看图片，PP-OCRv5_mobile_rec_infer PP-OCRv5_mobile_det_infer 这两个模型在我的机器上跑出这性能正常吗，OCRv5_mobile_rec看上去跑了6s

[aijyo]

ocr pipeline 配置如下：我就是想用它来识别截图，但是看上去。。这性能不知道哪里出了问题

// --------- Init OCR pipeline ----------
OCRPipelineParams config;
{
    config.text_detection_model_dir = det_dir;
    config.text_detection_model_name = "PP-OCRv5_mobile_det";

    config.text_recognition_model_dir = rec_dir;
    config.text_recognition_model_name = "PP-OCRv5_mobile_rec";

    config.use_doc_orientation_classify = false;
    config.use_doc_unwarping = false;
    config.use_textline_orientation = false;

    config.device = "cpu";
    config.cpu_threads = 1;
    config.thread_num = 1;

    config.precision = "fp32";
    config.enable_mkldnn = false;
    config.mkldnn_cache_capacity = 10;

    config.lang = "eng";
    config.paddlex_config = R"(D:\code\gitcode\PaddleOCR\deploy\cpp_infer\wow\ocr.yaml)";
}