CodeProject.AI Version 2.0

[David L]

yep seen those. Would cost 4 times that here in Uruguay and would get destroyed by the kids footballs in no time.
A 5 usd sprinkler activated by nodered suits my needs much better but thanks anyway.

I bet the sun would create false triggers for that kind of sensor also. CPAI would be (is) much more accurate imo

Yeah I had nothing to share so I threw that up , been getting their Ads. We just moved and all my CAMs are at the old house until we sell it. Figured I could retrieve SD-Card footage if needed. Sadly I was just getting all my AI setup to do most all I wanted.
I now live in the woods on a lonely County Road, not much going on there BUT I will be joining the Critter Captures in the months to come. This house has Zero Ethernet ran... We have seen a lot of Deer and Rabbits...yesterday morning a family of about 8 Deer was in our backyard. Pretty kewl...

 

[Daniel15]

You can probably do something similar with NodeRed automation.

I use Node-RED for my notifications and it works well.


The function node is:

I like it because you're not restrained in what you can do compared to configuring alerts in a UI. It's just JavaScript code so you can easily modify the code to do whatever you want.

 

[David L]

Home Assistant has been/was great on notifications for me...

 

[Pentagano]

I use Node-RED for my notifications and it works well.
View attachment 168080

The function node is:

I like it because you're not restrained in what you can do compared to configuring alerts in a UI. It's just JavaScript code so you can easily modify the code to do whatever you want.

Me too, have various flows that do a lot of things.
Also telegram and pushover pushes.
Use 1 node for an IFTTT job that turns up my volume to max for emergency notifications

 

[Pentagano]

My present understanding is that the Coral accelerator is unable to process medium and large models theirby reducing recognition accuracy. 110ms is an excellent response time for the Coral. Can you clarify if the 30ms and 110ms response times were using the same model size.

It is indeed a fact that software is limited in its capability to accurately determine power consumption. I am curious my self how much power is actually being consumed by the 1060 GPU and have decided to measure my dedicated system with and without the card. I will post the results tomorrow if I find the time.

As Frigate uses 320x320 for analysis I assume this is why the coral is so fast if I'm not mistaken.
Have you tried cropping with such small jpeg sizes from your substream with CPAI and coral but with larger model sizes?

 

[MikeLud1]

Below are the different resolutions for Large, Medium, Small, and Tiny

 

[spammenotinoz]

As Frigate uses 320x320 for analysis I assume this is why the coral is so fast if I'm not mistaken.
Have you tried cropping with such small jpeg sizes from your substream with CPAI and coral but with larger model sizes?

I am using Windows 11 with BlueIris and CodeProject AII, using the default model size of MEDIUM getting 110ms response times (inference times are much lower, but that is round-trip, so most of the latency is USB).
My Coral is connected via USB 3.1 (although it has a USB-C connector, response times were a lot slower when connected via USB-C).

Anyone have the PCIe version that connects to the WiFi M.2 slot? I expect latency will be even lower....

 

[105437]

What I'm seeing with the critters model is that it likes to identify flying insects at night as four-legged animals. In this case a dog.

 

[Pentagano]

What I'm seeing with the critters model is that it likes to identify flying insects at night as four-legged animals. In this case a dog.
View attachment 168130

Maybe try increasing the confidence and creating a zone closer. Do you need to see critters so far away?
Mine works well as it's in colour with lights on outside

 

[105437]

Maybe try increasing the confidence and creating a zone closer. Do you need to see critters so far away?
Mine works well as it's in colour with lights on outside

I agree about the confidence percentage. Still working on that. All of that grass area where the stepping stones are is what I'd like to watch for raccoons in this FOV.

 

[Pentagano]

Geez that's impressive. Spot the person. Ipcam-general using main stream

 

[Pentagano]

I appear to be seeing duplicate alert videos on my UI3 app within a second or of each other.
Is that because I have re-triggers ticked in alerts.. question mark

 

[Palmpilot]

For those of you running AMD rather than Intel/nVidia, I want to mention that the YOLOv5.NET integration with DirectML/DirectX 12 is working well for me, generally showing all sub-100ms times (most 40-60ms). I'm running on an AMD Ryzen 9 5900HX with built-in Radeon Graphics (arguably CPU overkill for Blue Iris alone). The built-in Radeon graphics lets me run in a mini PC form factor, so to get the GPU support without needing a discrete graphics card is (for me) fantastic. CodeProjectAI w/GPU support is showing a massive performance win over DeepStack that I was running until recently.

I also feel obligated to publicly thank @MikeLud1 again for everything he's done for the BI community. You're awesome!

@carteriii can you share how you got YOLOv5.NET to use your built-in Radeon graphics? is there some settings that I am missing? My PC has a Radeon RX Vega M GL Graphics (and I can see it under CodeProject.AI's System Info) but when click Enable GPU my YOLOv5.NET still shows "CPU".

Thank you!

 

[105437]

My critter detection is getting more dialed in. Last night a raccoon showed up and was about to start digging up my side yard looking for things to eat.


It was the ipcam-animal model that detected it.


And then, the PIR sensor on the Orbit Yard Defender detected the raccoon and engaged!

 

[rdxny]

yep seen those. Would cost 4 times that here in Uruguay and would get destroyed by the kids footballs in no time.
A 5 usd sprinkler activated by nodered suits my needs much better but thanks anyway.

I bet the sun would create false triggers for that kind of sensor also. CPAI would be (is) much more accurate imo

this is a big ask, maybe unrealistic, but is there a way to share the system you use to activate the sprinklers, assuming zero knowledge of nodered or any hardware you use? I guess asking for a tutorial, or maybe where to find that info without a whole new skill to learn (yes, asking a lot!) If too much I understand

 

[Pentagano]

this is a big ask, maybe unrealistic, but is there a way to share the system you use to activate the sprinklers, assuming zero knowledge of nodered or any hardware you use? I guess asking for a tutorial, or maybe where to find that info without a whole new skill to learn (yes, asking a lot!) If too much I understand

yep bit if a big ask unfortunately if you have nothing to start with. I have homeassistant (for the sonoff devices etc) and nodered, plus mqtt on proxmox containers and use these 3 objects to create the nodered flow.

 

[Pentagano]

this is a big ask, maybe unrealistic, but is there a way to share the system you use to activate the sprinklers, assuming zero knowledge of nodered or any hardware you use? I guess asking for a tutorial, or maybe where to find that info without a whole new skill to learn (yes, asking a lot!) If too much I understand

If you have no other pc to run linux then you could start with windows desktop docker

 

[rdxny]

If you have no other pc to run linux then you could start with windows desktop docker

I think I will need to aim lower Thanks tho...

 

[Pentagano]

I think I will need to aim lower Thanks tho...

Need to start somewhere. Have a look at windows docker desktop and watch some youtubes. You could get a few containers running - nodered/mqtt - what are you using as your BI mqtt broker?

 

[rdxny]

I have zero mqtt knowledge. Total noob, I can do BI, pushover, even manages a Kasa light alert with an exe by jdfromdb, but that's it. So I would have to start any home auto from scratch, which is why I suggested that I knew a tutorial for the sprinklers would be unlikely, but worth asking anyway. I might someday try to learn all this, but too big a task for now.