Hello,
I have recently started training my own face detection haar cascades with OpenCV. I have a large collection of positive images (~3000) and negative images (~3000) but I am getting very poor results. I am worried that I am not setting the parameters properly and it is having a strong negative effect on my accuracy. To verify this, I have been looking for the parameters OpenCV used to train the alt2 haar cascade model (the most accurate face detection model). I have looked through the source and I don't think OpenCV releases these parameters by default. Does anyone know what they are?
Thanks so much, Jordan Cheney
