I have pgie, nvtracker, sgie.
In pgie and sgie I do some postprocessing.
How do I better make my pipeline: pgie->nvtracker->sgie or pgie->sgie->nvtracker?
Where should I specify bInferDone=TRUE: in pgie or in sgie and why?
1 please refer deepstream sample deepstream-test2.
2 you can add callback fucntion, like parse-bbox-func-name, or you can process output-tensor-meta, please refer to deepstream sample deepstream-infer-tensor-meta-test.
If I have pgie and a lot of sgie, I should specify bInferDone=TRUE after all sgie outputs tensor postprocessing or immediately after pgie postprocessing?
/** Holds a Boolean indicating whether inference is performed on the frame. */
gboolean bInferDone;
As the comment shown, it is used for pgie, please refer to attach_metadata_detector,
1 Like
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.