I want to extract bounding box images of cars from the annotated video stream as they are inferred in the osd_sink_pad_buffer_probe function. I am currently using the deepstream-test2 for my main application. I need to know the relevant data structures or the approach to this end. Any help would be appreciated. Thank you.
Please check if https://docs.nvidia.com/metropolis/deepstream/python-api/NvDsMeta/pynvmeta.html has the informantion you want.
If there is not, please elaborate which data structure you want to know.
Found it (NvDsFrameMeta) in the imagedata multistream application. Thanks.