Web camera shut down after period without face


#1

We’re using affectiva to track facial expressions of subjects in out lab. We have a C# application running on windows 7 that writes data to a CSV file.

For a proportion of the machines (9 out of 25) the logging stopped long before the session ended. Some follow up tests appear to indicate that the issue is arising where the face is lost (say the subject is distracted and looks away) for longer than some period - which we believe about 5 minutes - whereupon the camera appears to shut down (light goes out) and no further information is tracked (there are no further call backs from the detector, even to say that there is no face). Is there a way to detect such a camera loss or shutown from the API?

My feeling is that the root problem is some setting on the problem machines that turns off the cameras, but would like to confirm that at the same time as fixing it.

thanks,

Shaun


#2

Hi, sorry for the delay in responding. I guess first I should ask if you’ve already resolved this issue, and if so, what the solution was (in case others encounter it).

If you’re still having this problem, can you tell us if you using the CameraDetector class or the FrameDetector? Are you able reliably reproduce the problem and confirm your theory that it is related to no face being found for a period of time after a face is lost?


#3

Hi,

thanks for the follow up. We’re using the CameraDetector, and I’ve tracked the root of the problem to the power saving setting on the PCs’ display. I found our lab machines have a default setting to power down the display after 5 minutes and it would appear that when the display powers down, so does the web cam. This was leading to the issue that we saw and it was tying to their being no face found as it was when our lab subjects were distracted from their PCs for extended periods. Extending the time before power down of the display appears to have removed the problem. Interestingly though - simply waking the display up did not appear to restart the CameraDetector callbacks.

regards,

Shaun


#4

Great, glad you solved it. Hopefully, your analysis may prove useful to others down the road.


#5