5.5.8 - June 13, 2022 - Code Project’s SenseAI,

fenderman

Staff member
Joined
Mar 9, 2014
Messages
35,091
Reaction score
17,063
More formal support for Code Project’s SenseAI, now our preferred no-extra-cost AI
provider over DeepStack. SenseAI is better supported by its developers and has been found
to be more stable overall. At this point, if you are relying heavily upon custom AI models,
you may need to stick with DeepStack until guidance can be provided on how to add these
to SenseAI.
 

d5775927

Getting the hang of it
Joined
Dec 11, 2019
Messages
168
Reaction score
72
Location
here
More formal support for Code Project’s SenseAI, now our preferred no-extra-cost AI
provider over DeepStack. SenseAI is better supported by its developers and has been found
to be more stable overall. At this point, if you are relying heavily upon custom AI models,
you may need to stick with DeepStack until guidance can be provided on how to add these
to SenseAI.
Does it require a GPU or a strong CPU?
 
Joined
Dec 28, 2019
Messages
10,978
Reaction score
25,996
Location
New Jersey
i find custom models for DS to be an excellent addition. I don't expect to have many laptops approaching my house so AI that scans for every object in its database isn't the best option, at least to me. I haven't tried this one, yet, but sure hope it can be customized to just look for what I really want to find.
 

Lucas Tam

n3wb
Joined
Apr 12, 2017
Messages
20
Reaction score
5
Does it require a GPU or a strong CPU?
Looks like CPU only - with a 8 cores of a Ryzen 5950X under a Docker VM, I only get 3.5 operations per second. I'm not sure if that is enough for real world usage yet.

Also, I like the admin screen and API tester, but it seems to have a few graphic display bugs, the API availability switches between available/unavailable, but otherwise works.
 

BORIStheBLADE

Getting comfortable
Joined
Feb 14, 2016
Messages
559
Reaction score
1,692
Location
North Texas
When they say CPU only are they utilizing the Intel cpu that has the integrated gpu? I didn't see it on their page but might have missed it..
 

BORIStheBLADE

Getting comfortable
Joined
Feb 14, 2016
Messages
559
Reaction score
1,692
Location
North Texas
I think I'll stick with DeepStack until this is supported. Not that I have a ton of cameras but my set up is working pretty good for me right now. I will be keeping an eye on others experience with the new software for now.
 

wittaj

IPCT Contributor
Joined
Apr 28, 2019
Messages
13,469
Reaction score
25,136
Location
USA
Yeah DeepStack with the custom models work for my purposes, but I'm totally cool with BI looking to incorporate other methods as one may turn out to be better.

Now, when will he add @MikeLud1 plate reading customization to BI so that it is just a check box!!!!!
 

MikeLud1

IPCT Contributor
Joined
Apr 5, 2017
Messages
1,072
Reaction score
2,293
Location
Brooklyn, NY
I started play with Code Project’s SenseAI I downloaded and installed it on my other Blue Iris server that I use to remote connect to my main server. I posted in their discussion board to see if they will support DeepStack custom models

DeepStack with general V3.0 90ms

1655156652225.png

Code Project’s SenseAI 230ms

1655156725802.jpeg

1655158188925.png
1655158220441.png
 
Last edited:

Sybertiger

Known around here
Joined
Jun 30, 2018
Messages
1,929
Reaction score
4,176
Location
Orlando
Intrigued, but damn, I was just getting used to working with DeepStack, LOL.
Me too....when I zag y'all zig then when I zig y'all zag....I'll stay on DeepStack until y'all figure it out....LOL. I literally just put DS on my i7-6700 system about 15 minutes before the OP post. :facepalm:
 
Last edited:

jrbeddow

Getting comfortable
Joined
Oct 26, 2021
Messages
206
Reaction score
256
Location
SF Bay Area, CA
Anyone have this truly working yet on a live BI camera (not just a test of the API running)? I was initially getting a status message along the lines of "AI not found" on Alerts with the default setup (general AI tab set to the new localhost address of 127.0.0.1 and port 82). That address would not resolve when running the "Open in browser" test button, so I subsequently changed it to run on Port: 5000, and that one opens the new SenseAI API Dashboard, and does show the number of queries happening (I was up to 17 in the green "SenseAI Object Detection - Enabled" box). However, back in the Alerts list in BI, each one now shows up as "AI:404", so effectively still not resolving or unreachable host, right?

Any tricks to get this working?
 

MikeLud1

IPCT Contributor
Joined
Apr 5, 2017
Messages
1,072
Reaction score
2,293
Location
Brooklyn, NY
Anyone have this truly working yet on a live BI camera (not just a test of the API running)? I was initially getting a status message along the lines of "AI not found" on Alerts with the default setup (general AI tab set to the new localhost address of 127.0.0.1 and port 82). That address would not resolve when running the "Open in browser" test button, so I subsequently changed it to run on Port: 5000, and that one opens the new SenseAI API Dashboard, and does show the number of queries happening (I was up to 17 in the green "SenseAI Object Detection - Enabled" box). However, back in the Alerts list in BI, each one now shows up as "AI:404", so effectively still not resolving or unreachable host, right?

Any tricks to get this working?
You need to set in the general AI tab to the port to 5000
 
Top