Configured the sensor to report every X-minutes, and it works. It appears the device will only wake up to take photo, run the selected algorithm and then report the results, at each regular reporting interval. Is this the only operation mode? Understood that continual image-taking will drain the battery; but it might be good to sample more often and report those results at each regular reporting interval.
How can people-counting function well if it is only a snapshot-count once every e.g. 20 minutes?
Is there an external input which could be used for a trigger to take an image and analyze?
Is there a firmware roadmap to share? How is the Audio AI sensor coming along?