Frigate runs motion capture and only sends images to the tpu when motion is captured. It's not pulling feeds from the cameras. 5 fps is enough to capture enough motion to send the image for recogition. How this would work in the blue iris method of pulling a constant stream of images I am not sure. I haven't ever used blue iris and don't really intend to because I refuse to run windows for server duties.
Hell Yeah! Direct Deepstack Integration - 5.4.0 - March 31, 2021
- Thread starter fenderman
- Start date
sebastiantombs
Known around here
My understanding of DS implementation in BI is that it sends captures, frames, once motion is detected by BI. The number of frames and timing are configurable in BI. Obviously upping the number of frames sent to DS and reducing timing can improve detection but at the cost of increased detection time and additional CPU/GPU loading.
Been playing with frigate with 4 cams set up. proxmox lxc container and homeassistant on an lxc container for the integration.
Was quite a pain setting up but actually pleasantly surprised at the detections as it is not so sensitive and does not get triggered with sun/shade events etc.
Picks up people very well without false triggers.
Played with masks and sensitivity etc. My small CPU suffered but coped peaking at 80% just. Without a coral tpu, I know. But they are all out of stock.
I Haven't managed to pick up the clips and send by nodered yet, just alerts etc.
CCTVCam
Known around here
- Sep 25, 2017
- 2,815
- 3,683
For what it's worth, I asked the developers of BI to consider supporting these given that Nvida seems to be retiring the Quadro workstation gpu's. They said they would look into it.
In the meantime. I guess if anyone one want integrate a Coral product, they're going to need to be a programmer.
For the money, Coral products look amazing.
In the meantime. I guess if anyone one want integrate a Coral product, they're going to need to be a programmer.
For the money, Coral products look amazing.
All... looking for an assist. I have installed my P400. I have downloaded and installed CUDA v10.1.
Is there anything else I need to do before I install the GPU version of Deepstack? Thanks!
Is there anything else I need to do before I install the GPU version of Deepstack? Thanks!
First install these files and then download and install DeepStack for GPU:
Install DeepStack GPU
To install and use DeepStack GPU version on your Windows machine, follow the steps below
>> Download GPU version for Windows
Install DeepStack GPU
To install and use DeepStack GPU version on your Windows machine, follow the steps below
- Ensure your machine has an NVIDIA GPU
- Download and Install CUDA 10.1 from NVIDIA
- Download and Install cuDNN from NVIDIA
>> Download GPU version for Windows
I extracted it to my desktop where it created a cuda folder. Does that folder need to be moved somewhere? Thanks
@wittaj I just did all of these steps for Windows except number 5, which I didn't understand. It seems to be running but I must have something not configured correctly... times are worse than using the CPU version without a GPU. Any thoughts?
3 11/17/2021 2:02:30.042 PM Front_Cam MOTION_A
3 11/17/2021 2:02:32.132 PM Side_Porch_Cam MOTION_A
1 11/17/2021 2:02:45.453 PM Front_Cam DeepStack: Server error 100
0 11/17/2021 2:02:45.453 PM Front_Cam DeepStack: Alert cancelled [DeepStack: 100] 15006ms
1 11/17/2021 2:02:46.026 PM Side_Porch_Cam DeepStack: Server error 100
0 11/17/2021 2:02:46.026 PM Side_Porch_Cam DeepStack: Alert cancelled [DeepStack: 100] 11752ms
0 11/17/2021 2:02:48.015 PM App DeepStack has been restarted
3 11/17/2021 2:03:16.232 PM Front_Cam MOTION_A
3 11/17/2021 2:03:18.309 PM Front_Porch_Cam MOTION_A
3 11/17/2021 2:03:24.330 PM Doorbell_Cam MOTION_A
0 11/17/2021 2:03:26.945 PM Front_Cam DeepStack: person:87% [64,77 128,210] 8934ms
3 11/17/2021 2:02:30.042 PM Front_Cam MOTION_A
3 11/17/2021 2:02:32.132 PM Side_Porch_Cam MOTION_A
1 11/17/2021 2:02:45.453 PM Front_Cam DeepStack: Server error 100
0 11/17/2021 2:02:45.453 PM Front_Cam DeepStack: Alert cancelled [DeepStack: 100] 15006ms
1 11/17/2021 2:02:46.026 PM Side_Porch_Cam DeepStack: Server error 100
0 11/17/2021 2:02:46.026 PM Side_Porch_Cam DeepStack: Alert cancelled [DeepStack: 100] 11752ms
0 11/17/2021 2:02:48.015 PM App DeepStack has been restarted
3 11/17/2021 2:03:16.232 PM Front_Cam MOTION_A
3 11/17/2021 2:03:18.309 PM Front_Porch_Cam MOTION_A
3 11/17/2021 2:03:24.330 PM Doorbell_Cam MOTION_A
0 11/17/2021 2:03:26.945 PM Front_Cam DeepStack: person:87% [64,77 128,210] 8934ms
Bit off-topic but I was playing around with Frigate on windows - on CPU. 4/5fps - 4 cams substreams low resolution.
