I would like to connect 4 cameras GigE of this type: https://thinklucid.com/product/phoenix-16-mp-imx273/
The cameras are not going to record video continuosly, they would be recording at the same time during 10 seconds aproximately in intermittent periods. This cameras don’t have video compression onboard capabilities, so I estimate a rate of transmission about 300 MB/second/camera.
My idea is introduce a module where connect the four cameras, and deploy software on it to compress the video and then send it to a server where it could be processed.
My question is if I could use a Jetson Tx2 for this task.
I would have up to 7 Workstation with 4 cameras each one. The idea is having one TX2 for each Workstation, each one would be connected to a Switch and finally to a central server where videos will be stored.
In the roadmap of the Project there would be two phases of work:
Phase 1 -> In each Workstation, four cameras GigE like Phoenix Lucid or FLIR Blackfly (the last one are known to be working -> https://elinux.org/Jetson/Cameras ). The cameras would be connected to a NVIDIA Jetson TX2, where we could compress the video before we send it to the server.
Phase 2 -> In this phase, we would like to deploy software for an intelligent analysis of that videos, so we would have to deploy that software in each TX2 for each Workstation, so it involves to do repetitive tasks of deploy for each Workstation :-(
I would like like to expose this ideas with yours and know if there would be another approachs to consider or risks that I am not considering.