Function my_reco

sample/cpp/main.cpp:184–234

@ MaaCustomRecognitionCallback

// @ MaaCustomRecognitionCallback
MaaBool my_reco(
    MaaContext* context,
    MaaTaskId task_id,
    const char* node_name,
    const char* custom_recognition_name,
    const char* custom_recognition_param,
    const MaaImageBuffer* image,
    const MaaRect* roi,
    void* trans_arg,
    /* out */ MaaRect* out_box,
    /* out */ MaaStringBuffer* out_detail)
{
    /* Get image */

    // Approach 1
    uint8_t* png_data = MaaImageBufferGetEncoded(image);
    size_t png_size = MaaImageBufferGetEncodedSize(image);
    // cv::Mat im = cv::imdecode({ png_data, png_size }, cv::IMREAD_COLOR);

    // Approach 2
    // This approach is more efficient, but may be difficult for some languages.
    // I recommend you to use approach 2 if you can.
    void* raw_data = MaaImageBufferGetRawData(image);
    int32_t width = MaaImageBufferWidth(image);
    int32_t height = MaaImageBufferHeight(image);
    int32_t type = MaaImageBufferType(image);
    // cv::Mat im(height, width, type, raw_data);

    // And do your computer vision...
    // Or you MaaContext API to run some recognition
    MaaRecoId reco_id = MaaContextRunRecognition(context, "MySecondReco", "{}", image);
    auto tasker_handle = MaaContextGetTasker(context);
    // MaaTaskerGetRecognitionDetail(tasker_handle, reco_id, /* ... */);

    /* Output recognition result */

    // Step 1: output box
    std::array<int, 4> my_box { 0 }; // your result
    MaaRectSet(out_box, my_box[0], my_box[1], my_box[2], my_box[3]);

    // Step 2: output anything you want
    MaaStringBufferSet(
        out_detail,
        "Balabala, this string will be used by MaaCustomActionCallback and "
        "MaaQueryRecognitionDetail. "
        "And for compatibility, I recommend you use json.");

    // Finally, if this task is hit and you want to execute the action and next of this task,
    // don't forget to return true!
    return true;
}