Thanks to visit codestin.com
Credit goes to github.com

Skip to content

Improve contours approximations in ChessBoardDetector::generateQuads #25145

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Merged
merged 1 commit into from
Mar 6, 2024
Merged

Improve contours approximations in ChessBoardDetector::generateQuads #25145

merged 1 commit into from
Mar 6, 2024

Conversation

@MaximSmolskiy
Copy link
Contributor

@MaximSmolskiy MaximSmolskiy commented Mar 3, 2024

Pull Request Readiness Checklist

Firstly, there were bugs/problems:

  1. we always approximated contour at least once even if it was originally quadrangle
  2. we broke loop only if approx_contour.size() == 4, which was wrong, approx_contour.size() <= 4 - is correct

Secondly, approximation can be sped up. Before we approximated source contour twice on each loop iteration. But we can approximate source contour sequentially (on current iteration approximate output from previous iteration). The previous idea that for getting quadrangle we may need several approximations of source contour is taken into account here. And it will be much faster - because current countour size will decrease quite quickly with iterations.

I tested this PR (on 3.2.0 code version with some modern patches) on two private datasets and it sped up chessboards detection by 24% with minor changes in detected chessboards number:

  1. from 58 seconds to 44 seconds (without changes in detected chessboards number)
  2. from 142 seconds to 108 seconds (from 536 detected chessboards to 532 detected chessboards)

And with benchmark

python3 objdetect_benchmark.py --configuration=generate_run --board_x=7 --path=res_chessboard --synthetic_object=chessboard

PR sped up chessboards detection by 12-16% with minor changes in detected chessboards number (majority is in favor of PR):

cell_img_size = 100 (default)

before
                                 category  detected chessboard  total detected chessboard  total chessboard  average detected error chessboard
                                      all             0.904167                      13020             14400                           0.600512
Total detected time:  127.96504 sec

after

                                 category  detected chessboard  total detected chessboard  total chessboard  average detected error chessboard
                                      all             0.904167                      13020             14400                           0.600512
Total detected time:  109.11559199999988 sec

----------------------------------------------------------------------------------------------------------------------------------------------

cell_img_size = 10

before
                                 category  detected chessboard  total detected chessboard  total chessboard  average detected error chessboard
                                      all             0.541875                       7803             14400                           4.199676
Total detected time:  3.6333469999999997 sec

after
                                 category  detected chessboard  total detected chessboard  total chessboard  average detected error chessboard
                                      all             0.539792                       7773             14400                           4.209964
Total detected time:  3.0567529999999987 sec

----------------------------------------------------------------------------------------------------------------------------------------------

cell_img_size = 9

before
                                 category  detected chessboard  total detected chessboard  total chessboard  average detected error chessboard
                                      all             0.512292                       7377             14400                           3.375697
Total detected time:  3.4287690000000013 sec

after
                                 category  detected chessboard  total detected chessboard  total chessboard  average detected error chessboard
                                      all             0.516458                       7437             14400                           3.378223
Total detected time:  2.9598780000000016 sec

----------------------------------------------------------------------------------------------------------------------------------------------

cell_img_size = 8

before
                                 category  detected chessboard  total detected chessboard  total chessboard  average detected error chessboard
                                      all             0.474444                       6832             14400                           3.246657
Total detected time:  3.0795969999999997 sec

after
                                 category  detected chessboard  total detected chessboard  total chessboard  average detected error chessboard
                                      all             0.474444                       6832             14400                           3.229681
Total detected time:  2.6868730000000016 sec

----------------------------------------------------------------------------------------------------------------------------------------------

cell_img_size = 7

before
                                 category  detected chessboard  total detected chessboard  total chessboard  average detected error chessboard
                                      all             0.398889                       5744             14400                           3.421867
Total detected time:  2.7457670000000003 sec

after
                                 category  detected chessboard  total detected chessboard  total chessboard  average detected error chessboard
                                      all             0.400972                       5774             14400                           3.422815
Total detected time:  2.4135210000000034 sec

----------------------------------------------------------------------------------------------------------------------------------------------

cell_img_size = 6

before
                                 category  detected chessboard  total detected chessboard  total chessboard  average detected error chessboard
                                      all             0.268472                       3866             14400                           3.240348
Total detected time:  2.388680000000001 sec

after
                                 category  detected chessboard  total detected chessboard  total chessboard  average detected error chessboard
                                      all             0.272708                       3927             14400                           3.236853
Total detected time:  2.0548080000000017 sec

----------------------------------------------------------------------------------------------------------------------------------------------

cell_img_size = 5

before
                                 category  detected chessboard  total detected chessboard  total chessboard  average detected error chessboard
                                      all             0.137500                       1980             14400                           3.028695
Total detected time:  1.9060430000000004 sec

after
                                 category  detected chessboard  total detected chessboard  total chessboard  average detected error chessboard
                                      all             0.137500                       1980             14400                           3.030068
Total detected time:  1.6219549999999998 sec

See details at https://github.com/opencv/opencv/wiki/How_to_contribute#making-a-good-pull-request

  • I agree to contribute to the project under Apache 2 License.
  • To the best of my knowledge, the proposed patch is not based on a code under GPL or another license that is incompatible with OpenCV
  • The PR is proposed to the proper branch
  • There is a reference to the original bug report and related work
  • There is accuracy test, performance test and test data in opencv_extra repository, if applicable
    Patch to opencv_extra has the same branch name.
  • The feature is well documented and sample code can be built with the project CMake

@asmorkalov asmorkalov added this to the 4.10.0 milestone Mar 4, 2024
@mshabunin
Copy link
Contributor

mshabunin commented Mar 4, 2024

Does this PR replaces #24760?

@MaximSmolskiy
Copy link
Contributor Author

MaximSmolskiy commented Mar 4, 2024

Does this PR replaces #24760?

Yes, I was going to close it after merging this PR

AleksandrPanov
AleksandrPanov approved these changes Mar 6, 2024
View reviewed changes
Copy link
Contributor

@AleksandrPanov AleksandrPanov left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

👍

@AleksandrPanov
Copy link
Contributor

AleksandrPanov commented Mar 6, 2024

@MaximSmolskiy, thank you for the good PR ^_^

@AleksandrPanov
Copy link
Contributor

AleksandrPanov commented Mar 6, 2024

I think that, before calling approxPolyDP(), we can check how close the contour is to a square/convex shape using boundingRect()/minAreaRect()/convexHull(). We can then drop some of the poor-quality contours at an earlier stage.

This is a suggestion for future improvements.

@asmorkalov asmorkalov merged commit 0e524ee into opencv:4.x Mar 6, 2024
@MaximSmolskiy MaximSmolskiy deleted the improve-contours-approximations-in-ChessBoardDetector-generateQuads-2 branch March 6, 2024 11:17
@asmorkalov asmorkalov mentioned this pull request Apr 1, 2024
Merged
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@AleksandrPanov AleksandrPanov AleksandrPanov approved these changes

Assignees

@AleksandrPanov AleksandrPanov

Labels

category: calib3d optimization

Projects

None yet

Milestone

4.10.0

Development

Successfully merging this pull request may close these issues.

4 participants

@MaximSmolskiy @mshabunin @AleksandrPanov @asmorkalov