Last Watch AI - a new platform-independent AI tool with web interface

[kleypot]

Introducing Last Watch AI



akmolina28/last-watch-ai is yet another open source tool for glueing Blue Iris to Deepstack AI, inspired by gentlepumpkin's AI tools.

I created this tool last year to address some of the gaps and frustrations I was experiencing with the other Deepstack tools available at that time, namely:

• Web Interface - No need to login/remote to your BI server, just open a web browser on any device in your network to manage profiles and view detection events. Desktop- and mobile-friendly!
• RESTful API - Everything can be managed from the API. In fact, the web interface is just a skin that runs on top of the API. For example, you could set up Home Assistant to enable/disable certain AI profiles when you leave the house or go to sleep.
• Platform Independence - everything runs in Docker (including Deepstack) and works on Windows and Linux. This also makes it easy to automatically start everything when your server boots.
• Forkability - the code is intentionally designed to make it easier to fork and add more features and automations. One of my frustrations with gentlepumpkin's project is actually just a frustration with windows UI code and how difficult it is to maintain over time.

Here are some links to get started:

• Full introduction - Introducing: Last Watch AI
• Setup Guide with Blue Iris - Last Watch AI + Blue Iris Walkthrough
• Github - akmolina28/last-watch-ai
• User Guide - Last Watch - Getting Started Guide
• API documentation - akmolina28/last-watch-ai
• Home Assistant Example - Car Presence Sensor with Home Assistant and Last Watch AI

More about myself and my setup:

• I am a professional software engineer based in Florida currently working in transportation logistics. I have a strong background in full stack web development. In my free time I work on music and audio engineering, and home automation. I started working on Last Watch AI during coronavirus lockdown last Summer and I have been using it in "production" for a few months now. So far, it has been remarkably stable and free from errors or issues, to the point that I'm running out of things to fix or tinker with. I have been lurking here for a while now and thought this was a good time to share with the community. I hope you all find this useful and welcome your feedback, suggestions, and contributions.

• I am running Blue Iris with 3 cameras on a dedicated windows server, and I have Last Watch running in a Ubuntu container on my Proxmox server. I originally had Last Watch running on my windows/BI server, but found the performance was much better on Linux and migrated over. I have a samba share set up on my ubuntu container and set that up as an output folder in Blue Iris. I allocated just 2GB of memory and 2 CPU cores (intel core i5) for Last Watch and that seems to be more than enough to handle everything I can throw at it.

Thanks for reading!

 

[The Automation Guy]

Perfect timing! I've procrastinated about adding AI to my system until now and it really is something I need to tackle.

 

[concord]

Thanks for developing and posting. I'm currently using Sentry, but plan on switching, when I have time.

 

[fenderman]

I made this thread a sticky.

 

[Mike A.]

Nicely done! Even comes with good guides and documentation.

 

[Holbs]

more AI tools out there, the better

 

[105437]

Ready to give this a try, thanks for doing this and looking forward to seeing how it gets developed.

 

[xjust]

just tested it
seems to do the job
any way to define masks within the app ?

 

[105437]

Not sure this will run as a Docker container on my QNAP.

 

[kleypot]

just tested it
seems to do the job
any way to define masks within the app ?

Not yet but I'm interested in adding that. Thanks for trying it out!

 

[kleypot]

Not sure this will run as a Docker container on my QNAP.

Do you have access to the command line? If you have sudo/root access you should be able to get it working. Note that it is not a docker container, it's a docker-compose project. You can think of it as a script that tells docker to run several containers which share resources with each other.

 

[105437]

Do you have access to the command line? If you have sudo/root access you should be able to get it working. Note that it is not a docker container, it's a docker-compose project. You can think of it as a script that tells docker to run several containers which share resources with each other.

Thanks for the reply. The QNAP is only able to run Docker containers unfortunately.

 

[Saargebeat]

Runs fine under Windows 10. Didnt yet set up the triggers from AI to BI but that will work, too i guess

How can i change to Deepstack-GPU? I have a nvidia gpu and like to use the GPU for Deepstack

thanks

 

[kleypot]

Runs fine under Windows 10. Didnt yet set up the triggers from AI to BI but that will work, too i guess

How can i change to Deepstack-GPU? I have a nvidia cpu and like to use the GPU for Deepstack

thanks

I can't find any good resources on how to run Deepstack-GPU using docker-compose, but you should be able to run Deepstack separately. Just remove the deepstack service from docker-compose.yml so that it does not start as a dependency for Last Watch. That way you can start the Deepstack-GPU service on your own using whatever parameters you want. You will also have to change the DEEPSTACK_BASE_URL setting in src/.env to something like http://localhost:5000 instead of the internal docker service address. After you change those two files you would have to restart the containers.

 

[wittaj]

So do you need Docker or not to run this? I am Windows 10 only and when I tried running Docker in the past, it kept giving me an error (that I do not recall now) but every internet search for it and possible solutions didn't work.

 

[kleypot]

So do you need Docker or not to run this? I am Windows 10 only and when I tried running Docker in the past, it kept giving me an error (that I do not recall now) but every internet search for it and possible solutions didn't work.

Yes this runs in Docker and it is developed and tested in Windows 10, Windows Server 2019 and Linux. Not sure what issue you were having with Docker but you may have better luck on Windows 10 if you install WSL 2 first.

Install WSL

Install Windows Subsystem for Linux with the command, wsl --install. Use a Bash terminal on your Windows machine run by your preferred Linux distribution - Ubuntu, Debian, SUSE, Kali, Fedora, Pengwin, Alpine, and more are available.

docs.microsoft.com

WSL is a low-level architecture that helps you runs Linux on Windows without using virtualization. If you install WSL, Docker will use that instead of Hyper-V. Hyper-V is a virtual environment in Windows -- you may have gotten errors if Docker tried to use Hyper-V but you had not enabled hardware virtualization in your BIOS. In that case, using WSL would make that problem go away.

 

[biggen]

So do you need Docker or not to run this? I am Windows 10 only and when I tried running Docker in the past, it kept giving me an error (that I do not recall now) but every internet search for it and possible solutions didn't work.

Gotta get with the times. All the cool kids are running things in containers nowadays.

 

[Saargebeat]

I did install Docker on Win10 with the provided manual of this fork and it worked like a charm.

I will try getting the Nvidia Container running

 

[wittaj]

Gotta get with the times. All the cool kids are running things in containers nowadays.

Oh I know LOL. I guess I will try again and post the error message here and see if someone has a solution. I should have done that the first time I tried LOL. And I will try the suggestion @kleypot provided. I seem to recall it was something similar and I remember going into the BIOS but that still didn't fix it. Sounds like this one will.

I hate missing out on all these great tools people here are creating!

 

[biggen]

Yeah give it another try. Hopefully we can help!