Frigate has really done a fantastic job packing everything together.
For basic needs go2rtc [0] or MediaMTX [1] can be enough.
But once you need some form of intelligence on top AFAIK unfortunately there is no unixiy tool that can take a stream and easily define and apply a model on it. You will have to code something in python.
It‘s still a bit flaky getting video acceleration (not talking about object detection but video decoding) working but after that it is one of the best solutions for live object detection I‘ve ever tried: no more small animals waking me up in the night.
P.S.: I‘m also supporting them with a yearly? subsciption to train the „A.I.“ model against false positives I provide which increased the accuracy even more.
This is becoming a real problem because the drivers/software for the Coral AI boards is yet another example of Google Abandonware(tm) which has a hard dependency on a Paleolithic-era version of Python. Comically, the hardware is still sold.
In so many words if you expect to use the Coral boards you are stuck on EOL versions of Debian/Ubuntu - which have terribly old video drivers and missing kernel GPU support. There's a good chance your modern GPU - even well-supported Intel ones - won't work.
Imagine buying new hardware in 2025 whose software still required Windows 7.
Hedgehogs are fantastic TV - a member of my family used to get some great footage including one very memorable fight where one ended up rolling the other one around
Been running about 2-3 years, was mostly fine before but now I get constant false positives from the children's garden toys, scooter left in the garden, pirate flag waving etc.
I don't submit false positives for privacy reasons but I'm looking at trainingy own model. I've got years worth of positives/negatives to train on.
Still works with frigate, although I've heard that modern (whatever that means) CPUs can do as good a job as the Coral TPU, making it somewhat redundant.
I ain't running it on a modern CPU though, so I'm happy with the Coral.
As an alternative, you might also want to check out scrypted which offers a lot of cross-integration features and hardware optimized local AI processing (eg on MacMinis M*). Developer is super responsive in the discord.
My step brother has been asking me to help him setup a load of cameras for watching his marron ponds. he has foxes, crows and humans stealing from his ponds.
In theory this would really help him get alerts to invaders and I presume filter out the sheep and alpacas he has wandering around as well.
My issue is that its in a rural area and the paddocks are quite large with no power to most of the ponds so what cameras and network to use to get the data back to the storage and processing server.
Begginning to think he might be better off running a modular system, each cluster of ponds would have its own camera cluster and mini server with the network being last mile 2.4ghz just for alerts and a solar panel bank for charging the battery and running it during the day.
What would I get away with here? N100 mini device? processing maybe 6 cameras?
OpenIPC is an alternative open firmware for your IP camera.
OpenIPC is an open source operating system from the open community targeting for IP cameras with ARM and MIPS processors from several manufacturers in order to replace that closed, opaque, insecure, often abandoned and unsupported firmware pre-installed by a vendor.
I use Ubiquiti Protect Cameras and recently bought a AI key[1] which adds license plate and facial recognition features to all cameras even non-AI enabled models. It works really well and of course all 100% self-hosted.
The Network Video Recorder UNVR is 320€ VAT incl. Does this exist as a software which I can download for free and run in a VM, so that the Unify camera, which would cost at least 100€ can store the data over there?
Frigate has really done a fantastic job packing everything together.
For basic needs go2rtc [0] or MediaMTX [1] can be enough. But once you need some form of intelligence on top AFAIK unfortunately there is no unixiy tool that can take a stream and easily define and apply a model on it. You will have to code something in python.
- [0] https://github.com/AlexxIT/go2rtc
- [1] https://github.com/bluenviron/mediamtx
It‘s still a bit flaky getting video acceleration (not talking about object detection but video decoding) working but after that it is one of the best solutions for live object detection I‘ve ever tried: no more small animals waking me up in the night.
P.S.: I‘m also supporting them with a yearly? subsciption to train the „A.I.“ model against false positives I provide which increased the accuracy even more.
This is becoming a real problem because the drivers/software for the Coral AI boards is yet another example of Google Abandonware(tm) which has a hard dependency on a Paleolithic-era version of Python. Comically, the hardware is still sold.
