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

Village Guy

Pulling my weight
Joined
May 6, 2020
Messages
223
Reaction score
100
Location
UK
FIxed this.

It seems that there is a max image resolution. Not sure what the limit is but using 640 x 320 works every time.
Not sure what you fixed. My images are 12 Mega pixels and are roughly 8 megabytes in size.
 

Jooced

n3wb
Joined
Jul 9, 2017
Messages
7
Reaction score
1
Location
California
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?
Is anyone able to offer help on the memo lines not working? All I ever see is this:
 

Attachments

seth-feinberg

Young grasshopper
Joined
Aug 28, 2020
Messages
54
Reaction score
4
The tutorial DOES clone and DOES USE DUAL STREAMS.
HAHA YES! thank gawd, that was bothering me so much and finally makes logical sense to me. How do most people on here do it? with the cloned low res streams and recording the hi res? Or just do everything on the HI res if computational/storage capacity isn't really a problem...
 

Netwalker

Getting the hang of it
Joined
Aug 8, 2017
Messages
45
Reaction score
28
HAHA YES! thank gawd, that was bothering me so much and finally makes logical sense to me. How do most people on here do it? with the cloned low res streams and recording the hi res? Or just do everything on the HI res if computational/storage capacity isn't really a problem...
After a few tries at different ways over the last months, I've settled on:

Using IVS detection on the cameras, limit decoding on the BI camera streams. Full res only, recording 24x7, all BI motion detection turned off.
When the Camera causes a motion alert, BI creates an event and saves an image to the AI_Input folder. *
AI_tool then sends it to deepstack for analysis and if it returns positive;
Ai_tool flags the event in BI &
Telegram sends me an alert with an image and the detected object outlined.

The only issue I have with this setup is that I almost feel guilty how little it uses BI. The positive is since BI isn't doing any motion detection, I have nearly nothing for an average CPU load with very close to perfect alerts (I actually haven't noticed anything that either should or shouldn't be in the alerts I'm getting from Telegram, but I'm sure it can't be perfect). Using the newest Deepstack (/deepquestai/deepstack:latest ) it's processing images in ~400ms on a dinosaur i5. Read this issue thread on how to update deepstack if you don't know: Future of DeepQuest and DeepStack? · Issue #65 · VorlonCD/bi-aidetection

* I tried having the cameras upload FTP images. I think that would be the best way, but there are a lot of small house keeping issues with it I don't have the time to fix. file permissions, cleaning directories, etc.
 
Last edited:

seth-feinberg

Young grasshopper
Joined
Aug 28, 2020
Messages
54
Reaction score
4
After a few tries at different ways over the last months, I've settled on:

Using IVS detection on the cameras, limit decoding on the BI camera streams. Full res only, recording 24x7, all BI motion detection turned off.
When the Camera causes a motion alert, BI creates an event and saves an image to the AI_Input folder. *
AI_tool then sends it to deepstack for analysis and if it returns positive;
Ai_tool flags the event in BI &
Telegram sends me an alert with an image and the detected object outlined.

The only issue I have with this setup is that I almost feel guilty how little it uses BI. The positive is since BI isn't doing any motion detection, I have nearly nothing for an average CPU load with very close to perfect alerts (I actually haven't noticed anything that either should or shouldn't be in the alerts I'm getting from Telegram, but I'm sure it can't be perfect). Using the newest Deepstack (/deepquestai/deepstack:latest ) it's processing images in ~400ms on a dinosaur i5. Read this issue thread on how to update deepstack if you don't know: Future of DeepQuest and DeepStack? · Issue #65 · VorlonCD/bi-aidetection

* I tried having the cameras upload FTP images. I think that would be the best way, but there are a lot of small house keeping issues with it I don't have the time to fix. file permissions, cleaning directories, etc.
This sounds amazing! Thanks so much for sharing. I def am not experiencing near perfect alerts, so I'm eager to try to replicate your setup. Did you follow a particular tutorial? If not, would you be willing to share any tips on how you implemented it? (I'm embarrassed to say I even needed to google the IVS acronym).

For clarity, you are recording the HD stream 24x7 but are you quickly discarding any "clips" that are not relevant? What does your long(er) term storage strategy look like? Currently, I keep the 6tb HDD on my Blue Iris box full which I think should get me CLOSE to a month of motion captured 4k h265 and then offloading anything older to a unraid server on the LAN...
 

balucanb

Getting the hang of it
Joined
Sep 19, 2020
Messages
141
Reaction score
23
Location
TX
Be sure you put a '-' before your Chat ID #, also be sure your bot is invited to the group.
Re reading the whole thread so that is why I am replying to one that is so far back- Anyway I have seen many reminders about including the "-" and it pops up in AI-Tool also but for me if I include it it will not work, I have to make sure and not use it. Curious if anyone knows why? the way I made the bot originally maybe?
 

Village Guy

Pulling my weight
Joined
May 6, 2020
Messages
223
Reaction score
100
Location
UK
Re reading the whole thread so that is why I am replying to one that is so far back- Anyway I have seen many reminders about including the "-" and it pops up in AI-Tool also but for me if I include it it will not work, I have to make sure and not use it. Curious if anyone knows why? the way I made the bot originally maybe?
I have never tried adding it and never had any issues either. The adage of If it is not broken don't try and fix it comes to mind.
 
