Prebuild trainer for video face recognition processing in opencv c++ -
currently trying on gender recognition/ classifcation using opencv , fisherface algorithm. @ moment, training program apporximately 2500+ images due different lightings.
problem takes me 35 minutes execute file. everytime execute file have re-train program. processing 2500+ images , takes 35 minutes before video face recognition program comes on.
my xecution line in terminal
./main haarcascade_frontalface_alt.xml image_to_train.txt 0
./main object build cpp/ .exe in windows
haarcascade_frontalface_alt.xml face detection provided opencv
image_to_train.txt 2500 images are
0 webcam
is there anyway pre-built .txt file on iamges dont have retrain them everytime. thinking database, database runs single image processing through whole database again(brute-force concept).
codes of training program these 2 lines in entire video recognition process. images passed these functions 1 one 2500 images execution time.looking opencv documentation still didnt idea on how can extract data out of them , process them make execution lot faster. prebuild excuting trainer.
ptr<facerecognizer> model = createfisherfacerecognizer(); model->train(images, labels);
Comments
Post a Comment