Here is a video of the ZED 2 running a Nvidia deepstream YOLO 3 model using the Video streams from the ZED 2 camera using gstreamer. There is also a link to a Github repo with the pre built gstreamer pipelines if you want to try it yourself:
Good job!
This is a really interesting application, thank you for sharing the repository with the ZED community 
When I run the folowing ZED 2 object detection model gstreamer pipeline: The results show a bounding box with the label of the object and it also displays the distance from the object to the camera. How do you extract that data?
That is my ultimate goal. To display the distance from the camera to object .Within the deepstream object recognition pipline results.
gst-launch-1.0 \
zedsrc stream-type=0 od-enabled=true od-detection-model=0 camera-resolution=2 camera-fps=30 ! queue ! \
zedodoverlay ! queue ! \
autovideoconvert ! fpsdisplaysinkThe depth information comes from the Object Detection module of the ZED SDK.
You are using a third-party detector, so in order to retrieve the depth information, you must create a new element plugin that gets a left/depth stream and the bounding box data as input and combines them all.
The zedodoverlay retrieves the information from the zed-meta-data:
Hi @joev,
Good job! would you please let me know how much delay you get with this solution?