command line and code get different result

[易鑫]

Hello,every one:

    I use the command line and C++ code to recognize the image text ,but get different result.The image file and traineddata file are in the attachments.

I used this command:

      tesseract test_20200617_1_1854.png stdout -l regular_rgb_layer

The result is -6X201,it is correct.

I also used this C++ code:

    tesseract::TessBaseAPI *tess_rgb_regular = new tesseract::TessBaseAPI();

    //  --oem 3   

    if (tess_rgb_regular->Init("./tessdata", "regular_rgb_layer", tesseract::OcrEngineMode::OEM_DEFAULT)) {

        std::cout << "OCRTesseract: Could not initialize tesseract." << std::endl;

        return EINITTESS;

    }

    tess_rgb_regular->SetVariable("save_best_choices", "T");

    image = cv::imread("test_20200617_1_1854.png");

    tess_rgb_regular->SetImage(image.data, image.cols, image.rows, image.step[1], image.step[0]);    

    tess_rgb_regular->SetSourceResolution(300);

    tess_rgb_regular->Recognize(0);

    string result = std::unique_ptr<char[]>(tessApi->GetUTF8Text()).get();

But the result is -6X20.

Does anyone know the reason,thanks in advabce.

[易鑫]

Does anyone knows,thanks.

在 2020年6月19日星期五 UTC+8下午5:27:55，易鑫写道：