I am a long time BI user that hasn't done much customization but since the current PC is flaking out I built a new Franken-box with 64G RAM, an Intel 14th Gen processor (28 cpus) 1x1TB and 1x4TB M.2 SSD's and a Gigabyte 3060 video card, and thought I might explore AI. Yes, I know the hardware is way overkill but I want it to last a while and I will probably multi-purpose the box. Large/long term storage is not a problem as I have a spare 12TB SATA drive in the "old BI box" as well as a ridiculous amount of NAS storage. I have 18 various vendors indoor and outdoor POE cameras from 2-4MP each, recording at default frame rates.
Since it is pretty beefy, I'm going to start this new build using Ubuntu Server with QEMU/Libvirt virtualization so I can leverage left over cycles and memory for other VM's. For the record I am a recovering Unix sysadmin (+40 years) and I really did put in a lot of effort into running some of the linux flavors of NVR/camera management (Shinobi, Zoneminder, BlueCherry, etc.) but it all comes down to this: BI is easier to configure, easier to use, more mature, more robust, has a great UI, and has better support. While I stumbled through and constantly tinkered with and googled for solutions to linux NVR problems that don't exist in BI, and I crashed and rebuilt the linux variants, the BI 5 Windows box kept running (BI is my only Windows instance in the house).
So I just laid down a fresh Windows 10 VM and thought I would ask this forum a few questions before I get in too deep.
First, my biggest recording issue has always been trees and shadows. I have no problem recording cars, people, and critters, but default motion sensing was always triggered by trees and shadows since I never really tuned well.
I do use dual streams for all cameras and since I have a lot of storage I was considering using continuous recording and just forgoing motion detection. What would be the downside to that?
Second, now that I have a pretty robust video card I was also considering AI to help with the shadows. I am still reading all the posts I can find on Deepstack and CodeProject but everyone seems to have a different opinion of each and I really haven't found two articles that agree with each other yet.
Given that I have all this hardware do I really need to immerse myself in AI?
If I don't need AI, do I need the graphics card at all (there is on-board Intel graphics)?
Hope you enjoyed the read and please do post your ideas on how you would move forward with this scenario. Thanks very much.
Since it is pretty beefy, I'm going to start this new build using Ubuntu Server with QEMU/Libvirt virtualization so I can leverage left over cycles and memory for other VM's. For the record I am a recovering Unix sysadmin (+40 years) and I really did put in a lot of effort into running some of the linux flavors of NVR/camera management (Shinobi, Zoneminder, BlueCherry, etc.) but it all comes down to this: BI is easier to configure, easier to use, more mature, more robust, has a great UI, and has better support. While I stumbled through and constantly tinkered with and googled for solutions to linux NVR problems that don't exist in BI, and I crashed and rebuilt the linux variants, the BI 5 Windows box kept running (BI is my only Windows instance in the house).
So I just laid down a fresh Windows 10 VM and thought I would ask this forum a few questions before I get in too deep.
First, my biggest recording issue has always been trees and shadows. I have no problem recording cars, people, and critters, but default motion sensing was always triggered by trees and shadows since I never really tuned well.
I do use dual streams for all cameras and since I have a lot of storage I was considering using continuous recording and just forgoing motion detection. What would be the downside to that?
Second, now that I have a pretty robust video card I was also considering AI to help with the shadows. I am still reading all the posts I can find on Deepstack and CodeProject but everyone seems to have a different opinion of each and I really haven't found two articles that agree with each other yet.
Given that I have all this hardware do I really need to immerse myself in AI?
If I don't need AI, do I need the graphics card at all (there is on-board Intel graphics)?
Hope you enjoyed the read and please do post your ideas on how you would move forward with this scenario. Thanks very much.
