Hi,
We have a deployment scenario in which we want to define an ROI on which our inferencing should run.
This ROI is not the same for all camera sources.
Is there any way we can define ROI in the source, which can be applied just after decoding of the stream and only the ROI(cropped and scaled) regions are used in the rest of the pipeline?
Regards
Pallab Sarkar
ROI based inferencing is not supported at the moment. However, u can filter out the detected objects outside ROI from metadata
Hi ,
Thanks for your response. Currently we are filtering the objects outside ROI using metadata.
But as we want to optimize it further, so we want to filter out the region outside ROI just after decoding.
Do you have any plans of adding this feature in future?
Regards