detect a specific person’s face on a captured frame?

And if so, what is the process?

If not, what is the use of Facenet-122 and what is it supposed to be trained for?



Facenet-122 in Dustin’s page is DetectNet trained for face detection.

You can re-train your own DetectNet with preferred database.
We haven’t trained DetectNet to recognize specific person before.
But you can give it a try.

The easiest way to re-train DetectNet is to use DIGITs.
DIGITs: https://developer.nvidia.com/digits
Object Detection: https://github.com/NVIDIA/DIGITS/tree/master/examples/object-detection

Happy Holiday : )

Hi Evi, that particular model was trained using FDDB.

You can also find an index of face databases available here: http://face-rec.org/databases/

Hi guys,

Thank you a lot for your answers.

I was confused initially that I could you it for face recognition.

Happy holidays to everyone :)