PPYOLOER使用C++推理（fastdeploy-win-x64-gpu-1.0.7）获取旋转目标数量为0

auto option = fastdeploy::RuntimeOption();
option.UseGpu();
auto model = fastdeploy::vision::detection::PPYOLOER(model_file, params_file,
config_file, option);
if (!model.Initialized()) {
std::cerr << "Failed to initialize." << std::endl;
return;
}

const cv::Mat im = cv::imread(image_file);

fastdeploy::vision::DetectionResult res;
if (!model.Predict(im, &res)) {
    std::cerr << "Failed to predict." << std::endl;
    return;
}

使用事例中的网络和图片，获取res.rotated_boxes.size() =0
请帮忙分析一下是哪里问题？

Start by reproducing the shown C++ PPYOLOER example with the referenced network and image, then inspect the model initialization, Predict result, and res.rotated_boxes.size(). The issue provides no file paths, error output, or expected detection count, so completion requires identifying the cause of the empty result and confirming the behavior with the same example.