So you decided to use OpenCV instead of JavaCV at the end?
Does your library require the OpenCV-app or are you including the native libs?
Enough questions....Look forward to trying it later on
So you decided to use OpenCV instead of JavaCV at the end?
Does your library require the OpenCV-app or are you including the native libs?
Enough questions....Look forward to trying it later on
Yes, It is OpenCv 3.20, with built-in native libs. Right now for armeabi-v7a and arm64-v8
There are a lot of methods to test yet, and a couple of modules which will need a harder work, as native camera. Example video is just CameraEx example with a few added lines
The modules available will be exactly those of OpenCV for android API, which are a more than enough for most of the needs/users
Not this one, since some modules depend on android and native libraries are built for arm.
But as the API must be (no idea, just guessing) quite the same, I think it will be possible to do something similar for B4J
@inakigarm
As @JordiCP wrote, it's a huge library (more like a framework) but it's already possible to use it with B4J. You could target just certain functionality and wrap necessary code for that functionality (guess you could use JavaObject/inlineJava too). I posted a video some time ago showing face-recognition doing just that.
Thanks wonder . Indeed, a new world of possibilities, only limited by imagination (and programming it, of course)
Unfortunately, not. I have used the already pre-built binaries for the current approach. Now I am making some experiments building the SDK again in order to learn more about the modules included in case I ever need to customize something
Well, I think that is exactly what @JordiCP is showing in the last video. You see the camera which is captioning the movements of the two objects. Before that you set up a configuration which determines what is the criteria of the object recognition(color, shape...)
Well, I think that is exactly what @JordiCP is showing in the last video. You see the camera which is captioning the movements of the two objects. Before that you set up a configuration which determines what is the criteria of the object recognition(color, shape...)
There are various "filters" one could apply but many of them require that you train them by furnishing lots of data....
If one could do what you suggested, I am not 100 % sure but it might be possible...
@inakigarm
As @JordiCP wrote, it's a huge library (more like a framework) but it's already possible to use it with B4J. You could target just certain functionality and wrap necessary code for that functionality (guess you could use JavaObject/inlineJava too). I posted a video some time ago showing face-recognition doing just that.
Just another test, facedetection example (original code ported from OpenCv java samples) using CascadeClassifier, and tuned to make a very simple AR app