I am on jetson xavier nx with jetpack 5.0.2. I am using it on MODE_20W_6CORE power model.
I am using the following pipeline to save camera input in .avi format. I have excluded properties and caps of the camera src.
gst-launch-1.0 -e v4l2src ! "video/x-raw" ! queue ! videoconvert ! avimux ! filesink location=name.avi sync=false async=true
I have not used encoding because I want raw frames in the video.
My issue is : In the saved file, I am not getting the FPS that the camera provides. I have tried the same stream on my laptop, and I don’t get any frame drop there. I have tried this with different cameras as well and got the same result.
Hi,
The performance may be capped due to CPU capability. Please run sudo tegrastats and check if certain CPU cores are in 100% loading.
Ad we have Jetpack 4.6.4 and 5.1.2 for Xavier NX. You may consider upgrade to latest version and give it a try.
I cannot upgrade the jetpack due to other dependencies.
In sudo tegrastats, it shows it has used a CPU core at 88% at one instance. I don’t see 100% anywhere.
Update: I also tried the same stream on an orin with jetpack 5.1.2, and it gave me the same result.
Is there any other way? Perhaps some changes in the stream or something?
Hi,
Please encode to h264 or h265 to reduce the data size. The bottleneck looks to be in file IO. Would suggest encode the data to have less data size.
I have compared the feed with and without encoding and found that the encoded feed looses a lot of information. It is not ideal for my usecase.
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.