Javacpp-presets: opencv_cudaoptflow produces no jniopencv_ximgproc in java.library.path

[n-kai-cj]

I make a opencv + cuda test code like below and confirmed to success of opencv_cudafeatures2d and opencv_cudaimgproc, but opencv_cudaoptflow failed with java.lang.UnsatisfiedLinkError: no jniopencv_ximgproc in java.library.path: [...].
Does anybody know why?

import org.bytedeco.opencv.global.opencv_core;
import org.bytedeco.opencv.global.opencv_cudaimgproc;
import org.bytedeco.opencv.global.opencv_imgproc;
import org.bytedeco.opencv.opencv_core.GpuMat;
import org.bytedeco.opencv.opencv_cudaoptflow.FarnebackOpticalFlow;
import org.bytedeco.opencv.opencv_cudafeatures2d.ORB;
import org.bytedeco.opencv.opencv_cudaimgproc.TemplateMatching;

public class Main {
    public static void main(String[] args) {
        GpuMat mat_cuda = new GpuMat();
        System.out.println("mat_cuda succeed");
        ORB orb_cuda = ORB.create();
        System.out.println("orb_cuda succeed");
        TemplateMatching temp_cuda = opencv_cudaimgproc.createTemplateMatching(opencv_core.CV_8UC1, opencv_imgproc.TM_SQDIFF);
        System.out.println("temp_cuda succeed");
        FarnebackOpticalFlow optflow_cuda = FarnebackOpticalFlow.create();
        System.out.println("optflow_cuda succeed");
    }
}

run

mat_cuda succeed
orb_cuda succeed
temp_cuda succeed
java.lang.UnsatisfiedLinkError: no jniopencv_ximgproc in java.library.path: [...]
    at java.lang.ClassLoader.loadLibrary (ClassLoader.java:2670)
    at java.lang.Runtime.loadLibrary0 (Runtime.java:830)
    at java.lang.System.loadLibrary (System.java:1870)
    at org.bytedeco.javacpp.Loader.loadLibrary (Loader.java:1631)
    at org.bytedeco.javacpp.Loader.load (Loader.java:1265)
    at org.bytedeco.javacpp.Loader.load (Loader.java:1109)
    at org.bytedeco.opencv.global.opencv_ximgproc.<clinit> (opencv_ximgproc.java:31)
    at java.lang.Class.forName0 (Native Method)
    at java.lang.Class.forName (Class.java:398)
    at org.bytedeco.javacpp.Loader.load (Loader.java:1176)
    at org.bytedeco.javacpp.Loader.load (Loader.java:1109)
    at org.bytedeco.opencv.opencv_cudaoptflow.DenseOpticalFlow.<clinit> (DenseOpticalFlow.java:57)
    at Main.main (Main.java:17)
    at org.codehaus.mojo.exec.ExecJavaMojo$1.run (ExecJavaMojo.java:254)
    at java.lang.Thread.run (Thread.java:834)
Caused by: java.lang.UnsatisfiedLinkError: C:\Users\-\.javacpp\cache\opencv-4.3.0-1.5.3-windows-x86_64.jar\org\bytedeco\opencv\windows-x86_64\jniopencv_ximgproc.dll: The specified procedure could not be found
    at java.lang.ClassLoader$NativeLibrary.load0 (Native Method)
    at java.lang.ClassLoader$NativeLibrary.load (ClassLoader.java:2442)
    at java.lang.ClassLoader$NativeLibrary.loadLibrary (ClassLoader.java:2498)
    at java.lang.ClassLoader.loadLibrary0 (ClassLoader.java:2694)
    at java.lang.ClassLoader.loadLibrary (ClassLoader.java:2627)
    at java.lang.Runtime.load0 (Runtime.java:768)
    at java.lang.System.load (System.java:1834)
    at org.bytedeco.javacpp.Loader.loadLibrary (Loader.java:1581)
    at org.bytedeco.javacpp.Loader.load (Loader.java:1265)
    at org.bytedeco.javacpp.Loader.load (Loader.java:1109)
    at org.bytedeco.opencv.global.opencv_ximgproc.<clinit> (opencv_ximgproc.java:31)
    at java.lang.Class.forName0 (Native Method)
    at java.lang.Class.forName (Class.java:398)
    at org.bytedeco.javacpp.Loader.load (Loader.java:1176)
    at org.bytedeco.javacpp.Loader.load (Loader.java:1109)
    at org.bytedeco.opencv.opencv_cudaoptflow.DenseOpticalFlow.<clinit> (DenseOpticalFlow.java:57)
    at Main.main (Main.java:17)
    at org.codehaus.mojo.exec.ExecJavaMojo$1.run (ExecJavaMojo.java:254)
    at java.lang.Thread.run (Thread.java:834)

[saudet]

You might not have the right version of CUDA installed. Please add a dependency on cuda-redist-platform.

[n-kai-cj]

Thank you for quick reply. As you said, my CUDA version (10.2.89) is different from javacpp-cuda version (10.2.76).
Unfortunately CUDA 10.2.76 seems not to be available for now, so I can't try to re-install the same version.
Could you try to run my above code on your PC if you have 10.2.76 version of CUDA ?

