How is everyone implementing Deepscan for BI 5.4.x

General discussion about Blue Iris
Post Reply
roadkill401
Posts: 16
Joined: Fri May 28, 2021 5:16 pm

How is everyone implementing Deepscan for BI 5.4.x

Post by roadkill401 »

I don't have a major server to run Blue Iris but want to get this setup correctly so I can walk away and just use it, rather than constantly having to come back and tinker.

I have an intel i5-6600 right now with 8gb ram, I have physically 2 camera's with a 3rd on its way. It is currently setup to run as a service and I use RDP to log into the unit if needed, but I find I can do most everything through the web interface. It shows cpu at 20% and ram at 33% so it should be able to handle a few more cameras if I want.

While asking. what build of windows 10 does everyone have their system running on. I originally got my PC with 21H1 but had serious problems with windows. So I reformatted with build 1909 and it seems to be solid. I am reformatting the computer to add in an SSD for the windows10 OS. (I found a power splitter so I can now run two drives). Windows shows even with 1909 that it's up to date. I was thinking of simply turning off windows update service and let it sit at 1909. What is the best version of Windows 10 to park your Blue Iris server at?

I am wondering how others are implementing DeepScan to properly detect people. I don't need cars or bears. Cats might be fun but not a necessity. I have an old Synology DS412+ that can run docker but its just a really slow Intel Atom D processor, so I gather that DeepScan won't run well on it. I have read that there is a build to run on a $100 Raspberry Pi4 8gb model.
Last edited by roadkill401 on Fri Jul 02, 2021 6:17 pm, edited 1 time in total.
SolarEclipse
Posts: 15
Joined: Fri Jun 28, 2019 2:31 pm

Re: How is everyone implementing Deepscan for BI 5.4.x

Post by SolarEclipse »

Just my 2 cents . . .

I'm running my 3 camera setup in a Virtual Machine on a i7-3770 with 5 GB of RAM dedicated to it.

I have implemented DeepScan and it sees about 75% of the vehicles that drive by and identifies them as such. People walking by is only about 25%. It completely fails at face identification.

So far I'm just leaving it going as I hope DeepScan will eventually be improved and get better results but I don't rely only on it for alerts. My max resolution on a camera is 1440 so it could be that it just doesn't get enough info to make a firm determination.

I do work from home, at least until fall, so it hasn't been an issue. I've also installed an alarm system so my guess for me I'll be relying on it more than any alerts from BlueIris going forward. And for a brief time today I got in my head maybe I'll just get rid of my cameras as I have to keep a computer on 24/7 and just managing it takes time I'd rather spend doing other things. But I'll save that decision for another day.
Post Reply