biggen
Known around here
- May 6, 2018
- 2,834
- 3,248
Is that a 24GB drive? Surely not and this must be RAM. How old is the hard drive though as that CPU is about a decade old now?
Last edited:
CodeProject.AI Version 2.0
- Thread starter MikeLud1
- Start date
Is that a 24GB drive? Surely not and this must be RAM. How old is the hard drive though as that CPU is about a decade old now?
I think that was it! There were lots of CP.AI temp files in this folder: users\username\APP Data\Local\temp. Maintenance and CCleaner didn't find them! I also found a ".Docker" folder that must have been left after uninstalling CP.AI. I deleted both of those and that may have done the trick! Thanks so much!
I think I'll need to do that. Which version of BI are you using with DS?
The 24GB is indeed RAM! Yes, the CPU is probably at least 10 years old. It's an old Dell machine but a few years ago I added an SSD for the OS and BI, along with a WD Purple hard drive for recording. After adding the two hard drives it's actually been running great....until now!
My grandson still has a 1st Gen PC I gave him several years ago. I also have a 2nd gen. for our HTPC, rarely use it much anymore but do keep Win10 updated on it.
So, if we want to use our old CPUs with CP.AI, could we just add a graphics cards such as the one David L suggested earlier (MSI NVIDIA GeForce GTX 1650 4G)? Or at this point, would it make more sense to upgrade to a PC with a newer CPU? I know the WIKI discusses CPU requirements for BI, but wondering what they would be for BI with CP.AI? Is the graphics card the key here? And, is CUDA necessary (it Iooks a little intimidating to me!) ?
My thought is with a GPU doing my AI, it is kind a like having two Processors, CPU for BI and GPU for AI...prior to AI my 4th gen CPU was handling BI but I don't have near the CAMs you have. I am recording 24/7 on 4 CAMs though and getting motion alerts...
If you do get a GPU card I would get at least 4Gigs memory, just my Opinion. I know in Gaming Memory on a Video Card is important so thinking AI would benefit too...
Last edited:
That's how I get by with my old i7, using my gaming pc hand downs so popped a 1060GTX into this BI PC. Have then undervolted and underclocked both the CPU\GPU and got the power usage down a large amount (the PC reports only 15w for CPU use and 9w for GPU while everything is running, reported from using CPUID HWMonitor, wouldn't say this is likely 100% accurate but compared to previous in stock mode readings it was more like 40w CPU and 60w GPU). Getting the upgrade itch again now on my gaming pc so the BI PC may receive a full update soon, once I convince the boss it's really needed lol
Just don't tell her it is a gaming PC, it is for buying shoes online
CodeProject.AI Version 2.1 coming soon. I am test this version right now. I do not know all the changes right now.
Changes
Install does not install modules, they get installed after running the install.
The ability to uninstall default modules
Update to ALPR (Version 2.2) fixes auto plate rotation, improves plate reading.
New ObjectDetection (TF-Lite) module, I thinks this is for Coral support, I could not test since I do not have a Coral
More Modules, list in the below screenshot
Changes
Install does not install modules, they get installed after running the install.
The ability to uninstall default modules
Update to ALPR (Version 2.2) fixes auto plate rotation, improves plate reading.
New ObjectDetection (TF-Lite) module, I thinks this is for Coral support, I could not test since I do not have a Coral
More Modules, list in the below screenshot
Well W11 gives me this, never use to.
But on CPU mode it did install fine and seems to run fine as always.
But on CPU mode it did install fine and seems to run fine as always.
Last edited:
Never did it before though and I have installed every version of AI since it was released.
CodeProject.AI version 2.1 changes
Release notes for 2.1
Release notes for 2.1
- All modules are now equal. On first startup we install Object Detection (Python and .NET) and Face Processing, but
these can be uninstalled. The Server installer itself no longer installs these modules: they are auto-installed on
first startup by the server itself. - Reworking of the Python module SDK. Modules are new child classes, not aggregators of our module runner.
- Reworking of the modulesettings file to make it simpler and have less replication
- Improved logging: quantity, quality, filtering and better information
- Addition of 2 modules: ObjectDetectionTFLite for Object Detection on a on Raspberry Pi using Coral, and Cartoonise for some fun
- Improvements to half-precision support checks on CUDA cards
- Modules are now versioned and our module registry will now only show modules that fit your current server version.
- Various bug fixes
- Shared Python runtimes now in /runtimes. No more AnalysisLayer folder. These are all now in /modules
- Tested on CUDA 12
ohh nice cant wait to try it when released....thanks for all the great work yall do
Feature Request?
@MikeLud1
My understanding is that the “+ real-time images: xx” setting will cause more than just the initial image to be analyzed, but that the analysis will stop after the first positive confirmation.
Is there a way to add an option to have the analysis continue to run on the remaining images as specified in "+ real-time images: xx" and "analyze one each: xx" sometimes even when one confirm happens? A second analysis may provide a better result, especially for ALPR where it doesn’t get the number correct on the first try, and also face rec I assume. I realize this will cause duplicate confirmations of the same image in most cases, but for those with the processing power it might be nice to have the additional option.
Or am I misunderstanding how it works?
@MikeLud1
My understanding is that the “+ real-time images: xx” setting will cause more than just the initial image to be analyzed, but that the analysis will stop after the first positive confirmation.
Is there a way to add an option to have the analysis continue to run on the remaining images as specified in "+ real-time images: xx" and "analyze one each: xx" sometimes even when one confirm happens? A second analysis may provide a better result, especially for ALPR where it doesn’t get the number correct on the first try, and also face rec I assume. I realize this will cause duplicate confirmations of the same image in most cases, but for those with the processing power it might be nice to have the additional option.
Or am I misunderstanding how it works?
