[tool] [tutorial] Free AI Person Detection for Blue Iris

xjust

n3wb
Joined
Jan 19, 2021
Messages
22
Reaction score
3
Location
romania
[GIN] 2021/02/05 - 20:59:29 | 200 | 70.9997ms | 127.0.0.1 | POST /v1/vision/detection
found the same beta21.01 version on the forum. it seems to work now.

is there any plan to add minimum object size to aitool ?
i have several cats that are detected as 'human' in certain positions in the yard :)
 
Joined
Jan 21, 2021
Messages
3
Reaction score
1
Location
Somers CT
i followed the instructions for windows installation of the gpu version
installed all the drivers for nvidia including creating the developer accounts to dowload everything
works pretty much the same as without hardware gpu acceleration for me (250-300ms) on medium
I had this same problem. You need to go to the deepstack.cc forums and download the beta release. It dropped my times from 350-400 ms to 70-80 ms. Let me see if I can find a link. pow...

 

Saargebeat

n3wb
Joined
Oct 28, 2020
Messages
9
Reaction score
3
Location
Germany
Hello Friends of AI

can someone explain how to setup dynamic masks? I like to get rid of detection of parked cars. I thought i did everything but it doesnt create any dynamic masks

I dont have static masks or anything.



Also i have 150-200ms response time with RTX3070
Using newest beta

I tried MODE High but then i got more than 300ms

Also CPU goes up to 100% starting with 13% but GPU only quite stable 1% - during processing maybe 5%


Maybe a bit OT: I get the 10sec Push Images, but most of the time they start to late (person is gone already). Any idea how to improve?
 
Last edited:

spammenotinoz

Pulling my weight
Joined
Apr 4, 2019
Messages
208
Reaction score
128
Location
Sydney
Hello Friends of AI

can someone explain how to setup dynamic masks? I like to get rid of detection of parked cars. I thought i did everything but it doesnt create any dynamic masks

I dont have static masks or anything.



Also i have 150-200ms response time with RTX3070
Using newest beta

I tried MODE High but then i got more than 300ms

Also CPU goes up to 100% starting with 13% but GPU only quite stable 1% - during processing maybe 5%


Maybe a bit OT: I get the 10sec Push Images, but most of the time they start to late (person is gone already). Any idea how to improve?
I literally just clicked the "enable dynamic mask" button on each camera. Try a different version of AI Tool, perhaps your version has a bug.
Pre-recording should stop you missing any footage, but I recommend recording 24x7 continuous and have AI Tool flag footage for you.
I don't recall having any dynamic masking issues, I am currently running a self compiled version.
 

Saargebeat

n3wb
Joined
Oct 28, 2020
Messages
9
Reaction score
3
Location
Germany
I literally just clicked the "enable dynamic mask" button on each camera. Try a different version of AI Tool, perhaps your version has a bug.
Pre-recording should stop you missing any footage, but I recommend recording 24x7 continuous and have AI Tool flag footage for you.
I don't recall having any dynamic masking issues, I am currently running a self compiled version.

I took the lastest versions of AITool and the Fork.
Maybe i can get a confirmation of somebody running the same software versions?! Or any non-self-compiled one :)

Also maybe there is a short manual how it should work. Whatever i try, i dont get any dynamic mask
 

Tinbum

Pulling my weight
Joined
Sep 5, 2017
Messages
432
Reaction score
117
Location
UK

mirceab

n3wb
Joined
Jan 22, 2021
Messages
3
Reaction score
0
Location
Romania
If there are 4 alerts in a row, separated by 2 seconds each, only the first alert is being sent to Telegram, for the other 3 alerts it fails with ERROR sending image to Telegram error, no further details. Chat id and telegram token are correct since at least one image is sent to telegram. Cooldown is set to 0.033. Should I just try different values for caption and/or filename or any other parameter? Does someone have the "magic" settings that really work?
FYI, I migrated to vorloncd from the original aitool. Aitool used to to be able to send all the images to telegram.
 

mirceab

n3wb
Joined
Jan 22, 2021
Messages
3
Reaction score
0
Location
Romania
If there are 4 alerts in a row, separated by 2 seconds each, only the first alert is being sent to Telegram, for the other 3 alerts it fails with ERROR sending image to Telegram error, no further details. Chat id and telegram token are correct since at least one image is sent to telegram. Cooldown is set to 0.033. Should I just try different values for caption and/or filename or any other parameter? Does someone have the "magic" settings that really work?
FYI, I migrated to vorloncd from the original aitool. Aitool used to to be able to send all the images to telegram.
After playing a while with the settings, I managed to fix the issue myself, therefore the ERROR sending image to Telegram error no longer shows up if check the Queue actions setting:
1612775791463.png
 

spammenotinoz

Pulling my weight
Joined
Apr 4, 2019
Messages
208
Reaction score
128
Location
Sydney
I took the lastest versions of AITool and the Fork.
Maybe i can get a confirmation of somebody running the same software versions?! Or any non-self-compiled one :)

Also maybe there is a short manual how it should work. Whatever i try, i dont get any dynamic mask
I see no one has replied, if you don't want to compile yourself, you may wish to try this version to deal with the masking issue.

