View my account

Best way to stream D435 over a network

Comments

4 comments

  • MartyG

    An Intel RealSense employee developed a project called EtherSense for using cameras with Ethernet.

    https://github.com/krejov100/EtherSense

    The CONIX Research Center have also done an advanced multi-camera Ethernet system.

    https://github.com/conix-center/pointcloud_stitching

    0
    Comment actions Permalink
  • Evan

    Hi Marty,

    Thank you so much for your response.

    My initial post may have been confusing -- I meant to stream remotely to the cloud.  We are trying to handle all our compute/model evaluation on a server at a different location than on-site.

    The above projects seem like they are great at streaming realsense data over a local network, but I don't think they handle encoding/decoding to help send less data over a network.  We'd love to be able to use some lossy codec and stream data over a network, as our connection is limited 18 Mbs.

    Thanks!

    0
    Comment actions Permalink
  • MartyG

    For doing vision computing in the cloud, I would think that Microsoft Azure would be the primary cloud computing platform to consider.  Intel themselves recommend Azure cloud services on their cloud info page, so I don't think I can get into trouble for suggesting Microsoft in this case.   :)

    In regard to how to get the data from a RealSense camera to an Azure cloud, I would think that it should be straightforward to transfer the data from local storage into the cloud once it has been captured and compressed.  

    A couple of years ago there was a discussion about applying compression to RGB-D data that may provide some useful guidance.

    https://forums.intel.com/s/question/0D50P0000490DvWSAU/compression-formats-for-rgbd-streaming

    A promising avenue to investigate would be streaming to the cloud with ROS, like in an example that does so with ROS and Amazon.  I recommend googling for 'ros stream to cloud' for further research leads.  RealSense cameras are fully compatible with ROS via the RealSense SDK's ROS support 'wrapper'.

    https://github.com/IntelRealSense/realsense-ros/releases

    *********

    I hope that the above information gives you a good starting point for further exploration.

    0
    Comment actions Permalink
  • MartyG

    BTW, I offer my apologies for the delayed response.   I spent a while considering your case very carefully, and then once I posted it yesterday, I found that hours later it was still in an awaiting moderation mode that was keeping it invisible.  So I edited links out of it and re-posted it to get past the moderation filter!

    0
    Comment actions Permalink

Please sign in to leave a comment.