$ nvcc --version
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2019 NVIDIA Corporation
Built on Wed_Oct_23_19:32:27_Pacific_Daylight_Time_2019
Cuda compilation tools, release 10.2, V10.2.89

my pom.xml is below.
```xml:pom.xml

4.0.0
org.bytedeco.opencv
optflowcuda
1.5.3

Main
11
11



org.bytedeco
opencv-platform
4.3.0-1.5.3

    <!-- Additional dependencies required to use CUDA and cuDNN -->
    <dependency>
        <groupId>org.bytedeco</groupId>
        <artifactId>opencv-platform-gpu</artifactId>
        <version>4.3.0-1.5.3</version>
    </dependency>

    <!-- Additional dependencies to use bundled CUDA and cuDNN -->
    <dependency>
        <groupId>org.bytedeco</groupId>
        <artifactId>cuda-platform-redist</artifactId>
        <version>10.2-7.6-1.5.3</version>
    </dependency>

    <!-- Additional dependencies to use bundled full version of MKL -->
    <dependency>
        <groupId>org.bytedeco</groupId>
        <artifactId>mkl-platform-redist</artifactId>
        <version>2020.1-1.5.3</version>
    </dependency>

    <!-- Optional dependencies to load the Python module -->
    <dependency>
        <groupId>org.bytedeco</groupId>
        <artifactId>numpy-platform</artifactId>
        <version>1.18.2-1.5.3</version>
    </dependency>

</dependencies>
<build>
    <sourceDirectory>.</sourceDirectory>
</build>

```

[saudet]

We don't need to install CUDA with cuda-redist-platform. Maybe there's some other dependency missing. Please let me know what the Dependencies tool says: https://github.com/bytedeco/javacpp-presets/wiki/Debugging-UnsatisfiedLinkError-on-Windows

[n-kai-cj]

Nice tool, thank you! I tested like below and it's not working yet...

It seemed to load my prebuilt OpenCV libraries that are different version from javacpp-opencv.

1. remove my prebuilt OpenCV from PATH and run.

   

   missing libraries are:

   • opencv_imgcodecs430.dll
   • opencv_video430.dll
   • opencv_calib3d430.dll
   • opencv_features2d430.dll
   • opencv_flann430.dll
   • opencv_imgproc430.dll
   • opencv_core430.dll

2. copy all dll from cache\opencv-4.3.0-1.5.3-windows-x86_64-gpu.jar\org\bytedeco\opencv\windows-x86_64-gpu

   I have some of missing libraries in opencv-gpu cache, so copy all dll to working directory and run.

   

   missing libraries are:

   • opencv_video430.dll

   error message has changed

   mat_cuda succeed
   orb_cuda succeed
   temp_cuda succeed
   java.lang.NoClassDefFoundError: Could not initialize class org.bytedeco.opencv.global.opencv_ximgproc
       at java.lang.Class.forName0 (Native Method)
       at java.lang.Class.forName (Class.java:398)
       at org.bytedeco.javacpp.Loader.load (Loader.java:1176)
       at org.bytedeco.javacpp.Loader.load (Loader.java:1109)
       at org.bytedeco.opencv.opencv_cudaoptflow.DenseOpticalFlow.<clinit> (DenseOpticalFlow.java:57)
       at Main.main (Main.java:32)
       at org.codehaus.mojo.exec.ExecJavaMojo$1.run (ExecJavaMojo.java:254)
       at java.lang.Thread.run (Thread.java:834)
   

I have 2 questions.

1. Why didn't they load opencv-gpu cache?
2. Why is opencv_video430.dll missing in opencv cache?

[saudet]

Those are probably the files that have issues loading. Please check the dependencies of those modules.

[n-kai-cj]

I totally forgot the documentation saying (https://github.com/bytedeco/javacpp-presets/tree/master/opencv#documentation).

∗ Call Loader.load(opencv_java.class) before using the API in the org.opencv namespace.

Finally my fixed code works well.

```java:Main.java
import org.bytedeco.javacpp.Loader;
import org.bytedeco.opencv.global.opencv_core;
import org.bytedeco.opencv.global.opencv_cudaimgproc;
import org.bytedeco.opencv.global.opencv_imgproc;
import org.bytedeco.opencv.opencv_core.GpuMat;
import org.bytedeco.opencv.opencv_cudaoptflow.FarnebackOpticalFlow;
import org.bytedeco.opencv.opencv_cudafeatures2d.ORB;
import org.bytedeco.opencv.opencv_cudaimgproc.TemplateMatching;
import org.bytedeco.opencv.opencv_java;

import java.io.IOException;

public class Main {
public static void main(String[] args) {
Loader.load(opencv_java.class);
try {
Loader.load(org.bytedeco.opencv.opencv_cudaoptflow.FarnebackOpticalFlow.class);
} catch (UnsatisfiedLinkError e) {
try {
String path = Loader.cacheResource(org.bytedeco.opencv.opencv_cudaoptflow.FarnebackOpticalFlow.class, "windows-x86_64/jniopencv_ximgproc.dll").getPath();
new ProcessBuilder("c:/DependenciesGui.exe", path).start().waitFor();
} catch (IOException|InterruptedException e1) {
e1.printStackTrace();
}
}
GpuMat mat_cuda = new GpuMat();
System.out.println("mat_cuda succeed");
ORB orb_cuda = ORB.create();
System.out.println("orb_cuda succeed");
TemplateMatching temp_cuda = opencv_cudaimgproc.createTemplateMatching(opencv_core.CV_8UC1, opencv_imgproc.TM_SQDIFF);
System.out.println("temp_cuda succeed");
FarnebackOpticalFlow optflow_cuda = FarnebackOpticalFlow.create();
System.out.println("optflow_cuda succeed");
}
}

and run

mat_cuda succeed
orb_cuda succeed
temp_cuda succeed
optflow_cuda succeed
```

Thank you for your help @saudet !!

[saudet]

We shouldn't need to load opencv_java for this. I've narrowed down the issue to opencv_ximgproc actually depending on opencv_video, so we can call Loader.load(opencv_video.class) to work around this. I'll be fixing this. Thanks for reporting!

[saudet]

Fix included in version 1.5.4! Thanks for reporting