Not a bad CPU Ryzen 5 3400. But look at the peaks during heavy activity!
Coral tpu devices are still all out of stock!!
Was doing this for possibly building systems for some friends. Blueiris windows + GPU or good intel = expensive. Frigate + coral tpu USB = much cheaper.
Also testing Frigate on my small intel atom minipc - seems to do ok. 80% peaks when detecting,
Not a bad CPU Ryzen 5 3400. But look at the peaks during heavy activity!
Coral tpu devices are still all out of stock!!
Was doing this for possibly building systems for some friends. Blueiris windows + GPU or good intel = expensive. Frigate + coral tpu USB = much cheaper.
Also testing Frigate on my small intel atom minipc - seems to do ok. 80% peaks when detecting,
The best bet I have found for getting a coral is to either put in a pre-order on mouser where I got mine and wait it out. The other option is to suck it up and pay a scalper, one seems to be releasing them on amazon every few weeks at $150ish. M.2s seem to be more available but you need a motherboard that supports the A+E key. Other than that you're right on the money, any newer intel chip with quick sync and coral is much more affordable for most people.
sebastiantombs
Known around here
@105437 I could swear that you need to copy some files from the cuDNN subdirectories into the CUDA directory structure but I can't remember where I saw that. I'll look in my office later, I think I printed out the directions and that will have a URL on it.
kklee
Pulling my weight
I had made a note of these instruction from somewhere (can't recall where, maybe from another post here).
Deepstack GPU Installation
You must install.
Nvidia driver of your video card.
The CUDA Toolkit 10.1
The CUDA Toolkit 10.1 update 1
The CUDA Toolkit 10.1 update 2
Extract the cuDNN NVIDIA lib into your CUDA Toolkit folder.
Don't forget to read the windows instruction to how extract cudnn in cuda.
You have to set a environment variable also.Before issuing the following commands, you'll need to replace x.x and 8.x.x.x with your specific CUDA and cuDNN versions and package date.
Where:
The CUDA directory path is referred to as C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\vx.x
The cuDNN directory path is referred to as <installpath>
Procedure
Navigate to your <installpath> directory containing cuDNN.
Unzip the cuDNN package.
cudnn-x.x-windows-x64-v8.x.x.x.zip
or
cudnn-x.x-windows10-x64-v8.x.x.x.zip
Copy the following files into the CUDA Toolkit directory.
Copy <installpath>\cuda\bin\cudnn*.dll to C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\vx.x\bin.
Copy <installpath>\cuda\include\cudnn*.h to C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\vx.x\include.
Copy <installpath>\cuda\lib\x64\cudnn*.lib to C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\vx.x\lib\x64.
Set the following environment variables to point to where cuDNN is located. To access the value of the $(CUDA_PATH) environment variable, perform the following steps:
Open a command prompt from the Start menu.
Type Run and hit Enter.
Issue the control sysdm.cpl command.
Select the Advanced tab at the top of the window.
Click Environment Variables at the bottom of the window.
Ensure the following values are set:
Variable Name: CUDA_PATH
Variable Value: C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\vx.x
Include cudnn.lib in your Visual Studio project.
Open the Visual Studio project and right-click on the project name.
Click Linker > Input > Additional Dependencies.
Add cudnn.lib and click OK.
sebastiantombs
Known around here
kklee
Pulling my weight
Visual Studio is a software development tool, not needed for BI/DS deployment. I should have left that part out, the text was a cut and paste I grabbed from somewhere.