<div dir="ltr"><div><div><div><div><div>Hi all,<br><br></div>as you probably know by now, OpenCV3 got released a few months ago. To summarize what matters to the ROS community, OpenCV3 is:<br></div>- a big cleanup of the C API (which also comes with forward compatibility brought to OpenCV 2.4.9+; meaning it's usually easy to write code that compiles with OpenCV 2 and 3, like Python did between 2.7 and 3.x)<br>- more performance (many algorithms optimized, transparent GPU utilization ...)<br></div></div>- new algorithms especially in opencv_apps, cf the GSOC videos since 2013: ( [1], [2], [3] ), especially a few that seem very relevant to the ROS community: RGBD processing, 3d visualization, deep learning support, object recognition, calibration (faster, omni-cam support ...), better/faster stereo ....<br></div><div>And more details in the announcement [4]<br></div><div><br></div>Now, all that seems like a very good reason to switch to OpenCV3 by default for the next release of ROS: Kinetic. Please, let's discuss it on the SIG as it might get technical:<br><a href="https://groups.google.com/forum/#!topic/ros-sig-perception/K5__71SX7eU">https://groups.google.com/forum/#!topic/ros-sig-perception/K5__71SX7eU</a><br><div>(and feel free to create non-technical threads about the subject on the SIG too)<br></div><div><br><br>[1] <a href="https://www.youtube.com/watch?v=_TTtN4frMEA">https://www.youtube.com/watch?v=_TTtN4frMEA</a><br>[2] <a href="https://www.youtube.com/watch?v=3f76HCHJJRA">https://www.youtube.com/watch?v=3f76HCHJJRA</a><br>[3] <a href="https://www.youtube.com/watch?v=OUbUFn71S4s">https://www.youtube.com/watch?v=OUbUFn71S4s </a><br>[4] <a href="http://opencv.org/opencv-3-0.html">http://opencv.org/opencv-3-0.html</a><br></div></div>