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

I have this working, however now I've got a few thousand spam alerts that are just jpgs. If I turn OFF "Capture an Alert List Image", it should still take the jpeg snapshot trigger pic, just not add it to the alerts, right?
View attachment 75162 View attachment 75163

I tested this, it still takes the jpeg snapshot, but it still shows in the list for both all cameras and the individual. I also tried turning this off but no change:

1605628744814.png

EDIT: aesterling is right, POST seems to do the trick. It keeps it out of the clip list. I turned off the "JPEG snapshot each" option on the Record tab.

1605629414634.png
 
Last edited:
Thanks for the input. For my setup, disabling both "Capture Alert List Image" on the trigger tab and unchecking "Include JPEG's on All Clips and Timeline" together keeps the stills from the alerts list/window but the trigger still sends the jpeg to the processing folder.

What I am now curious about is if it would increase performance to try the dual-streams option so we're processing lo-fi for recognition vs high def? I kind of suspect double cameras is probably faster too..24 hour recording at low-fi and triggering high-fi recording on alert, that has the added bonus of always running in case a trigger is missed...
 
I followed The Hook Ups Youtube tutorial, and created dual camera streams before realizing that 1.67 recommends using the flag option. What sort of issues should I expect with this mistake? I assume it's still recommended that switch over all my configurations to include the flag method, but am curious if that is what's causing missing alerts or if it's because I'm running deepstack on a separate unraid server on the LAN (or some other reason)
 
Hi all and thanks for the great thread. I've been following and using AITool since the earlier days but only just today upgraded to the Chris Dodge version.

I apologize if this has been asked before, I've only read through the first 88 pages of this thread.

I keep getting errors in AITool when using Telegram Cooldown. Error message is "ERROR sending image to telegram". This happens when I have the Telegram cooldown set to say 20 seconds but I have Blue Iris dumping JPEGS every 5 seconds while triggered. I want those images to continue to be processed and flags issued to Blue Iris, I just only want Telegram messages every 20 seconds. I don't understand why it's throwing an error in this situation. The behavior is as expected and it's not supposed to be sending to Telegram, so why does it throw an error?
 
I followed The Hook Ups Youtube tutorial, and created dual camera streams before realizing that 1.67 recommends using the flag option. What sort of issues should I expect with this mistake? I assume it's still recommended that switch over all my configurations to include the flag method, but am curious if that is what's causing missing alerts or if it's because I'm running deepstack on a separate unraid server on the LAN (or some other reason)
I have been running v1.67 with clone cameras and only moved over to the latest version yesterday! I assume you meant Clone and not simply dual stream? Single stream / Dual stream has very little effect on aitool operation. If setup correctly dual stream will enable higher resolution triggered images to be passed to deepstack for analysis.

If you are setup to record continously then flags are normally used. If like myself you only capture video after a triggered event there little if any advantage to using the flag option.
 
  • Like
Reactions: seth-feinberg
Hello,

First time poster here. :)
I've played for a few days with BI, AITool and Deepstack on a clean and fresh install of Windows 10 LTSC (chinese NUC with i7-8550U CPU).

Nice little piece of software! ;)

I have only 2 cameras for the moment (Hikvision and Annke), 4 more on the way.

With the latest version 1.67 and the flag system, is there any way to :
  • Use single camera, no duplicate, no clone, or whatsoever
  • Use dual-stream
  • Do not record video/image continuously
  • Have motion detection enable on BI
  • Save images on motion detection
  • Have these images analyzed by AITool+DeepStack
  • Trigger the video recording of the main stream in BI ?
I can't find how to do that. For the moment the only solution I've found is to duplicate cameras : one used for motion detection and jpg recording, and the other one fro video recording triggered externally by AITool. It works, but I'd like better to use a single camera if possible.

Thanks !


One evolution idea : it would maybe be interesting to forbid more than one instance of AITool. It happened to me a few times, double-clicking on its icon by mistake while it was already running, and it seems to not working well :)
 
You could instead save the JPEG using the “Post” tab in BI.
I have this working, however now I've got a few thousand spam alerts that are just jpgs. If I turn OFF "Capture an Alert List Image", it should still take the jpeg snapshot trigger pic, just not add it to the alerts, right?
View attachment 75162 View attachment 75163
I had this problem with a couple of different makes of ONVIF CantonK based Chinese cameras. It took ages for me to realize that the internal motion detector was firing continuously and triggering via ONVIF. In BI check the on the configure page of the cameras video settings page and uncheck "Get ONVIF trigger events". If this fixes it then you know the cause. The BI motion detection with AiTool works fine with them.
 
  • Like
Reactions: GreyTheOld
I had this problem with a couple of different makes of ONVIF CantonK based Chinese cameras. It took ages for me to realize that the internal motion detector was firing continuously and triggering via ONVIF. In BI check the on the configure page of the cameras video settings page and uncheck "Get ONVIF trigger events". If this fixes it then you know the cause. The BI motion detection with AiTool works fine with them.
This is a great tip...but I already had it off.

Now I'm working awesomely, EXCEPT that my alert videos are only showing up via the webUI, none are showing up in the app. Weird.
 
Still wondering what kind of issues I should expect using the <1.65 implementation on a 1.67 installation, but in the meantime I was watching some Youtube videos tutorials for general use and I was learning about the zone tool. Seems like I could use the Zone feature to mask my feeds and that seems easier than creating the custom picture in the original post. Is this something anyone has tried? Is it discouraged?
 
