[OpenCV 3.0.1] CascadeClassifier not working with HOG (Bug #4342)

Added by Alex D almost 10 years ago. Updated almost 10 years ago.

Status:Cancelled Start date:2015-05-15
Priority:Normal Due date:
Assignee:Vadim Pisarevsky % Done:

0%
Category:objdetect
Target version:3.1
Affected version:branch 'master' (3.0-dev) Operating System:Linux
Difficulty:Hard HW Platform:x64
Pull request:

Description

Program crashes when loading cascade.xml trained with opencv_traincascade with HOG feature.

I attached a file cascadeHOG.xml so you can test and make sure the problem exists. Any help strongly appreciated.

Training params:
$ opencv_traincascade -data haarCascade -vec samples.vec -bg Negative/negative.dat -numStages 24 -minhitrate 0.999 -maxFalseAlarmRate 0.5 -numPos 13000 -numNeg 28336 -w 50 -h 50 -mode ALL -precalcValBufSize 1500 -precalcIdxBufSize 1500 -featureType HOG

cascadeHOG.xml (159.7 kB) Alex D, 2015-05-15 03:48 pm

History

#1
Updated by Vadim Pisarevsky almost 10 years ago

  • Priority changed from High to Normal
  • Category set to objdetect

#2
Updated by Vadim Pisarevsky almost 10 years ago

We decided to drop the current HOG cascades in OpenCV 3.x. The implemented HOG features are quite weird - different from Dalal's interpretation of HOG, different from P. Dollar integral channel features. In xobjdetect we slowly grow superior ICF/ACF+WaldBoost-based detector, which is there already and will be improved during 2015.

  • Status changed from New to Cancelled
  • Assignee set to Vadim Pisarevsky
  • Difficulty set to Hard

Also available in: Atom PDF