codingTornado's profile - activity

Your error message appears to indicate you are using OpenCV 2.4.7 which is not the most recent, have you tried the latest 2.4 branch version? Maybe your bug is already fixed there

You need to add the libraries to the VS project like they explain in this tutorial
tl;dr you just need to add the include directory to the additional include directories, your library folders to the additional library directories and all .lib you need to the additional dependencies. Also, as stated here, you will need to add opencv to your path.
This article might come in handy too.

Reading the documentation helps a lot ;)

Do you mean the value for the 3 channels at each even/even position or do you mean something weird like "even byte"?

After a thorough search in the issue tracker I found a bug report on this very same issue, thanks @berak

Hello, I'm trying to find faces in a video using the CascadeClassifier. Since I migrated from 2.4 to 3.0 I've been having an issue with detectMultiScale when I hand it a RoI bounded Mat.
I use the following line to search for the faces

classifier_->detectMultiScale(region, found, 1.35, 5, 0, Size(25, 25));

Where region is a Mat bounded by a Rect I defined to reduce the search area based on a background subtractor and found is a new vector.

When run I get the following:

Debug Assertion Failed!
Program: C:\Windows\system32\MSVCP110D.dll
File: C:\Program Files (x86)\Microsoft Visual Studio 11.0\VC\include\vector
Line: 1124

Expression: vector subscript out of range

When debugging the application I find that this error comes when executing line 1265 on cascadedetect.cpp

if( !featureEvaluator->setImage(gray, scales) )

because scales, a vector initialized on line 1249, reaches setImage with size 0 no matter what I do.
What should I do about it? Should I add to that line another condition stating that scales must be greater than 0? Is this a bug that should be reported or am I doing something wrong?

I'm posting this answer so this looks a bit more tidy for searches since there is a big conversation in the comments but no "answer" to the thread.

What has been said is that some extra modules are broken, for example saliency(which up to this date 06/Jan/2015 still won't compile on windows), and won't compile properly. Fixing needs to be done on this modules, so unless you really need them, just exclude them from the cmake configuring process.

EDIT:
As pointed out in the comments, I forgot about the flags to exclude the non-compiling modules.

Add -DBUILD_opencv_saliency=OFF to the CMake command to remove saliency.
I also found trouble with missing headers and a non-portable suffix for a number (used LLU instead of ULL, which Visual Studio doesn't recognize as valid), so if you come into trouble building line_descriptor add -DBUILD_opencv_line_descriptor=OFF to the CMake command.

In case you would like to remove any other extra modules that don't depend on 3rd parties like matlab here is the full list of flag names:

• BUILD_opencv_adas for adas
• BUILD_opencv_bgsegm for bgsegm
• BUILD_opencv_bioinspired for bioinspired
• BUILD_opencv_ccalib for ccalib
• BUILD_opencv_datasets for datasets
• BUILD_opencv_face for face
• BUILD_opencv_latentsvm for latentsvm
• BUILD_opencv_line_descriptor for line_descriptor
• BUILD_opencv_optflow for optflow
• BUILD_opencv_reg for reg
• BUILD_opencv_rgbd for rgbd
• BUILD_opencv_saliency for saliency
• BUILD_opencv_surface_matching for surface_matching
• BUILD_opencv_text for text
• BUILD_opencv_tracking for tracking
• BUILD_opencv_xfeatures2d for xfeatures2d
• BUILD_opencv_ximgproc for ximgproc
• BUILD_opencv_xobjdetect for xobjdetect
• BUILD_opencv_xphoto for xphoto