It’s difficult to extract metadata from qtimlvdetection, so I’m planning to open the GStreamer pipeline with OpenCV and handle the post-processing using OpenCV. Is this possible?
Theoretically feasible. It is recommended that the customer integrate the OpenCV processing within the qtimlvdetection plugin.
I successfully got OpenCV working, and object detection is also functioning within OpenCV.
However, there’s still an issue where the OpenCV window only displays a black screen.
To address this, I managed to display both the Wayland display and the OpenCV window at the same time.
Specifically, the camera feed shows up correctly on the Wayland display,
while OpenCV shows only a black screen but still successfully renders the detection bounding boxes.
So for now, I’m treating Wayland as the visual display and using OpenCV for processing.
I think I need to look deeper into the remaining black screen issue, but overall, it’s a good working setup.
To display using OpenCV, the relevant Wayland interfaces need to be utilized for creation.
Alternatively, the information output by OpenCV, such as bounding boxes, can be drawn on the corresponding buffers in qtimlvdetection.
thank you for anwser
i’ll try again and share the result
Okay,we are looking forward to your good news. 