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

Jooced said:
To me it looks like Deepstack didn't close correctly and it starting multiple instances.

Make sure to close down Server.exe, Redis-server.exe and Python.exe (I think). Then you can start up without errors.
Click to expand...
What would the right way be? I ask because I am running the VolronCD version that starts DS (and I assume everything else) automatically so I just click the X in the corner to close it if I need to because I need to re-boot for example.
 
Hi @Chris Dodge do you think the VorlonCD version can be modified to work with DOODS (Dedicated Open Object Detection Service) instead of (or in addition to) deepstackai? I have a lot of false positive alerts with deepstackai (latest version that was released this week either cpu or gpu) at night with IR vision, and I am curious if DOODS would perform better.

More info: snowzach/doods

CC: @Zach Brown the author of DOODS

Thanks!
 
Hi everyone. I feel like I'm in a little over my head on this one and could use some handholding. I have an aggressive neighbor who recently threatened my family to the extent that I filed a sheriff report. Today she ran her car through my front yard to cause damage because she thought nobody was watching. I got it on video and it will also go to the sheriff. I really want to get this working ASAP with phone alerts because right now it just records 24/7 and no alerts. Could I pay someone to help hand hold me through this setup and get it running. I am sure if I spent the time I could eventually arrive at a working solution but I'd really like to get this working better before things deteriorate further.
Maybe it’s the adrenaline making this so urgent but either way thanks for reading.
 
Last edited:
  • Like
Reactions: WildDoktor
I feel for you @Simcole; I hope somebody can help! That somebody, unfortunately, won't be me...I've been trying to get that working for 2 months, with very little success. :-(
 
He hasn't published a new release for 27 days, but it looks like he's actively making changes. I wish I knew how to compile a github project, as he suggested a few pages back that somebody compile his work on their own if they want the latest rev.
 
  • Like
Reactions: balucanb
WildDoktor said:
He hasn't published a new release for 27 days, but it looks like he's actively making changes. I wish I knew how to compile a github project, as he suggested a few pages back that somebody compile his work on their own if they want the latest rev.
Click to expand...

Here is a compiled version based on his latest code (Last update was 6 days ago). Just be aware it may have unfinished or broken features.

www.mediafire.com

AiTool_Oct27_Release

MediaFire is a simple to use free service that lets you put all your photos, documents, music, and video in a single place so you can access them anywhere and share them everywhere.
www.mediafire.com
 
  • Like
Reactions: balucanb
mayop said:
Here is a compiled version based on his latest code (Last update was 6 days ago). Just be aware it may have unfinished or broken features.

www.mediafire.com

AiTool_Oct27_Release

MediaFire is a simple to use free service that lets you put all your photos, documents, music, and video in a single place so you can access them anywhere and share them everywhere.
www.mediafire.com
Click to expand...
Thanks MAYOP! @WildDoktor let me know how it works so I can try it. :cool::p
 
No joy here. What a friggin' nightmare. I can't look at this any more tonight or I'm gonna say/do something immature. I'm just gonna let my sd camera continue to record 24x7 and go to bed...hope nobody comes to burn my house down because I won't get an alert.

Working from home tomorrow, so maybe I'll have a chance to work on this with a clearer head tomorrow during the day.
 
  • Like
Reactions: balucanb
Hey Guys,

is there any way of training the AI? It sometimes detects a Boat but for sure we dont even have water :) Also some Persons are detected but actually its a plant.

Most problems occur during night, especially when light is switched on due to motion detector (i mean a hardware device, not blue iris motion detection).


Besides that, is there some hardware which is known as doing fast AI processing? I am now using only 2 cameras, but it will be 5-7 later. At this moment processing one image (AI) takes 1,5-2 Seconds.
I am wondering how to operate lets say 5 cameras which might send images the same time - then it would take up to 10 seconds until all images (of one single motion detection) are processed?! But during this process, more images are sent to the folder...
 
@GentlePumpkin
Could help me out with you code. I'm trying to change cooldown minutes to cooldown seconds and after going through your code have been unable to understand how it works. I can change the text label but that's not going to do too much :) Any pointers would be much appreciated. I'm able to program in Assembler and Basic but never got around to C+. I can compile your code so no issue there.
 
All... what is the update process exactly? I'd like to try the new version but not lose my camera configs and history etc. Thanks
 
Had a few minutes this morning. I thought, "maybe deepstack running on the same vm as bi and aitool is too much; I'll install it in a docker container on a brand new Photon VM and see how that goes."

So I got it installed, and started it with the command
docker run --gpus all -e VISION-DETECTION=True -v localstorage:/datastore -p 80:5000 deepquestai/deepstack:cpu-x6-beta

Then in the aitool DeepStack tab I unchecked the Automatically Start DeepStack box. In the aitool Settings tab I set the new deepstack url to the ip : port of the new photon box.

Guess what? Still doesn't work.

Aitool log says:
"Got http status code 'Forbidden' (403) in 3ms: Forbidden"
"Problem with AI URL: 'http://(photon ip and port)/v1/vision/detection"
"All url's are in use or disabled, waiting..."

If I open http://(photon ip and port), the deepstack splash screen appears with no issues. If I add /v1 to the end, I get a 404 not found error.
 
Ok; apparently I also had to install a web server on the photon box (that makes sense);
docker run -d -p 80:80 vmwarecna/nginx got it installed.

Then I ran docker run --gpus all -e VISION-DETECTION=True -v localstorage:/datastore -p 80:5000 deepquestai/deepstack:cpu-x6-beta again, and aitools was able to access the deepstack url.

So I walked outside and hung out in front of the camera for about a minute.

And I suddenly got 7 text messages in a row, all showing the exact same snapshot, with the exact same timestamp, of my empty front yard.

Sheesh. I'm never gonna get this to work correctly!
 
You must log in or register to reply here.
Top Bottom