OpenCV 3.0 and Python segfault with detectMultiScale (Bug #4267)

Added by Mike Feist almost 10 years ago. Updated almost 10 years ago.

Status:Done Start date:2015-04-07
Priority:Normal Due date:
Assignee:Vadim Pisarevsky % Done:

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

Description

So we first trained our data using opencv_traincascade, this worked fine. But then when we loaded the cascade into our python application and tried to detect objects it gave us a segfault. This issue went away when we moved to Python 2.7 and OpenCV 2.4.10. We also tried both Python 2.7 and Python 3.4 with OpenCV 3.0 with the same results.

netCascade = cv2.CascadeClassifier('cascade.xml')

...

nets = netCascade.detectMultiScale(gray, 1.3, 5)

History

#1
Updated by Philip L almost 10 years ago

Please test the current opencv code if you have time, it should now work in 3.0 too.

#2
Updated by Vadim Pisarevsky almost 10 years ago

checked the current facedetect.py - it works fine. However, the call to the nested cascade should be disabled when the nested cascade is not used. Will submit PR that fixes the problem.

  • Status changed from New to Done
  • Assignee set to Vadim Pisarevsky
  • Target version set to 3.0

Also available in: Atom PDF