Why not openface about face_recognition HOT 5 CLOSED

Sure, I would be interested to hear how it goes. Closing this issue for now since the basic question is covered.

from face_recognition.

OpenFace uses a mix of several programming languages and frameworks to run. That can make it difficult to deploy and use (in my opinion). The goal here was to make something much simpler to deploy and develop with. But both have their place.

I haven't extensively tested the accuracy of both libraries head to head, but both have performed well in my use cases. So it's more a matter of picking the right tool for the right job.

from face_recognition.

yes, thats absolutely true. Your package is kinda out-of-the-box compared to openface. It would be really great if you can keep a comparision/benchmarking with these two. I am writing a complete python only version of openface, so was just wondering if dlib already has comparable performance, its not worth spending time on openface python version. I'll also try to do some benchmarking as time allows. Thanks for the reply 👍
Cheers

from face_recognition.

In theory, the dlib model used here should be similar to or slightly more accurate than OpenFace, but I have no real numbers to back that up. It also depends on the specific training set used for both models. I.e. you might get slightly better results if you trained your own model from scratch from a larger dataset instead of using the included models from either library.

from face_recognition.

Maybe I should spend more time on dlib before coding up openface-python. Thank you so much for the feedback. I'll keep you posted incase you are interested

from face_recognition.