About FaceRecognizer
Dear all, Right now I am developing a face recognition system by using OpenCV, and what I want to
implement is to find the similarity of the given face and the faces in database, and can list
the result like this:
face1: 95%
face2: 93%
face3: 92%
face4: 80%
I found a class named FaceRecognizer in opencv, and it has some interface like train and
predict, but finally I found that this two interface only can judge whether a face is belong to
the database or not, can not do what I want to do.
Is there any other interface can achieve my goal?
If you're interested in solutions outside of OpenCV, I can check this post which summarizes 40+ API's for face detection and recognition:
http://blog.mashape.com/post/53379410412/list-of-40-face-detection-recognition-apis
"Is there any other interface can achieve my goal?"
unfortunately, not so. it's not only the interface, that does not meet your requirements, but also all 3 implementations do a single nearest neighbour search, what you want sounds more like knn. you'll also have to devise your own heuristics for expressing something like : 95%.
no real idea, what's the best solution here, either hack the interface/library, or just 'steal' the relevant parts and devise your own ( you probably don't need all 3(eigen,fisher,lbp), lbp seems to be the most easy of them )
Thanks for you reply berak, can you explain more about how can I steal the relevant parts and devise my own? I want more details about this....
look at modules/contrib/src/facerec.cpp
pls let me know where can I get the source code of opencv in android? or I can build my own android sdk from the master branch?