Have you switched from DeepStack to CodeProject.AI? Poll

Have you switched from DeepStack to CodeProject.AI?

  • I am not using any AI and never will

    Votes: 0 0.0%

  • Total voters
    209

MikeLud1

IPCT Contributor
Joined
Apr 5, 2017
Messages
2,141
Reaction score
4,118
Location
Brooklyn, NY
I am creating this poll to help users decide if they should switch from DeepStack to CodeProject.AI.

Please post your likes and dislikes, your posts will help improve CodeProject.AI. Also you can go to their discussing board if you find any bugs or have a feature request.

Link to CodeProject.AI

Link to discussing board
 

Futaba

Pulling my weight
Joined
Nov 13, 2015
Messages
220
Reaction score
153
I am still using deep stack because it is still working fine, with your custom models and your license plate recognition tool and models. I plan to move to CodeProject.AI but want to make sure it is a complete replacement of deep stack and has all the features I am using. Since deep stack is fully working except when it pukes due to too many orphaned temp files. And that all 8GB of video memory is fully utilized when deep stack is running.

I would switch over if BI complete removes deep stack integration or there is an easy way to move to CodeProject.AI and back to deep stack if needed. I am sure there is but I haven't spent time to investigate yet.

I guess my answer is I am in a "wait and see" mode.
 

tech101

Known around here
Joined
Mar 30, 2015
Messages
1,472
Reaction score
2,125
Location
SF BayArea, USA
I will definitely be moving from DS to CodeProject in the near future. Just waiting for the implementation or switch to be a little easier will wait maybe until the 2.0 :)
 

kklee

Pulling my weight
Joined
May 9, 2020
Messages
187
Reaction score
201
Location
Vancouver, BC
I switched over from DS to CodeProject a few months ago because people were reporting faster detection times. The switchover was straight forward, minor tweaking was required in BI for the different custom model names. CodeProject reduced my detection times by 50% and seems to be more accurate.

Edit: Forgot to mention that it's running on a Quadro P400.
 
Last edited:

wittaj

IPCT Contributor
Joined
Apr 28, 2019
Messages
24,436
Reaction score
47,561
Location
USA
For me, Deepstack just plain works. But I also have spent a lot of time working on my field of views and making sure I am not doing to much with one field of view. I average under 150ms which is totally fine for my use case.

Sure Deepstack had some issues and I was on the leading edge of trying it out for our members by literally downloading that latest update the moment it became available.

But we never had to go and do json edits and edits to the registry as folks have had to do with Code Project. The biggest issues with Deepstack were was it car,person or Car or was there a space after the comma or was it capitalized LOL.

There just seems to be way more issues for folks with Code Project than with Deepstack, although I acknowledge that many have implemented Code Project without any issues.. Sure we still had active posts of members Deepstack not working, but those were more user error in putting the files in the wrong directory or setting up motion detection wrong, but once they got that straightened out, they were good to go.

My concern is the whole flashy webpage and userface and feel and what not for SenseAI that leads me to believe this is an initiative those folks started with the hope that someone buys them out and then it becomes a paid service. Kind of like what OpenALPR did and now we have to pay $8 per camera per month when Rekor bought it. Deepstack on the other hand has been around awhile and really looked like a grassroots hobby community and as such would be less likely to be attractive to a company wanting to get into AI. I might be wrong and hopefully I am wrong, but it just has that feel of being a startup hoping to get bought out by an Apple or Google or some other entity looking to get into AI. Maybe Rekor will buy it LOL.

Now with that said, I suspect at some point Deepstack will break with BI and we will be forced to either go with CodeProject or whatever the flavor is at that time or go without the integrated AI.

I will probably take a day at some point after the ALPR rolls out and make the switch as that would be the only driver for me in wanting to switch.
 

spuls

Getting the hang of it
Joined
May 16, 2020
Messages
89
Reaction score
68
Location
at
Since there are already multiple API wrapper for Tensorflow, it would be also interessting how many users already moved to solutions with hardware support for Coral or non-CUDA solutions?
 

