OpenCV in Android Studio

Question

The below steps for using Android OpenCV sdk in Android Studio. This is a simplified version of this(1) SO answer.

1. Download latest OpenCV sdk for Android from OpenCV.org and decompress the zip file.
2. Import OpenCV to Android Studio, From File -> New -> Import Module, choose sdk/java folder in the unzipped opencv archive.
3. Update build.gradle under imported OpenCV module to update 4 fields to match your project build.gradle a) compileSdkVersion b) buildToolsVersion c) minSdkVersion and d) targetSdkVersion.
4. Add module dependency by Application -> Module Settings, and select the Dependencies tab. Click + icon at bottom, choose Module Dependency and select the imported OpenCV module.
   • For Android Studio v1.2.2, to access to Module Settings : in the project view, right-click the dependent module -> Open Module Settings
5. Copy libs folder under sdk/native to Android Studio under app/src/main.
6. In Android Studio, rename the copied libs directory to jniLibs and we are done.

Step (6) is since Android studio expects native libs in app/src/main/jniLibs instead of older libs folder. For those new to Android OpenCV, don't miss below steps

• include static{ System.loadLibrary("opencv_java"); } (Note: for OpenCV version 3 at this step you should instead load the library opencv_java3.)
• For step(5), if you ignore any platform libs like x86, make sure your device/emulator is not on that platform.

OpenCV written is in C/C++. Java wrappers are

1. Android OpenCV SDK - OpenCV.org maintained Android Java wrapper. I suggest this one.
2. OpenCV Java - OpenCV.org maintained auto generated desktop Java wrapper.
3. JavaCV - Popular Java wrapper maintained by independent developer(s). Not Android specific. This library might get out of sync with OpenCV newer versions.

Integrating OpenCV v3.1.0 into Android Studio v1.4.1, instructions with additional detail and this-is-what-you-should-get type screenshots.

Most of the credit goes to Kiran, Kool, 1", and SteveLiles over at opencv.org for their explanations. I'm adding this answer because I believe that Android Studio's interface is now stable enough to work with on this type of integration stuff. Also I have to write these instructions anyway for our project.

Experienced A.S. developers will find some of this pedantic. This answer is targeted at people with limited experience in Android Studio.

1. Create a new Android Studio project using the project wizard (Menu:/File/New Project):

   • Call it "cvtest1"
   • Form factor: API 19, Android 4.4 (KitKat)

   • Blank Activity named MainActivity

     You should have a cvtest1 directory where this project is stored. (the title bar of Android studio shows you where cvtest1 is when you open the project)

2. Verify that your app runs correctly. Try changing something like the "Hello World" text to confirm that the build/test cycle is OK for you. (I'm testing with an emulator of an API 19 device).

3. Download the OpenCV package for Android v3.1.0 and unzip it in some temporary directory somewhere. (Make sure it is the package specifically for Android and not just the OpenCV for Java package.) I'll call this directory "unzip-dir" Below unzip-dir you should have a sdk/native/libs directory with subdirectories that start with things like arm..., mips... and x86... (one for each type of "architecture" Android runs on)

4. From Android Studio import OpenCV into your project as a module: Menu:/File/New/Import_Module:

   • Source-directory: {unzip-dir}/sdk/java
   • Module name: Android studio automatically fills in this field with openCVLibrary310 (the exact name probably doesn't matter but we'll go with this).

   • Click on next. You get a screen with three checkboxes and questions about jars, libraries and import options. All three should be checked. Click on Finish.

     Android Studio starts to import the module and you are shown an import-summary.txt file that has a list of what was not imported (mostly javadoc files) and other pieces of information.

     But you also get an error message saying failed to find target with hash string 'android-14'.... This happens because the build.gradle file in the OpenCV zip file you downloaded says to compile using android API version 14, which by default you don't have with Android Studio v1.4.1.

5. Open the project structure dialogue (Menu:/File/Project_Structure). Select the "app" module, click on the Dependencies tab and add :openCVLibrary310 as a Module Dependency. When you select Add/Module_Dependency it should appear in the list of modules you can add. It will now show up as a dependency but you will get a few more cannot-find-android-14 errors in the event log.

6. Look in the build.gradle file for your app module. There are multiple build.gradle files in an Android project. The one you want is in the cvtest1/app directory and from the project view it looks like build.gradle (Module: app). Note the values of these four fields:

   • compileSDKVersion (mine says 23)
   • buildToolsVersion (mine says 23.0.2)
   • minSdkVersion (mine says 19)
   • targetSdkVersion (mine says 23)

7. Your project now has a cvtest1/OpenCVLibrary310 directory but it is not visible from the project view:

Use some other tool, such as any file manager, and go to this directory. You can also switch the project view from Android to Project Files and you can find this directory as shown in this screenshot:

Inside there is another build.gradle file (it's highlighted in the above screenshot). Update this file with the four values from step 6.

8. Resynch your project and then clean/rebuild it. (Menu:/Build/Clean_Project) It should clean and build without errors and you should see many references to :openCVLibrary310 in the 0:Messages screen.

   

   At this point the module should appear in the project hierarchy as openCVLibrary310, just like app. (Note that in that little drop-down menu I switched back from Project View to Android View ). You should also see an additional build.gradle file under "Gradle Scripts" but I find the Android Studio interface a little bit glitchy and sometimes it does not do this right away. So try resynching, cleaning, even restarting Android Studio.

