search

btwardow / tf-face-recognition

Ral-Time Tensor Flow Face Detection and Recognition
MIT License
67 stars 20 forks source link

About modifying frames #6

Open LiuShaw opened 5 years ago

LiuShaw commented 5 years ago

How to modify the number of detected frames and where to modify?

btwardow commented 5 years ago

What do you mean by the number of detected frames?

Face detection algorithm don't have a limit.

Or maybe you have in mind number of examples in the face recognition model?

LiuShaw commented 5 years ago

Thank you for your algorithm and reply 👍 . I found that the detect box couldn’t catch people when people move fast. I thought you set variables to limit it at first. After looking through the code, I found it just depends on the performance of CPU. In addition, as your comment, the problems in the “make” method are caused by "pre-trained models not imported properly". More specific, the loading of tf images has some problems. May it can be solved by reload tf in embedding and recognition .py. The problems do shows the magic of docker. :-)

sumit1212 commented 5 years ago

Hey btwardow, Capture faces is too slow how to increases.

Any way Thank for great system.