align-toparrow-leftarrow-rightbackbellblockcalendarcamerachatcheckchevron-downchevron-leftchevron-rightchevron-small-downchevron-small-leftchevron-small-rightchevron-small-upchevron-upcircle-with-crosscrosseditfacebookglobegoogleimagesinstagramlocation-pinmagnifying-glassmailmoremuplabelShape 3 + Rectangle 1outlookpersonplusImported LayersImported LayersImported Layersshieldstartwitteryahoo

Jigsaw Challenge

This time we'll be showing our progress on writing software to solve jigsaw puzzles.  Bring your own work to share.

Join or login to comment.

  • JJ S.

    Here is a start on the OCR challenge: http://www.jjsland.com/opencv/challenge_06/

    August 31, 2014

  • JJ S.

    We didn't show much progress on the jigsaw. I did show how my Spot-it solution was able to show the puzzle pieces via countouring. But we notice that the background tends to be difficult to remove - especially since the puzzle pieces themselves have a lot of details that are picked up via contours.

    We did review some results of the Spot-it Challenge. Soheil was able to find the Clover from the sample by using contouring and moments. Soheil shared his results at https://github.com/seizadi/spotit

    We did review some of the previous work John was working on for the Object Tracking Challenge.

    Arun provided much input about algorithms for both computer vision and machine learning including the recommendation for Tessaract OCR which may be used in the next challenge.

    We decided to work on OCR for the next challenge. Specifically (or generally), read the receipt. More to come, due date Sept 29th.

    August 30, 2014

    • JJ S.

      August 30, 2014

    • JJ S.

      August 30, 2014

  • JJ S.

    We were fortunate to have a few members of the ROS team drop in on our discussion. They showed us the Kinect and we learned how it can be used as a depth-camera. Can this be used for the Object Tracking Challenge? John was a bit hesitant, wanting to continue focus without a depth camera. I currently am without a depth camera as well - although the Kinect is certainly an affordable option.

    http://www.meetup.com/SV-ROS-users/

    are participating in

    http://www.iros2014.org/program/kinect-robot-navigation-contest

    August 30, 2014

  • Stoney V.

    It takes a long time to compile openCV 2.4.9 or openCV 3.0. It is not necessary to use the newer openCV 3.0. One issue that I encountered while compiling openCV 3.0 was the need to use ffmpeg instead of the forked libav. ffmpeg developers say that they frequently pull request libav to maintain compatibility.
    You can add John Severinsson's ffmpeg PPA for the Ubuntu version that you are using so that you get updates. This site has instructions for installing ffmpeg on multiple platforms. https://www.ffmpeg.org/download.html

    August 24, 2014

    • JJ S.

      Its like 10x faster on a SSD over a HD

      August 30, 2014

  • John B.

    Thanks to everyone who came out. Our next meeting will be on Monday, September 29th, where we'll be looking at scanning and OCR-ing receipts. Look for a formal meeting announcement soon.

    August 25, 2014

  • JJ S.

    http://opencv.org/opencv-3-0-alpha.html

    Can we discuss the new features of 3.0 that are new compared to 2.4.9?

    -----

    I just installed opencv on my desktop and had / am having some difficulty.

    Problem Solved: $ cmake opencv-master fails because ippicv cannot download.

    It appears that sourceforge.net was unable to serve the download.

    Solution: I used sourceforge.net to download the .tgz from another mirror and put the file in the 3rdparty/ippicp/linux-d80c.... directory.

    Now the cmake succeeds.

    The make and make install succeeds without glitches.

    $ pkg-config --modversion opencv
    3.0.0

    I was expecting 2.4.9, which I have on my laptop.

    -----

    New Problem: At this point, my source fails to compile due to, first: CV_GRAY2BGR was not declared in this scope.

    I am assuming I need to modify my code to work in 3.0.0 versus 2.4.9

    August 23, 2014

    • JJ S.

      from imgproc.hpp
      CV_GRAY2BGR -> COLOR_GRAY2BGR
      CV_BGR2GRAY -> COLOR_BGR2GRAY
      CV_HOUGH_GRADIENT -> HOUGH_GRADIENT
      CV_RETR_EXTERNAL -> RETR_EXTERNAL
      CV_CHAIN_APPROX_SIMPLE -> CHAIN_APPROX_SIMPLE from core.hpp
      CV_AA -> LINE_AA

      from highgui.hpp
      CV_WINDOW_AUTOSIZE -> WINDOW_AUTOSIZE

      August 23, 2014

    • JJ S.

      those are the changes that effect my current code

      August 23, 2014

  • John B.

    Sorry, forgot to add that. We'll be meeting at Hacker Dojo in Mountain View.

    August 20, 2014

  • A former member
    A former member

    location ?

    August 19, 2014

  • JJ S.

    I have a couple small puzzles here. They are a little grainy, but they should be good to start.

    http://www.jjsland.com/opencv/challenge_05/

    August 14, 2014

8 went

People in this
Meetup are also in:

Sign up

Meetup members, Log in

By clicking "Sign up" or "Sign up using Facebook", you confirm that you accept our Terms of Service & Privacy Policy