Fubduck

Getting the hang of it
Joined
Jul 10, 2018
Messages
109
Reaction score
72
Location
colorado
I'm sticking with DeepStack. If CodeProject.AI ever supports the nVidia Jetson Nano I'd consider giving it a shot.
 

MikeLud1

IPCT Contributor
Joined
Apr 5, 2017
Messages
2,141
Reaction score
4,118
Location
Brooklyn, NY
Since there are already multiple API wrapper for Tensorflow, it would be also interessting how many users already moved to solutions with hardware support for Coral or non-CUDA solutions?
I added the below response

1670092543890.png
 

kaltertod

Getting the hang of it
Joined
Jul 30, 2022
Messages
65
Reaction score
46
Location
BFE
I am happy with codeproject ai other than the facial recognition and a few hiccups with delivery on it.... but I think as a whole the hiccups that I am experiencing are due to some camera changes as well.
 

CCTVCam

Known around here
Joined
Sep 25, 2017
Messages
2,660
Reaction score
3,480
I'm waiting to put my camera up although that's due to other diy projects physically in the way of installation.

Putting that aside, the 2 factors stopping me from using CodeProject AI atm are:

1. Lack of support for Coral TPU - this to me is a ground breaking piece of kit, especially in these expensive energy times - several times the performace of high cuda count cards for only 2 watts of power, saving literally tens if not hundreds of watts per hour!!! The issue here is the lack of a 1 click installer that can download and configure everything required. I curently understand the Coral TPU can be used but it requires mulitple pieces of additional software and some code programming skills to make it work. I think for most, a simple "click to install support for Coral's TPU" is what is needed.

2. A custom model (I haven't seen one so stand to be corrected) for objects for Codeproject. Not quite sure where the developers are going with this - maybe the standard object AI list is for general AI developers and not CCTV usage, but objects such as girafe, banana, cup, Zebra etc etc are of little use to most people unless trying to use an internal camera to monitor which objects were moved or stolen or for zookeepers to see which animals were stolen. For most, with external cameras, little more is needed other than person, and basic vehicles eg car, truck, motorbike, bicycle, or maybe a few animals to add for those who wildlife spot. But for most of us, the basic 5 objects are probably all that are needed - some might argue knife or gun, but these would be hard to identify in a hand anyway as partially obscured by being held. Beyond those basic 5 objects, all the others are doing is using wasting power and cpu time trying to decide if it;s a person or a Zebra stealing your lawn mower. Needless to say, the police aren't too interested when you report a thieving zebra! They may send assistance through other agencies with persons with white coats though if you report too many Zebras or Giraffe making off with your garden tools or Gnomes.
 

actran

Getting comfortable
Joined
May 8, 2016
Messages
784
Reaction score
697
@CCTVCam

#2: CodeProject.AI includes @MikeLud1 custom models for IPcam use cases. In fact, he's part of the CodeProject.AI team. So your #2 is a non issue. And adding additional custom models like the delivery truck detection is dead simple by dropping them into the expected folder.

As for #1, if you require Coral support, you're probably not going to find it with Deepstack either (especially if you need a 1-click installer).

I would encourage you to give CodeProject.AI a try. Their one-click Windows install is pretty solid and will run fine on CPU, especially if you have an i7 or higher.

Side note, there is no need for regedits, that's old news and was only required for earlier iterations.
 
Last edited:

spuls

Getting the hang of it
Joined
May 16, 2020
Messages
89
Reaction score
68
Location
at
BI asked my in my last deployment if it should install Codeproject AI, i guess they already build in a "1-click" installer?

There are multiple Tensorflow based projects with coral support, they provide the same api as deepstack and can be used in BI without a problem. But all of them need quite some knowledge to implement.
 

fenderman

Staff member
Joined
Mar 9, 2014
Messages
36,897
Reaction score
21,250
BI asked my in my last deployment if it should install Codeproject AI, i guess they already build in a "1-click" installer?

There are multiple Tensorflow based projects with coral support, they provide the same api as deepstack and can be used in BI without a problem. But all of them need quite some knowledge to implement.
That is useless to 99.99999 percent of blue iris users. That is the beauty of codeproject, easy to install and there is active support from the developers to the blue iris developer. Codeproject has already indicated that they will likely work on Coral support.
 
Top