Improving Face Recognition Features with OpenCV / EmguCV

Question

I am currently using EmguCV (OpenCV C # wrapper) to detect faces in real time (webcam). I get about 7 FPS.

Now I want to improve performance (and save processor cycles), and I'm looking for options, here are my ideas:

• Detect the face, take the face functions and try to find these functions in the next frames (using the SURF algorithm), so that it becomes “face detection + tracking”. If not found, use face recognition again.

• Face detection in the next frame will try to identify the face in the ROI where the face was previously (i.e., look for the face in a smaller part of the image). If the face is not found, try to find it again in the whole image.

• Side idea: if a face is not detected by 2-3 frames and no movement in the image, do not try to detect more faces until movement is detected.

Do you have any suggestions for me?

Thank.

Answer 1

• You can try for the SURF algorithm, but I'm not sure that it provides the appropriate functions on the face, perhaps around the eyes, or if you are close and have skin disorders, or, again, maybe in the hair if there is enough permission. Moreover, SURF is actually not very fast, and I would just avoid making it more calculable if you want to save CPU time.

• - , , camshift, CPU, , camshift - . , , , : , ...

• , (, ), , , ... , , ... , , , n-1 frame n, , , , .... ? 5min , # matchTemplate()...

, ( ) , , , ...

,

Answer 2

. , Haar , . 20 , 10-15 . . : OpenCV haartraining ( - ).

, SURF - . P-N: . YouTube , , .

Answer 3

, .

B.Tech CS , MSB 70% . , , .

, , , . , .