Joined
Nov 24, 2020
Messages
2
Reaction score
0
Location
Chicago
Am I doing something wrong that would prevent Telegram text from showing? All I see is a picture.

I deleted the registry at HKEY_CURRENT_USER\Software\AppDataLow\AITool and also the files in AppData\Local\WindowsFormApp2 and setup 1 camera from scratch and still only get a picture. Am I missing something here? I'm running version 2.0 also.
 

Attachments

barnyard

n3wb
Joined
Aug 9, 2020
Messages
23
Reaction score
5
Location
United States
Am I doing something wrong that would prevent Telegram text from showing? All I see is a picture.

I deleted the registry at HKEY_CURRENT_USER\Software\AppDataLow\AITool and also the files in AppData\Local\WindowsFormApp2 and setup 1 camera from scratch and still only get a picture. Am I missing something here? I'm running version 2.0 also.
I think there's a big in that version. I tried to build it from the latest source because I believe Chris said it's fixed since the release, but I get errors. I had to roll back to the previous version

Sent from my Pixel 3 using Tapatalk
 
Joined
Nov 24, 2020
Messages
2
Reaction score
0
Location
Chicago
I think there's a big in that version. I tried to build it from the latest source because I believe Chris said it's fixed since the release, but I get errors. I had to roll back to the previous version

Sent from my Pixel 3 using Tapatalk
Looks like you were right! I tried version 1.81 and I receive Telegram captions as expected. But now I get an error when creating a custom mask lol. Looks like I'll just use 2.0 till there is an update which hopefully fixes this.

Thanks for the help!!
 

Village Guy

Pulling my weight
Joined
May 6, 2020
Messages
223
Reaction score
100
Location
UK
Heads up: I just found out that if you remove "Spiny Lumpsucker, Pink Fairy Armadillo, Tasselled Wobbegong" from Additional relevant objects under the camera's tab, it causes aitool to trigger on anything that it recognises within it's database! This anomaly is present in build Version 2.0.149.7621 but may be present in other builds.
 
Last edited:

joshwah

Pulling my weight
Joined
Apr 25, 2019
Messages
216
Reaction score
104
Location
australia
For dynamic masking, say I add "Cars" under objects, which works great. It adds the mask under "active masks". Can someone please confirm, will any other object be detected in that mask? i.e. a person walking past? will it mask out all objects or only "cars"?
 

Village Guy

Pulling my weight
Joined
May 6, 2020
Messages
223
Reaction score
100
Location
UK
For dynamic masking, say I add "Cars" under objects, which works great. It adds the mask under "active masks". Can someone please confirm, will any other object be detected in that mask? i.e. a person walking past? will it mask out all objects or only "cars"?
If you are trying to simply ignore cars you would be better advised to use the object detection facility and simply choose 'Person'
A masked out area will for the time it is masked stop anything being recognised in that area.
 

dnorris78

n3wb
Joined
Oct 26, 2020
Messages
2
Reaction score
0
Location
Laurinburg, NC
I have everything setup and it is all working great ... except when I attempt to run DeepStack from the AI Tool I get errors when DeepStack launches. Otherwise DeepStack is processing images.

1606309302189.png
 

JL-F1

Getting the hang of it
Joined
Jun 12, 2020
Messages
50
Reaction score
35
Location
USA
Is everyone still using 'simple' motion detection or has 'edge vector' been fixed?
 

Village Guy

Pulling my weight
Joined
May 6, 2020
Messages
223
Reaction score
100
Location
UK
Is everyone still using 'simple' motion detection or has 'edge vector' been fixed?
I can't speak for anyone else, I am still using simple mode. I have not heard anything about a fix.
 

seth-feinberg

Young grasshopper
Joined
Aug 28, 2020
Messages
54
Reaction score
4
So after combing through this thread and the tutorials for a few days now and I think I have the resources/it makes sense for me to simply record on the 4k streams 24x7 (I currently have all my cameras cloned and am recording 24x7 on the low res stream and sending alerts to record the 4k stream)

I have a few questions:

For Trigger URL's, the [Camera] variable doesn't seem to work for me in any format of Long or sho
rt name of the camera, so I just hardcoded it in the URL:
http://localhost:47201/admin?camera=BackYardHD&trigger&user=aiuser&pw=password http://localhost:47201/admin?camera=BackYardHD&flagalert=1&trigger&memo=[summary]&user=aiuser&pw=password
1. If I'm only using the 4k stream recording 24x7 and then flagging events that AITool finds, why would I still need the first URL? I guess I'm wrong to have guessed that one calls BI to start recording?
2. And is the "[summary]" supposed to be included in the url verbatim? and for that matter is leaving it as [camera] supposed to pull in the camera name you set above in settings? Currently I have them all "hard coded" all as the same Short Name i entered above. So a different URL for each camera.

Finally, I have one clarification on the process:

3. Will i be able to use these "flags" to define what clips i save going forward? For example. If I'm recording the 4k stream 24x7 but AI detects a motion/person on that camera only for 5 minutes in a 24 hour period, will i be able to separate out the 5 minute clip for longer term storage and delete (or allow to be overwritten) the other 23 hours and 55 minutes?
 
Top