Description
Hello there, I’d like to get inference on frames from AWS Rekognition whenever a person is detected. I have a deepstream graph that gets RTSP source, gets analytics using object detection and tracker, and sends the analytics metadata to a message broker. I’ve thought of adding a probe that saves the image for later boto3. Or sending the frame itself over kafka. But I’m not sure if any of both is applicable.
Can you recommend an approach?
Thanks in advance