Gesture Solver

This is an example to recognize gestures with the Android Sensors. The doing stuff:

1. Noise reducing of the sensor data (Pre-Gesture Classification)
2. Centroid Classification with the help of DTW, because we have time series (sensor data)
   • Hierarchical clustering (gesture are far apart (for example, a small and a tall person makes a gesture))
   • Centroid Calculation with DBA (DTW Barycenter Averaging)
   • Calculation Average distance to the to the Centroid for every Centroid
3. real-time-data classified with step 2 and the result is a recognized gesture

(!) It's more a collection of useful Code pieces than a working/runnable App (backend-side).

Include in Android APK

Because this project relies on JNI the JNI libs have to be included in the APK. Add following lines to your build.gralde. This will extract the shared object files from the JAR file into the intermediate directory.

task copyLocationSolverSharedObjectFiles(type: Copy)

copyLocationSolverSharedObjectFiles {
    description = 'Copies required shared object files into android jniLibs'
    from zipTree(System.getProperty("user.home") + "/.m2/repository/com/github/mleenings/navi/location-solver/$naviVersion/location-solver-${naviVersion}.jar")
    into "$buildDir/intermediates/locationSolverJniLibs/"
    include 'android/**'
}

preBuild.dependsOn copyLocationSolverSharedObjectFiles

And make sure that from the intermediate directory the files will be included as JNI libs. Add following lines to android {...}

android {
    sourceSets {
         main {
             jniLibs.srcDirs "$buildDir/intermediates/locationSolverJniLibs/android"
         }
     }
}