SVM预测图像OpenCV

时间:2014-05-21 07:23:33

标签: java opencv svm

我能够训练系统但是当我尝试预测时,会引发Bad参数异常。

OpenCV错误:cvPreparePredictData中的错误参数(示例不是有效向量),文件........ \ opencv \ modules \ ml \ src \ inner_functions.cpp,第1099行 线程“main”中的异常CvException [org.opencv.core.CvException:cv :: Exception:........ \ opencv \ modules \ ml \ src \ inner_functions.cpp:1099:错误:( - 5)该示例不是函数cvPreparePredictData中的有效向量 ]

这是我的代码:

        System.loadLibrary(Core.NATIVE_LIBRARY_NAME);

        Mat classes = new Mat();
        Mat trainingData = new Mat();
        Mat trainingImages = new Mat();
        Mat trainingLabels = new Mat();
        CvSVM clasificador;
        String path="C:\\java workspace\\ora\\images\\Color_Happy_jpg";
       for (File file : new File(path).listFiles()) {
            Mat img=new Mat();   
            Mat con = Highgui.imread(path+"\\"+file.getName(),Highgui.CV_LOAD_IMAGE_GRAYSCALE);
            con.convertTo(img, CvType.CV_32FC1,1.0/255.0);

                img.reshape(1, 1);
                trainingImages.push_back(img);
               trainingLabels.push_back(Mat.ones(new Size(1, 75), CvType.CV_32FC1));

            }
        System.out.println("divide");
        path="C:\\java workspace\\ora\\images\\Color_Sad_jpg";
          for (File file : new File(path).listFiles()) {
                Mat img=new Mat();
                Mat m=new Mat(new Size(640,480),CvType.CV_32FC1);
                Mat con = Highgui.imread(file.getAbsolutePath(),Highgui.CV_LOAD_IMAGE_GRAYSCALE);

                con.convertTo(img, CvType.CV_32FC1,1.0/255.0);
                img.reshape(1, 1);
                trainingImages.push_back(img);

                trainingLabels.push_back(Mat.zeros(new Size(1, 75), CvType.CV_32FC1));

              }

            trainingLabels.copyTo(classes);
            CvSVMParams params = new CvSVMParams();
            params.set_kernel_type(CvSVM.LINEAR);
            CvType.typeToString(trainingImages.type());
            CvSVM svm=new CvSVM();



            clasificador = new CvSVM(trainingImages,classes, new Mat(), new Mat(), params);

            clasificador.save("C:\\java workspace\\ora\\images\\svm.xml");
            Mat out=new Mat();

            clasificador.load("C:\\java workspace\\ora\\images\\svm.xml");
            Mat sample=Highgui.imread("C:\\java workspace\\ora\\images\\Color_Sad_jpg\\EMBfemale20-2happy.jpg",Highgui.CV_LOAD_IMAGE_GRAYSCALE);

           sample.convertTo(out, CvType.CV_32FC1,1.0/255.0);               
            out.reshape(1, 75);
            System.out.println(clasificador.predict(out));

1 个答案:

答案 0 :(得分:1)

1

你的火车标签仍然是错误的。

你需要一个带有numrows == numimages和1 col的浮动垫。所以,每张图片标签为1个。

所以你的悲伤面孔应该有:

trainingLabels.push_back(-1.0);

你的快乐应该有:

trainingLabels.push_back(1.0);

2

必须以与训练相同的方式处理预测样本。

sample.convertTo(out, CvType.CV_32FC1,1.0/255.0);               
out.reshape(1, 1);