   You should see the openCVLibrary310 module with all the OpenCV functions under java like in this screenshot:

   

9. Copy the {unzip-dir}/sdk/native/libs directory (and everything under it) to your Android project, to cvtest1/OpenCVLibrary310/src/main/, and then rename your copy from libs to jniLibs. You should now have a cvtest1/OpenCVLibrary310/src/main/jniLibs directory. Resynch your project and this directory should now appear in the project view under openCVLibrary310.

   

10. Go to the onCreate method of MainActivity.java and append this code:

    if (!OpenCVLoader.initDebug()) {
        Log.e(this.getClass().getSimpleName(), "  OpenCVLoader.initDebug(), not working.");
    } else {
        Log.d(this.getClass().getSimpleName(), "  OpenCVLoader.initDebug(), working.");
    }
    

    Then run your application. You should see lines like this in the Android Monitor: (I don't know why that line with the error message is there)

11. Now try to actually use some openCV code. In the example below I copied a .jpg file to the cache directory of the cvtest1 application on the android emulator. The code below loads this image, runs the canny edge detection algorithm and then writes the results back to a .png file in the same directory.

    Put this code just below the code from the previous step and alter it to match your own files/directories.

    String inputFileName="simm_01";
    String inputExtension = "jpg";
    String inputDir = getCacheDir().getAbsolutePath();  // use the cache directory for i/o
    String outputDir = getCacheDir().getAbsolutePath();
    String outputExtension = "png";
    String inputFilePath = inputDir + File.separator + inputFileName + "." + inputExtension;
    
    
    Log.d (this.getClass().getSimpleName(), "loading " + inputFilePath + "...");
    Mat image = Imgcodecs.imread(inputFilePath);  
    Log.d (this.getClass().getSimpleName(), "width of " + inputFileName + ": " + image.width());
    // if width is 0 then it did not read your image.
    
    
    // for the canny edge detection algorithm, play with these to see different results
    int threshold1 = 70;
    int threshold2 = 100;
    
    Mat im_canny = new Mat();  // you have to initialize output image before giving it to the Canny method
    Imgproc.Canny(image, im_canny, threshold1, threshold2);
    String cannyFilename = outputDir + File.separator + inputFileName + "_canny-" + threshold1 + "-" + threshold2 + "." + outputExtension;
    Log.d (this.getClass().getSimpleName(), "Writing " + cannyFilename);
    Imgcodecs.imwrite(cannyFilename, im_canny);
    

12. Run your application. Your emulator should create a black and white "edge" image. You can use the Android Device Monitor to retrieve the output or write an activity to show it.

The Gotchas:

• If you lower your target platform below KitKat some of the OpenCV libraries will no longer function, specifically the classes related to org.opencv.android.Camera2Renderer and other related classes. You can probably get around this by simply removing the apprpriate OpenCV .java files.
• If you raise your target platform to Lollipop or above my example of loading a file might not work because use of absolute file paths is frowned upon. So you might have to change the example to load a file from the gallery or somewhere else. There are numerous examples floating around.

For everyone who felt they want to run away with all the steps and screen shots on the (great!) above answers, this worked for me with android studio 2.2.1:

1. Create a new project, name it as you want and take the default (minSdkVersion 15 is fine).

2. Download the zip file from here: https://sourceforge.net/projects/opencvlibrary/files/opencv-android/ (I downloaded 3.2.0 version, but there may be a newer versions).

3. Unzip the zip file, the best place is in your workspace folder, but it not really matter.

4. Inside Android Studio, click File->New-> Import Module and navigate to \path_to_your_unzipped_file\OpenCV-android-sdk\sdk\java and hit Ok, then accept all default dialogs.

5. In the gradle file of your app module, add this to the dependencies block:

    dependencies {
        compile project(':openCVLibraryXYZ')
        //rest of code
    }
   

Where XYZ is the exact version you downloaded, for example in my case:

    dependencies {
        compile project(':openCVLibrary320')
        //rest of code
    }

Android Studio 3.4 + OpenCV 4.1

1. Download the latest OpenCV zip file from here (current newest version is 4.1.0) and unzip it in your workspace or in another folder.

2. Create new Android Studio project normally. Click File->New->Import Module, navigate to /path_to_unzipped_files/OpenCV-android-sdk/sdk/java, set Module name as opencv, click Next and uncheck all options in the screen.

3. Enable Project file view mode (default mode is Android). In the opencv/build.gradle file change apply plugin: 'com.android.application' to apply plugin: 'com.android.library' and replace application ID "org.opencv" with

   minSdkVersion 21
   targetSdkVersion 28
   

   (according the values in app/build.gradle). Sync project with Gradle files.

4. Add this string to the dependencies block in the app/build.gradle file

   dependencies {
       ...
       implementation project(path: ':opencv')
       ...
   }
   

5. Select again Android file view mode. Right click on app module and goto New->Folder->JNI Folder. Select change folder location and set src/main/jniLibs/.

6. Select again Project file view mode and copy all folders from /path_to_unzipped_files/OpenCV-android-sdk/sdk/native/libs to app/src/main/jniLibs.

7. Again in Android file view mode right click on app module and choose Link C++ Project with Gradle. Select Build System ndk-build and path to OpenCV.mk file /path_to_unzipped_files/OpenCV-android-sdk/sdk/native/jni/OpenCV.mk.

   path_to_unzipped_files must not contain any spaces, or you will get error!

To check OpenCV initialization add Toast message in MainActivity onCreate() method:

Toast.makeText(MainActivity.this, String.valueOf(OpenCVLoader.initDebug()), Toast.LENGTH_LONG).show();

If initialization is successful you will see true in Toast message else you will see false.