1 | initial version |
the hog cascades were removed in 3.x (and won't come back)
the hog detector is using a linear SVM (not a boosted cascade) to predict candidate hog features from sliding windows, so the serialized xml file contains the single support vector (amd -rho for thresholding the dot product later).
2 | No.2 Revision |
the hog cascades were removed in 3.x (and won't come back)
the hog detector is using a linear SVM (not a boosted cascade) to predict candidate hog features from sliding windows, so the serialized xml file contains the single support vector (amd -rho for thresholding the dot product later).
you can use the train_HOG sample to train your own classifier there.
3 | No.3 Revision |
the hog cascades were removed in 3.x (and won't come back)
the hog detector is using a linear SVM (not a boosted cascade) to predict candidate hog features from sliding windows, so the serialized xml file contains the single support vector (amd -rho for thresholding the dot product later).
you can use the train_HOG sample to train your own classifier there.
4 | No.4 Revision |
the hog cascades were removed in 3.x (and won't come back)
the hog detector is using a linear SVM (not a boosted cascade) to predict candidate hog features from sliding windows, so the serialized xml file contains the single support vector (amd (and -rho for thresholding the dot product later).
you can use the train_HOG sample to train your own classifier there.