Here’s my take on this. To clear thing up I think this is the biggest mis conception of these AI tools like CPAI. They think that because it’s marking something wrong it’s their fault. When in reality it’s just an engine. They are using YOLO models by Ultralytics. If you are using MikeLuds...
No. It can be used for GPU. MikeLud1 has mentioned himself.
Here is a link to a post where he states his RTX3060 is faster on .NET than 6.2. He shows it on the benchmark test that is built in.
https://ipcamtalk.com/threads/codeproject-ai-version-2-5.74304/post-878936
Yeap Im aware of the .NET saying without Cuda on its description, but if you follow MikeLud1, he always has been saying the same thing about how the .NET module works faster on his RTX 3090 I believe he has. I never got it to be faster, but previously I had the Quadro P1000 card in here and the...
So my Dell SFF has a GPU slot but it but up against the power supply and cannot fit a dual slot GPU in it. So there is an x4 slot next to it. I put the 3050 in there and so far it’s been working fine in that slot. Getting about 25ms times on v5 NET.
I think im answering my own question that 4060 is probably out given that I dont think they have a version that can be powered by PCI slot alone, I believe they all need minimum 115w power, so they would need a 6 or 8pin connector which basically throws that card out of contention for lower...
Looking at just tensor cores, looks like 3050 has 73 AI TOPS and 4060 has about 242 AI TOPS. So thats a decent bump. Cuda Cores wise we have a decent bump of 3072 Cuda cores for the 4060 and 2304 Cuda cores for the 3050 (6GB) cards or 2560 Cuda Cores for the 3050 (8GB) card. Cuda Compute is...
I’ll have to put the computer on a Kill a Watt meter to see. Right now it’s on a UPS which also has my PoE switch for the cams so I can only see total wattage.
I’m using the default YOLO v8 setup and its model set as Large. Avg times I see are about 125ms for my cams which are sending it 720p...
I was looking into potentially upgrading my P1000 to either the RTX 3050 ones that @MikeLud1 mentioned or the RTX 4060 you are using here. I also do want to say in the low power mode and PCI slot powered GPU only as the box Im using is a SFF Dell that doesnt have any 6 or 8pin GPU power plug...
I cannot get YOLOv8 to run. Cant even uninstall the module now. Ive been using v8 for awhile but this 2.9.3 just wont let it run. Had to go back to using YOLOv5.6.2. Now I cant even run it either, theres no restart/start/stop buttons anymore even after shift refreshing it.
@EMPIRETECANDY
Im going to replay all my outdoor Amcrests (best time now here). I got 9. I was thinking of the IPC-Color4K-T and/or the IPC-T58IR-SE. I seen the review and what folks say about the Color4K-T, i have a few areas that always have lights on that will help, but there are other...
@MikeLud1
Question on this. On the "Open Images Dataset V7" what portion should we download? This Dataset seems huge and semi confusing to me being totally new at this on what to get, im assuming it has to be the ones that are already annotated.
Is it possible to use the COCO dataset? I...
Alright updating this that somehow now it just started working. I just let it sit for about 15-20mins after it went into Lost Contact, and now its started and processing.
Initial thoughts are its not faster since its showing like 200-300ms where as the Yolov6.2 and 8 were only taking about...
Yea unfortunately that doesnt do much, this is happening on this box (which ive had BI/CPAI on it for a year now and keeps happening any version of CPAI). Then also got same issue on a brand new fresh install for another location. Brand new Win11 Pro, latest BI and CPAI 2.3.4 (at that time a...
It is 1.9.2. Should be there on the info portion of the .NET module I posted. But Ive seen this issue on every version of CPAI Ive used. Going back to like 2.x to latest. I just remember you mentioning that you got better performance on .NET module and wanted to try it, but for some reason I...
I still cant get the .NET YoloV5 module to run. Ive even done this on a fresh PC (another build) and always the same thing, where it starts, then ends up saying LOST CONTACT. Posted in Codeproject discussions but that didnt get anywhere unfortunately. Im wondering am I missing something like...
I would test both the YOLOv5 6.2 and the .NET one. See which one gives you better performance. Some state .NET gives better. For me it was opposite so I use the 6.2 one instead. I could never get the .NET one to properly use the GPU in a stable fashion so I stopped messing with it. I get...
I can’t remember off the top my head but I believe if you attempt to install 2.17 it will tell you must uninstall the old one manually first. So just go to add remove and uninstall the old one and then rerun the installer for the latest one.
You will be forced to uninstall it due to the change in how the installer works from 2.1x.
It will remove most things except potentially any custom files/models you may have added in those directories.
Authors say it’s not that easy due to the number of different platforms like arm64 x64 then is it Linux or docket or windows native etc. So it would then require like 15-20 different offline installers for each different flavor.
The way they have it now is looks for pieces it needs based on...
There’s a thread on codeproject forum. Basically I noticed that it wasn’t installing .NET and ASP.NET and because of that installer was skipping steps like creating the service for AI.
So install those two modules and repair the installer and should be good.
So using ipcam-general vs using default object detection went for almost 2days without an issue (not saying at 2days there was an issue, i just stopped it to try my next test). So that sort of rules out the install being wonky, so it looks good. Before as mentioned with default object...
What model you using? Tonight Im going to test something by using a different model than ipcam-general. Going to try to make it use the yolo5vl which is what i was using before fine on 1.6.8.
Ive seen the same issue. Constantly crashes on that. Im testing it now not using the default object detection (using ipcam-general instead now) and its been stable now for 20hrs where before it would barely every go past 8-12hrs. Mine, like yours, would show same error, multiple it would...
So far Im seeing somewhat better performance now, its not memory leaking like it was before. Only change Ive made so far was to stop using the yolov5l default object detection and go back to using ipcam-general, and so far its not been using wicked memory on python and spawning off multiples...