Real-time data storage using d455 depthcam.
I want to get a depth value of 15 fps.
However, the amount of data is vast and very slow. Currently, 15 fps has been set based on the height and width of 640 * 480, and the value is to be stored using Jetson Nano. I hope you can tell me both csv and MongoDB.
And, I'm using pyrealsense with Python3
Thank you.
-
If you are saving to .csv format then that is known to be slow, as mentioned by a RealSense team member in the link below. Writing a bag format file is the most efficient RealSense recording method, though bag files can be gigabytes in size. Bag files - which are like a video recording of camera data - can be recorded using pyrealsense2 scripting.
https://github.com/IntelRealSense/librealsense/issues/1485#issuecomment-380728118
It is possible though to save every frame as an individual bag file using an RealSense SDK instruction called save_single_frameset()
Frames can also be stored in memory during recording with the Keep() instruction instead of continuously recording to file and then have batch-processing actions performed on all the frames when the recording is closed, such as alignment, post-processing and saving to file. This method is suited to short recording durations though because it progressively consumes the computer's memory capacity. On a Nano, 10 seconds would be a reasonable recording duration, whilst higher memory capacity computers could record to memory for 30 seconds.
Please sign in to leave a comment.
Comments
1 comment