Dynamic masking - nothing created · Issue #77 · VorlonCD/bi-aidetection - Dynamic masking - nothing created #77

 

JL-F1

Young grasshopper
Joined
Jun 12, 2020
Messages
31
Reaction score
8
Location
USA
I have the cameras and the Server time synced with nettime

Everytime I look at Aitools history the time on the first motion event jpg is about 4 seconds behind the aitools history listed time, every jpg after the first are all in the same second.

The jpg time is the same second as the file timestamp

AMD 3700
64GB ram
NVME for OS and aitools and blueiris
SSD for jpgs
Hardwired camera
Nvidia 1660 super

deepstack GPU 2021.02.1
same issue with previous two betas as well

AiTools 2.0.627.7696
but it's been the same time gap with the previous versions as well

Any ideas why the time gap or how I can improve it?
 

seth-feinberg

Young grasshopper
Joined
Aug 28, 2020
Messages
54
Reaction score
4
y'all, i did it! 157 pages read. It took a month (did it when I had a sec here and there), and I'm not sure it was worth it (over a more targeted approach that is) but I def have a much better understanding of how everything works together and where the project has gone. Little embarrassed about some earlier questions (thanks for your patience @Village Guy and @austwhite !). I put links to the posts of everything I thought might be useful into an email draft as I went, and I'm currently making a 2nd pass but I did have 1 clarification before i get too far:

Is it required to be continuously recording on SOME stream (Main stream with flags or sub stream with triggers) in order for this to work? Or is it possible to simply have jpegs taken on motion without any continuous recording and then sent to AI Tool for confirmation and triggering?
 

Saargebeat

n3wb
Joined
Oct 28, 2020
Messages
9
Reaction score
3
Location
Germany
I see no one has replied, if you don't want to compile yourself, you may wish to try this version to deal with the masking issue.

Dynamic masking - nothing created · Issue #77 · VorlonCD/bi-aidetection - Dynamic masking - nothing created #77



Solution. Thanks



Is there somewhere a good explanation about these dynamic masks?
I see that it creates them now, but apparently not correct (due to misconfig i think).

So it masks a parked car, but also some car driving across the street. Later the parked car is gone but still driving cars are masked, too.
 
Last edited:

Village Guy

Pulling my weight
Joined
May 6, 2020
Messages
220
Reaction score
100
Location
UK
Is it required to be continuously recording on SOME stream (Main stream with flags or sub stream with triggers) in order for this to work? Or is it possible to simply have jpegs taken on motion without any continuous recording and then sent to AI Tool for confirmation and triggering?
1. No it is not required to continuously record.

2. If you ONLY want to record when AITool recognises something of interest you need to operate with a cloned camera configuration. The reason for this is that BI does not have an option to only record when externally triggered. I have asked the developer to add the facility of only recording when receiving an external command but alas he is not receptive to the idea.

3. If you do not use a clone camera, BI will record for the period of time set within the configuration every time the camera triggers no matter what AITool does or does not recognise. AITool's job is to send external flag events to enable you to search through the recordings for those that have been flagged by AITool.

Personally I use clone mode for all my cameras as I fail to understand the merit of recording from false triggers.

I urge everyone that would like to do away with Clone mode to send the developer a request to add an option to only enable recording when an external request is sent.
 

spammenotinoz

Pulling my weight
Joined
Apr 4, 2019
Messages
208
Reaction score
128
Location
Sydney
1. No it is not required to continuously record.

2. If you ONLY want to record when AITool recognises something of interest you need to operate with a cloned camera configuration. The reason for this is that BI does not have an option to only record when externally triggered. I have asked the developer to add the facility of only recording when receiving an external command but alas he is not receptive to the idea.

3. If you do not use a clone camera, BI will record for the period of time set within the configuration every time the camera triggers no matter what AITool does or does not recognise. AITool's job is to send external flag events to enable you to search through the recordings for those that have been flagged by AITool.

Personally I use clone mode for all my cameras as I fail to understand the merit of recording from false triggers.

I urge everyone that would like to do away with Clone mode to send the developer a request to add an option to only enable recording when an external request is sent.
Peace of Mind - You really want to go to all the effort of setting up CCTV, Configuring BI, DeepStack and IT Tools only to likely miss a key event. In all seriousness that is the outcome you can expect.

You are entitled to your view, however anyone with experience can only advocate for 24x7 recording,
It's an absolutely terrible feeling to come back after an event to find, that for whatever reason you didn't capture the footage and with the recent sales of 12TB drives, tell me again how storage isn't cheap.
In your proposed case BI + AI Tool + Deepstack all need to be working perfectly without issue. Too many variables.

