Is there a way to merge the face recogniser with the person detected in the next unit? I mean, can we obtain the position in 3D of the person with the face recogniser?
For example, if you are detecting people with the leg detector, the upper body and the mld trackerr, if you want to follow one of them with the detection from the kalman filter (because is more robust), is there a way to use the face recogniser to know who you want to follow?.
The fast response is no. The face recogniser is menat only for knowing who it is, not where it is. Use the face detector for that.
You could infer the detected face position based on the bounding box size, because human faces are always the sime size more or less, so you can calibrate it and do that. Although having the face detector that generated already a TF and everything its much easier.
I see, i was looking for a face detector but that can recognise the faces. Maybe i will try to do something with the bounding box size, it is interesting and maybe i can get something out of that idea.
Thank you very much