Recording Video efficiently with a Jetson Orin Nano Dev Kit. Compression issue.

Software
1

Hello,

I am using a ZED X Mini camera with a Jetson Orin Nano Developer Kit. I have noticed in community discussions that I am not the first person to encounter this issue. I understand that because the Jetson Orin Nano lacks hardware GPU encoders (NVENC), recording is restricted to H.264 Lossless or Lossless formats which are CPU heavy.

Unfortunately, this results in a high number of dropped frames because the CPU cannot keep up with the data throughput (I am testing at HD1200 at 30 fps).

I was wondering if you have a solution for this. For example, I would be interested in trying raw video acquisition since I am using an SSD, but this is currently not available as a “Compression Mode” option.

Additionally, since I only require the video feed from one “eye” of the ZED X Mini, are there any options to record in monocular mode only? Along with the frames and IMU measurements, I am also embedding data from a 1D LiDAR range sensor into the SVO2 files.

Thanks in advance for your input.

Best regards,

Simone

2

Hi @Simone,

Thanks for reaching out.
That is correct, the Orin Nano does not have hardware encoders available, which limits recording and streaming capabilities to be CPU-bound.

Starting in ZED SDK 5.2, we have added a software encoder for H264 lossy recording that should activate by default, and prints a message in the verbose logs showing that it is activated. This can help reduce the number of dropped frames.

We currently don’t have a “one camera only” recording mode, as the stereo format allows to replay all of the ZED SDK algorithms out of the box.