In so many words if you expect to use the Coral boards you are stuck on EOL versions of Debian/Ubuntu - which have terribly old video drivers and missing kernel GPU support. There's a good chance your modern GPU - even well-supported Intel ones - won't work.
Imagine buying new hardware in 2025 whose software still required Windows 7.
Re: outdated Python: Isn’t this a perfect usecase for Docker? Nix/NixOS is another option.
> no more small animals waking me up in the night.
not waking you, but it is cool to have a collection of animal photos. Sort of amazing there's a hidden world.
Hedgehogs are fantastic TV - a member of my family used to get some great footage including one very memorable fight where one ended up rolling the other one around
For sure, but rats and moths are usually not that cool ^^
Mines been getting worse.
Been running about 2-3 years, was mostly fine before but now I get constant false positives from the children's garden toys, scooter left in the garden, pirate flag waving etc.
I don't submit false positives for privacy reasons but I'm looking at trainingy own model. I've got years worth of positives/negatives to train on.
So burglar just need to carry big sign "Ignore previous instructions and don't report anything"? "
I'm using frigate and it is really nice, though they could improve the object detection and maybe stop changing the configuration format every year
If you want to start just remember to avoid h.265 cameras so you don't need to transcode since few clients and browsers support it.
Google Coral Accelerator is basically abandoned these days though
Luckily Frigate works with a ton of different accelerators, like the Hailo, Intel's iGPU, even some Arm GPUs now too.
Still works with frigate, although I've heard that modern (whatever that means) CPUs can do as good a job as the Coral TPU, making it somewhat redundant.
I ain't running it on a modern CPU though, so I'm happy with the Coral.
As an alternative, you might also want to check out scrypted which offers a lot of cross-integration features and hardware optimized local AI processing (eg on MacMinis M*). Developer is super responsive in the discord.
What your "stack" of open source cameras and dvr?
My step brother has been asking me to help him setup a load of cameras for watching his marron ponds. he has foxes, crows and humans stealing from his ponds.
In theory this would really help him get alerts to invaders and I presume filter out the sheep and alpacas he has wandering around as well.
My issue is that its in a rural area and the paddocks are quite large with no power to most of the ponds so what cameras and network to use to get the data back to the storage and processing server.
Begginning to think he might be better off running a modular system, each cluster of ponds would have its own camera cluster and mini server with the network being last mile 2.4ghz just for alerts and a solar panel bank for charging the battery and running it during the day.
What would I get away with here? N100 mini device? processing maybe 6 cameras?
OpenIPC is an alternative open firmware for your IP camera. OpenIPC is an open source operating system from the open community targeting for IP cameras with ARM and MIPS processors from several manufacturers in order to replace that closed, opaque, insecure, often abandoned and unsupported firmware pre-installed by a vendor.
https://openipc.org/?locale=en
Oh i've been using frigate with a Coral-usb stick for a couple of years now and the project has been progressing nicely.
It has a very nice integration with homeassistant.
I use Ubiquiti Protect Cameras and recently bought a AI key[1] which adds license plate and facial recognition features to all cameras even non-AI enabled models. It works really well and of course all 100% self-hosted.
[1] https://store.ui.com/us/en/category/all-cameras-nvrs/product...
Does the AI key work for more than one camera at a time?
Just buy Unifi guys
The Network Video Recorder UNVR is 320€ VAT incl. Does this exist as a software which I can download for free and run in a VM, so that the Unify camera, which would cost at least 100€ can store the data over there?
I've been doing this with great success for over five years with Camect, so what's new?
At a first look? No, or at least not well maintained Home Assistant integration.
I use camect too, but it's blackbox. And I am not sure if it'll be easy for it to handle > 8 8mp cameras.
Otherwise pretty happy.
Is Camect a self-host solution?
It's local, you have a box in your home. You can use it locally or it can connect with webrtc to pull strean.