Still wondering what kind of issues I should expect using the

I use the masking feature in Chris Dodge’s fork for a bunch of my cameras and I recommend it over using the old method.


Sent from my iPhone using Tapatalk
 
Still wondering what kind of issues I should expect using the <1.65 implementation on a 1.67 installation, but in the meantime I was watching some Youtube videos tutorials for general use and I was learning about the zone tool. Seems like I could use the Zone feature to mask my feeds and that seems easier than creating the custom picture in the original post. Is this something anyone has tried? Is it discouraged?
Did you even bother to read my response to your earlier post?:confused:
 
I have been running v1.67 with clone cameras and only moved over to the latest version yesterday! I assume you meant Clone and not simply dual stream? Single stream / Dual stream has very little effect on aitool operation. If setup correctly dual stream will enable higher resolution triggered images to be passed to deepstack for analysis.

If you are setup to record continously then flags are normally used. If like myself you only capture video after a triggered event there little if any advantage to using the flag option.

Again, i'm sorry for not responding to this, I really don't know how I screwed that up, this is VERY helpful!

So i must confess I don't fully understand the difference between Clone and Dual Stream. My amcrest cam allow for multiple streams, I (think) i send snapshots from the substream to AI Tool and then send a trigger url to start recording on the HD/Main stream. Just based on the name I would think this would be Dual Stream, but I added each camera twice in Blue Iris so i really don't know. Sorry for my confusion.

As for your second point, I thought I'm just triggering recording of the HD stream when an alert is recognized but looking in BlueIris ALL of my SD cameras have a red circle in the top right and none of the HD ones do (all same cameras just added twice) so i don't believe I'm experiencing the setup I was intending to, i'll need to go back over my setup.
 
My program was working very well but a few days ago I noted it says there were many errors, so I checked the log and I can see that sometimes it tells me that:
"the process cannot access thephotofile.jpg because it is being used by another process"
And then
"ERROR: Processing the following image failed"

It is very strange because it only happens sometimes and I've never seen that error before.
I tried by rebooting and also by re downloading everything, updating the nuget packages and many errors appeared. But the "old version" without the nuget updates is working normally except for this can't access the file issue. I'm seeing that about 1/3 of the pictures to be analyzed receive that error, the others are processed without problems.

Hope someone can give me a clue what could be happening.
Thanks
 
Last edited:
Can anyone help me get the memo field working? I feel like I've tried everything and I have yet to see it work.

My trigger URLs:
Code:
http://192.168.x.x:x/admin?trigger&camera=Front&user=x&pw=x
http://192.168.x.x:x/admin?camera=Front&trigger&memo=[summary]&user=x&pw=x

Either one results in:
Code:
signal=green
profile=1
lock=0
camera=Front

In Alerts or Flagged I always see the thumbnail and:
Code:
Front
11/18/2020 6:41:33 PM
11 sec jpg H265

..never the memo. Any ideas what I'm doing wrong?
 
Again, i'm sorry for not responding to this, I really don't know how I screwed that up, this is VERY helpful!

So i must confess I don't fully understand the difference between Clone and Dual Stream. My amcrest cam allow for multiple streams, I (think) i send snapshots from the substream to AI Tool and then send a trigger url to start recording on the HD/Main stream. Just based on the name I would think this would be Dual Stream, but I added each camera twice in Blue Iris so i really don't know. Sorry for my confusion.

As for your second point, I thought I'm just triggering recording of the HD stream when an alert is recognized but looking in BlueIris ALL of my SD cameras have a red circle in the top right and none of the HD ones do (all same cameras just added twice) so i don't believe I'm experiencing the setup I was intending to, i'll need to go back over my setup.
OK Before you move forward you need to decide how you want your security camera's to operate.

1. The camera's record continously so you will never in theory miss anything and you wll have a full history for the period the files remain on disk. This method will rely on events being flagged. The disadvantage is that you may need significant storage space depending on how long you intend to keep your history.

2. Your camera's are setup to only record when an event happens. This has the advantage of taking considerably less storage space since it will only record to disk when an event is triggered by a camera. The disadvantage is that you will not have a full time elapsed record.

Config 1. Only needs one camera per security area being covered.

Config 2. Require's effectively two camera's, one to feed a trigger to aitool and the other for aitool to send a trigger. This only happens when the event is qualified by aitool when it will initiate a record for the desired period. To reduce cpu load this normally is accomplished by setting up BI to have a Master and a Clone feed for each camera.

Single or Dual stream has very little if any change required to the aitool configuration, it is specific to BI. It can assist aitool to capture high definition triggered images when a camera stream is monitoring in a lower resolution.
 
Thanks. You answered my question ! :)
For the "Operating mode 2", you need "Solution 2", and so 2 different virtual cameras for each physical camera. Period.
 
Anyone have any idea why i keep getting these errors? I have turned off 'queue actions' in settings and also each individual camera and still the same?

Skipping image because queue (101) is greater than '100'. (Adjust 'MaxImageQueueSize' in .JSON file if needed): C:\aialerts\frontdoor.20201119_223359298.jpg
 
Anyone have any idea why i keep getting these errors? I have turned off 'queue actions' in settings and also each individual camera and still the same?

Skipping image because queue (101) is greater than '100'. (Adjust 'MaxImageQueueSize' in .JSON file if needed): C:\aialerts\frontdoor.20201119_223359298.jpg
I have personally never experienced this issue, could it be that deepstack is unable to process the images fast enough and it simply runs out of alloted buffer memory? What is the maximum number of events that you estimate are being processed per second?