sunzuolei / mvo_android

The monocular visual odometry implemented on an Android phone.

mvo_android

The monocular visual odometry implemented on an Android phone.

1.Dependency

• Eclipse with ADT、CDT;

• android-ndk-r10e(NDK);

• OpenCV4Android 3.0.0 or 3.1.0;

  Here to download OpenCV4Android

• OpenCV Manger

2.BuildProject

To build mvo_android, you can simply import cloned project into eclipse.I've already imported and set opencv3.1.0 as library,so you can jump this step.After that,you need to change the path to your OpenCV.mk in Android.mk. Config and run ndk-build .

3.Function

Through this application , you can calcuate the position and attitude of camera.Moreover,it provides feature detection constantly.

• Camera Mode

  Using mobile camera to get image inputFrame.

  camera mode runtime screenshot

  

• Local Mode

  Get inputFrame through local storage directory.

  local mode runtime screenshot

  

  The KITTI benchmark data* is used for MVO test.
  *KITTI data is well known as benchmark data for computer vision algorithm evaluation,it is available at http://www.cvlibs.net/datasets/kitti/

4.Demo

Here is the video demo for this application

5.APK(armeabi-v7a)

If you want to try Visdo simply,you can download and install Visodo.apk and OpencvManager.apk on your android phone