Memory configuration in Jetson Nano

Hi,

As I will be using Jetson nano to capture video as well to process the video, I am wondering how can we configure memory so that, once it reaches the max internal memory limit, it automatically falls back to storage for processing.

My question is, how can we configure the same? Also, can I use any USB drive for this purpose?

Hi,
We don’t have the mechanism like this. May see if there is existing implementation in Linux that you can leverage.

Not sure how you capture video. If you capture video through v4l2 , you can refer to

/usr/src/jetson_multimedia_api/samples/12_camera_v4l2_cuda

Most memory usage in running the application is to create NvBuffer for putting video frames.

Hi,

I am using below command to capture MP4 format.
gst-launch-1.0 nvarguscamerasrc ! ‘video/x-raw(memory:NVMM), width=(int)3264, height=(int)1848, format=(string)NV12, framerate=(fraction)28/1’ ! nvv4l2h265enc bitrate=8000000 ! h265parse ! qtmux ! filesink location=/media/D6D86ADBD86AB97F/Vidd/h265enc/3264X1848X28.mp4 -e

I can see that all the codes are in C++. Could you please provide example in python?

Hi,
For launching gstreamer pipline in python, you may refer to

Hi,

The link you mentioned helps to find the fix for TX2. I am using Nano. Currently, I am using below command,
gst-launch-1.0 nvarguscamerasrc ! ‘video/x-raw(memory:NVMM), width=(int)3264, height=(int)1848, format=(string)NV12, framerate=(fraction)28/1’ ! nvv4l2h265enc bitrate=8000000 ! h265parse ! qtmux ! filesink location=/media/D6D86ADBD86AB97F/Vidd/h265enc/3264X1848X28.mp4 -e

For recording the video for development purpose. Planning to have below python code to read frames and process the images.

With this python code, do you think resource utilisation be optimised for capturing videos. I also tried with 4 TB hard disk, and found that system was overloaded. Now, I am using 64GB pen drive, which seems to be working well for me. Please suggest the best approach to capture video/frames to have the good quality images as well optimising resource utilisation.

Hi,
For running OpenCV on Jetson platforms, need to have software conversion from BGRx to BGR. Please refer to explanation in the post:

We would suggest run pure gstreamer pipeline to pass NVMM buffers from source to sink. Please check examples in gstreamer user guide

1 Like