I have a pipeline that is streaming a flir camera over rtsp like this.
/usr/bin/gst-launch-1.0 aravissrc ! video/x-raw,format=RGB ! videoconvert ! video/x-raw,format=I420 ! x264enc speed-preset=ultrafast ! rtspclientsink protocols=tcp location=rtsp://localhost:8554/flir
I am running human detection experiments on this and normally it was working okay. Recently it started giving no detections at night times so I checked the frames. They were almost all dark. I manually increased the exposure and the gain like this
/usr/bin/gst-launch-1.0 aravissrc gain=25 exposure=100000 ...
and it returns the expected frames with this setting but this setting is not compatible to the daylight. So I decided to use AutoExposure like this
/usr/bin/gst-launch-1.0 aravissrc gain=25 features="BalanceWhiteAuto=Continuous ExposureAuto=Continuous" ...
but it does not return the frame with expected light. So I am out of options here except to write a service to dynamically changing exposure value based on time but I think it is the job of the camera software. What should I do here? Is there any other plugins or features with gstreamer that can fix my problem?
I am not able to find great documentation on aravissrc and its features.