features 2D : DenseFeatureDetector fails when using OpponentSIFT descriptor but works with SIFT, SURF and OpponentSURF (Bug #1824)

Added by Alexandre Benoit almost 13 years ago. Updated almost 13 years ago.

Status:Cancelled Start date:2012-04-19
Priority:Blocker Due date:
Assignee:Alexander Shishkov % Done:

0%
Category:features2d
Target version:2.4.0
Affected version: Operating System:
Difficulty: HW Platform:
Pull request:

Description

Hi all,
When using a Descriptor extractor based on the OpponentSIFT descriptor and using the DenseFeatureDetector keypoint detector, a segmentation fault occurs.

This error is not encountered when the descriptor is SIFT, SURF or OpponentSURF

The problem seems to come from the fact that the number of keypoints returned by the SIFT descriptor on each opponent channel is not the same for each channels... which is strange if a dense grid is chosen

Explanations :
if the Feature detector object used is a DenseFeatureDetector and if the feature descriptor is the OpponentSIFTDescriptor :
in modules/features2d/src/descriptors.cpp : void OpponentColorDescriptorExtractor::computeImpl
the single component extractor (SIFT/SURF, etc.) is applied on a dense grid on each Opponent channel in the for loop.
However, in the case of the OpponentSIFT the number of "channelKeypoints" is not the same for each channel
Then, the remaining of the function fails at the while loop level : 

1while( channelKeypoints[0][idxs[0][cp[0]]].class_id < maxInitIdx && cp[0] < channelKeypoints[0].size() ) { cp[0]++; }
2while( channelKeypoints[1][idxs[1][cp[1]]].class_id < maxInitIdx && cp[1] < channelKeypoints[1].size() ) { cp[1]++; }
3while( channelKeypoints[2][idxs[2][cp[2]]].class_id < maxInitIdx && cp[2] < channelKeypoints[2].size() ) { cp[2]++; }

here is the segmentation fault location.

hope it helps.

Best regards.
Alex

Associated revisions

Revision ab15b595
Added by Alexander Shishkov almost 13 years ago

fixed #1824

Revision 8d1a8df1
Added by Andrey Pavlenko over 11 years ago

Merge pull request #1824 from vpisarev:ocl_experiments5

History

#1
Updated by Andrey Kamaev almost 13 years ago

  • Description changed from Hi all, When using a Descriptor extractor based on the OpponentSIFT descripto... to Hi all, When using a Descriptor extractor based on the OpponentSIFT descripto... More

#2
Updated by Vadim Pisarevsky almost 13 years ago

  • Assignee changed from Vadim Pisarevsky to Maria Dimashova

#3
Updated by Alexander Shishkov almost 13 years ago

I can't reproduce your problem on 2.4 branch.
My code sample:

 1    Mat im = imread(argv[1],1);
 2
 3    DenseFeatureDetector df;
 4    vector<KeyPoint> kpts;
 5    df.detect(im, kpts);
 6    Ptr<DescriptorExtractor> sift = new SiftDescriptorExtractor();
 7    OpponentColorDescriptorExtractor desc(sift);
 8    Mat descriptors;
 9    desc.compute(im, kpts, descriptors);

We done some updates in the opponent descriptor code, please try your code on 8263 revision. If the problem still occurs, please add code sample and image for reproducing.

  • Status changed from Open to Cancelled
  • Assignee changed from Maria Dimashova to Alexander Shishkov

Also available in: Atom PDF