OpenCV 3.0 is getting closer, much closer. We are glad to announce OpenCV 3.0 rc, which will be followed by 3.0 final quite soon.

The release main goal was to finalise the API, simplify migration from OpenCV 2.4 and stabilize the code. We are still in the bug fixing mode, so the upcoming 3.0 should be even more stable. It's a good moment now to try the rc and report the found issues at http://code.opencv.org (or even fix them!)

These are the main changes since 3.0 beta (see also http://code.opencv.org/projects/opencv/wiki/ChangeLog):

  • Improved compatibility with 2.4, including smart pointer (cv::Ptr) API, support for trained opencv_ml 2.4 models etc. There are still some differences, assumed by the major version increment; check this transition guide: http://docs.opencv.org/3.0-rc1/db/dfa/tutorial_transition_guide.html
  • The new acceleration layer, OpenCV HAL, will help to accelerate OpenCV on various platforms. It will grow substantially during OpenCV 3.x lifetime; now it can be considered as technology preview thing. Yet, it can be already useful because of the so-called "universal intrinsics" that will let you to write a code that is optimized for both SSE and NEON.
  • Self-contained motion jpeg codec has been added, which could be a nice replacement for ffmpeg when it's not available or can not be used due to licensing restriction. While not providing the best compression ratio, it's quite efficient, even on ARM. Thanks to Alexander Kalistratov for the decoder part and to Yury Gorbachev for optimizing parts of the encoder!
  • There is solid Android support and even OpenCV3 Manager, thanks to Alexander Smorkalov, Maksim Shabunin and other guys.
  • Microsoft team, in particular Max Kostin, have contributed several important improvements and bug fixes for WinRT port of OpenCV.
  • Over 500 pull requests have been merged in since 3.0 beta. Big thanks to everybody who contributed the patches to opencv (github login's listed, you guys know who you are):

    opencv:
    Vladislav Vinogradov, Maksim Shabunin, Ilya Lavrenov,
    Erik Karlsson, Olexa Bilaniuk, Vadim Pisarevsky, Dmitry-Me,
    orestis, StevenPuttemans, Maxim Kostin, Yan Wang,
    Alexander Alekhin, Roman Donchenko, Ashod Nakashian,
    Joe Howse, Nisarg Thakkar, Andrey Kamaev, Dikay900,
    Alexander Nitsch, Evgeny Agafonchikov, TobyWanKenobi,
    theodore, Сергей Нужный, Andrew Senin, Artur Wieczorek,
    Hamid Bazargani, Pavel Rojtberg, berak, Alexander Smorkalov,
    Ben Hagen, Deanna Hood, Marvin Smith, Nicu Știurcă,
    Philipp Hasper, Simon Heinen, Takahiro Poly Horikawa,
    Ying Xiong, ASUS, Ahmatnurov Dinar, Alexander Duda,
    Ana Huaman Quispe, Eric Sommerlade, Naer Chang,
    Nils Plath, Sync-my-L2P, VBystricky, guopei, ippei ito,
    kalistratovag, karelknoest, sanuj, vbystricky, Adam Borowski,
    Adam Greig, Alexander Bohn / FI$H2k, Alexander Karsakov,
    Andrey Pavlenko, Ari Suwendi, Basheer Subei, Boaz Stolk,
    Chris Mayo, Chuanbo Weng, Claudio Caraffi, Daeyun Shin,
    Daniel Angelov, Daniel Sjölie, Dmitriy Anisimov, Florian Verdet,
    Guo Yejun, HelenWong, Jay Bosamiya, Jiri Drbalek,
    Jonathan Viney, Josep Bosch, Julien Nabet, Kai Wolf,
    Kevin Squire, LorenaGdL, Max Kostin, Michae Gharbi,
    Mike Izbicki, Nicolas Riebesel, Owen Healy, Pat O'Keefe,
    Pavel Vlasov, Prof. Dr. Rudolf Haussmann, Quentin Minster,
    RafaRuiz, Rok Mandeljc, S. Garrido, Samyak Datta,
    Sancho McCann, Scott Graybill, Shiquan Wang,
    Tim D. Smith, Timur Sattarov, Tzafrir, UQ Times,
    Yang Fan, burun, eltermann, exeltior, fhsm, greensea,
    hahne, mariadeanton, potpath, sergei, soyer, tcavallari,
    tt, ttagu99, vincentweb

    opencv_contrib:
    Maksim Shabunin, cbalint13, Auron-X, Dikay900, Str3iber,
    Dmitriy Anisimov, Jay Bosamiya, Rok Mandeljc,
    Vadim Pisarevsky, berak, Alfonso Sanchez-Beato,
    Antonella Cascitelli, Beat Küng, Dinar Ahmatnurov,
    Nisarg Thakkar, Sh0, Shuda Li, StevenPuttemans,
    Vladimir, Yan Wang, jiaolong_cvc228, pmkalshetti, yifita

    Special thanks goes to Steven Puttemans, who takes active part in submitting & reviewing patches, as well as in actually putting some answers to answers.opencv.org in order to make the site name more or less relevant :)

As usual, the latest version can be downloaded from SourceForge:

Enjoy!