arcticworker's profile - activity

In the Opencv there are some haar cascades:

• haarcascade_frontalface_alt.xml - Stump-based 20x20 gentle adaboost frontal face detector. Created by Rainer Lienhart.
• haarcascade_frontalface_alt2.xml - Tree-based 20x20 gentle adaboost frontal face detector. Created by Rainer Lienhart.

What is the difference between Stump-based and Tree-based frontal face detector? Inipractice, they give very similar results.

@tuannhtn, thanks for reply. What about optimal image resolution for recognition? I should use the highest resolution or are there any restrictions?

What format is best use for facial recognition e.g. png, jpg, pgm ot another? i use opnecv methods LBPHFaceRecognizer, FisherFaceRecognizer and EigenFaceRecognizer.

i try to use PredictCollector:

recognizer = cv2.face.createLBPHFaceRecognizer()
collect = cv2.face.PredictCollector()
recognizer.predict(predict_image, collect)

but on first step program is crashing:

Traceback (most recent call last):
  File "faces.py", line 62, in <module>
    collect = cv2.face.PredictCollector()
AttributeError: 'module' object has no attribute 'PredictCollector'

When i use:

collect = cv2.face.VectorPredictCollector()

it work well, but i need PredictCollector. what am I doing wrong? i use opencv v. 3.1.0

i use 3.1 version for linux

I use method cv::face::FaceRecognizer::predict from here, in python i write this:

nbr_predicted, conf = recognizer.predict(predict_image)

it predicts a label and associated confidence (e.g. distance) for a given input image, it return only 1 label (the closest for a given input image). How i can get all labels and associated confidence (e.g. distance) for my input image?

On same page i find this method:

virtual void cv::face::FaceRecognizer::predict  (   InputArray  src,
                                                    Ptr< PredictCollector >     collector,
                                                    const int   state = 0
                                                    )       const

If I'm not mistaken this is what I need. But i can't figure out what this is "Ptr< PredictCollector > collector", and how to describe (use) this in python?

p.s. sorry for my bad English.

alexylem, I have the same problem. If you know how to fix it, please help.