I don't use clones, just 24x7 recording, with AI flagging events. Then just select the Flag Filter in BI to review the Flagged events. So many times I have then turned off the flag to view what was happening on the other cameras.

That being said I will give you a way to achieve your outcome;
  • Trigger -> Disable Motion Sensor
  • Post -> Maintain Ring of JPEGS say 20, but not too large
Queue an Image Update every x seconds
Suggest changing the filename &CAM.%Y%m%d_%H%M%S%t
Select an appropriate path

Under this approach, the system will create and maintain a ring of constant jpegs per camera, which you can configure AI tool to review and detect and have AI Tool trigger BI.
If you still get motion triggers need to make sure ONVIF \ camera external triggers are disabled (can't remember where, I think video tab)

Suggest using a different folder to one already used by BI
For fast analysis use an SSD, but may shorten the lifespan. Modern SSD's shouldn't be an issue.

Also what is the issue with clones? They don't appear to use any additional measurable resources when I tested them.
 
Last edited:

Village Guy

Pulling my weight
Joined
May 6, 2020
Messages
220
Reaction score
100
Location
UK
@spammenotinoz Respectfully you omit to mention the need for mirrored or RAID storage configurations which will be essential if you wish to record 7 x 24 reliably not to mention the waste of power.

Judging from your long post I appear to have touched a nerve! Are you defending the developers stance of not providing an option to only start recording upon receipt of an external command?

My experience within this forum has been that setting up clone cameras is confusing for some members. If the option to only record upon external request was made available it would greatly simplify installations for those of us that do not want 24 x 7 recording or to record upon every triggered event.

I have personally been operating multiple cloned cameras since this project started and have never missed capturing anything of interest. Due to the low storage capacity requirements of operating cloned cameras I am able to take advantage of solid state storage which is inherently more reliable than hard disks and uses considerably less power.
 

CAL7

n3wb
Joined
Nov 26, 2020
Messages
21
Reaction score
7
Location
Florida
@spammenotinoz , @Village Guy This is a great discussion and I thank you each for your POV. I'd like to learn a lot more about these approaches but my grasp of the terminology is insufficient to completely follow the arguments.

My configuration is for each physical camera to be configured as one logical camera with a high definition mainstream and another logical camera with a VGA substream. I record the VGA substream 24/7. The VGA substream also provides motion-generated images to AITOOL/DEEPSTACK which then triggers the mainstream camera for short high-def recordings.

(Note that I have avoided the term "cloned camera" because, ever since beginning this project a couple of months ago, I have been unsure whether what I describe above is a cloned camera or just a second logical camera for BI.)

I'm content with the results but also would welcome any hints on how to improve. I'd be grateful for even some clarification of terms so I can better follow this thread. Thanks to all!
 

Village Guy

Pulling my weight
Joined
May 6, 2020
Messages
220
Reaction score
100
Location
UK
@CAL7
I can confirm that you are not using a cloned camera configuration. A cloned camera configuration is required if you want to record events that are ONLY triggered via AITOOL. Presently you are recording 24 x 7 so cloning makes no sense for you unless you wish to change your configuration to only record AITOOL triggered events.

If you are happy to record everything then you have no need to configure a clone setup.

I hope that clarifies the option.
 

seth-feinberg

Young grasshopper
Joined
Aug 28, 2020
Messages
54
Reaction score
4
I agree with @CAL7 , even if it sounds testy, i greatly appreciate the back and forth on this subject. I've never had a home surveillance system, and personally, it does seem wasteful to record 24x7 BUT I can totally envision the scenario where I'm livid I missed the ONE thing I wanted to catch at a given time.

2. If you ONLY want to record when AITool recognises something of interest you need to operate with a cloned camera configuration. The reason for this is that BI does not have an option to only record when externally triggered. I have asked the developer to add the facility of only recording when receiving an external command but alas he is not receptive to the idea.
Thank YOU! this is exactly what was confusing me so much. After 157 pages I simply couldn't figure out why people were still recording the substream while taking pictures and not just taking pictures on motion. Makes total sense now to find out that Blue Iris simply does allow you to ONLY record on external trigger. What is the best procedure for logging the request with Ken (i think that's the BI guy)? Seems that whatever avenue i go, having that option sure would be nice.

@CAL7 fwiw it also appears to me that you didn't explicitly clone your cameras, just created double instances of each physical camera. I assume this because i believe that you'd need to have explicitly cloned the doubled up cam during creation and then made changes (and thus would've been sure). I also assume this cause it sounds like you followed The Hook Up's video, and that simply creates double instances (I believe).

So in the best of both worlds, would I be able to record 24x7 on the substream for, say, 24 hours, while concurrently recording 4k clips of AI Tool confirmed events for longer term storage? I believe that's what @austwhite suggested to me 10 pages back or so, but haven't had a chance to make all the necessary changes. The question here is if BI is able to simultaneous record 2 of the same stream (the ongoing 24x7 low res record and the AI Tool deemed pertinant 4k clip)?
 
Top