Ask Your Question

PhilLab's profile - activity

2020-10-23 13:34:23 -0500 received badge  Good Question (source)
2017-05-18 04:12:34 -0500 asked a question Contrib module in Android Release

On http://docs.opencv.org/3.1.0/d5/df8/t... it states

According to this approach all OpenCV binaries are included into your application package. It is designed mostly for development purposes. This approach is deprecated for the production code, release package is recommended to communicate with OpenCV Manager via the async initialization described

which (for me) translates to

Building Android OpenCV yourself is deprecated.

The problem is, that the official releases (neither in the manager nor in the provided libs) don't contain the contrib modules despite the build bot seems to build them everytime.

To the release management team:

Could you include the pre-built OpenCV contrib modules in the next release packages? If not in the OpenCV manager, at least in the libs?

2014-11-27 03:26:23 -0500 commented question Removal of BRIEF from main repository

Ah, thanks for the update :-)

2014-11-26 04:59:19 -0500 received badge  Nice Question (source)
2014-11-25 03:28:25 -0500 asked a question Is cv::flip working in place?

Hi, is cv::flip(matA,matA,0) guaranteed to work?

2014-11-23 07:02:23 -0500 received badge  Scholar (source)
2014-11-10 11:20:45 -0500 commented question newbie: lookin for process realtime camera capture image processing

(I know this is not an answer helping with your problem, but:) I would strongly recommend to use the new C++ interface - especially when you are a newby! The new Interface does not have IplImage and cvXYZ, but cv::XYZ !

2014-09-24 01:48:01 -0500 received badge  Nice Answer (source)
2014-09-23 16:24:19 -0500 received badge  Teacher (source)
2014-09-23 12:16:48 -0500 received badge  Editor (source)
2014-09-23 12:14:50 -0500 commented question NDK or SDK for the image processing

I think this is a valid question because in this forum there are experienced OpenCV users - and maybe some of them already have dealt with this design decision. Whether to swtich to the NDK highly depends on the use case - Google even says: "the NDK will not benefit most apps" - so this question is indeed an OpenCV question.

2014-09-23 12:10:52 -0500 answered a question NDK or SDK for the image processing

If you can stick with Java, then I would recomend it. Because then the nice guys from OpenCV have already saved you the work of setting up JNI and the JNI wrapper.

Debugging for your PC is much, much more convenient than debugging for native Android NDK code! So if you want to use the NDK (because of reasons rwong already mentioned), I would highly recommend to develop a desktop wrapper for your computer vision code so that you can run it both on Android and your PC. (Pro tip: this keeps your code modular).

2014-09-23 11:59:00 -0500 commented question Vision Challenge is unnoticed

Yeah, I already spread the word in my research group :-P Ok, looking forward to it!

2014-09-22 01:09:40 -0500 received badge  Good Question (source)
2014-09-21 09:28:28 -0500 received badge  Nice Question (source)
2014-09-21 03:04:10 -0500 received badge  Supporter (source)
2014-09-21 02:52:56 -0500 asked a question Vision Challenge is unnoticed

Hi, why is the ongoing OpenCV Vision Challenge not advertised? I only found it by coincidence in the development area. Why not announcing it on opencv.org, Twitter, ... ?

This seems like a waste of the $50.000 prize money.

2014-09-03 02:50:42 -0500 commented question PointsClouds Generated by Reporjectimageto3D is So Bad

In order to help you, you should provide parts of the relevant code. Debugging from just two images is hard :-)

2014-09-01 02:45:52 -0500 received badge  Student (source)
2014-09-01 02:37:04 -0500 asked a question Removal of BRIEF from main repository

Does anyone know why the BRIEF feature descriptor has been moved to the contrib repository? Is it nonfree? BRIEF is well-matured and used in tons of OpenCV examples both in the official documentation and the internet. Plus, the combination FAST+BRIEF is like a "Hello World" for feature detection IMHO :-).

EDIT: Additionally, the integration of the contrib repository seems still a little buggy - e.g. using xfeatures2d together with the world module fails in the CMake generation.