What does HackerNews think of frigate?

NVR with realtime local object detection for IP cameras

Language: Python

#2 in Tensorflow
Bosch, Hanwha, Axis, Ubiquiti/UniFi.

I personally prefer Bosch and Hanwha cameras. Great optics, low light performance and solid firmware. Axis tends to be expensive and low light performance is not as good, for the price. UniFi Protect cameras are decent, but the standalone firmware is rather limited.

I connect these cameras to Frigate[1] locally.

Some example model numbers: Bosch Flexidome IP Starlight 6000 Dome Security Camera - NIN63023A3; Hanwha Techwin XNO-8080R WiseNet X Series Network Bullet Camera 5MP 3.7-9.4mm; AXIS P1468-LE Bullet Camera

I usually find these on eBay.

[1] https://github.com/blakeblackshear/frigate

These have been huge for things like Home Assistant and object detection in real time video processing via Frigate and Google Coral TPU Accelerator. It sure cut down on the false positives I was getting for detecting cars and people coming up my driveway and not deer or my dog walking across it. They're only about $100 and come in M.2 and USB form factors.

https://github.com/blakeblackshear/frigate

If you want a truly local camera system with all the fancy features, check out Home Assistant (homeassistant.io) and Frigate (https://github.com/blakeblackshear/frigate).
Wyze Cam v3 (wifi) -> Its 35$ with very good night vision ( amazon link )

Make sure to get a SD card for local recording

Then use this docker app to get raw RTSP urls and LAN only -> https://github.com/mrlt8/docker-wyze-bridge

Next use Frigate as your NVR -> https://github.com/blakeblackshear/frigate --------

The wyze cams use about 3-5 watts of energy over USB. This make perfect for using a USB battery bank, a solar "generator" like the comments mention, or any place you can get "usb" power. This assumes you have Wifi, but if you have say a 4g modem, id assume it has wifi hotspot on it, most do. Then you still can connect the wyze cams to them.

If you are concerned about privacy, thats where the docker app + sd card comes to play. You are recording to SD and also RSTP streaming via the docker app. If you are ultra concerned, just block the api calls with your favorite firewall/adguard/pi hole.

I have this exact setup and very happy. Runs on docker, I self host via traefik and VPN into my box for access.

I haven't looked into it, but someone else recommended https://github.com/blakeblackshear/frigate which integrates with Home Assistant.
Rolling your own is probably your best bet as being able to check remotely typically means your camera streams are running through a company's servers.

Personally, I moved from BlueIris and I've been incredibly happy with Frigate which is configured to pull from my local IP cameras - none of which have direct access to the internet: https://github.com/blakeblackshear/frigate

You can check in via browser assuming you have some method of accessing your local network. Paired with a google coral device (or a beefy CPU) alongside local storage, it also has the ability to detect objects in real-time which often is a feature that requires video processing on a company's server.

For any camera that provides an rtsp stream, you can use frigate[1], which has some excellent detection algorithms (it differentiate between people, dogs, ...). It also has hooks to integrate with HASS. I use it with a LaView system to monitor my external cameras and get notifications (including push to my iphone) through HASS.

[1] https://github.com/blakeblackshear/frigate

Frigate is another open-source object detection engine that works amazingly well with Home Assistant: https://github.com/blakeblackshear/frigate
> open-source . . . home security camera solution

i (deeply) reject your inquiry as a harmfully narrow, reductive consumerist ask, when this could be a much more neutral, open ended friendly question. you have qualified this from the start as a "solution," which is against the best premises, the best strengths of open source: "small pieces, loosely coupled."

the best software is small, purposeful, targetted. interwoven with other systems. only then is open source able to continue to focus on innovation & advancement, without becoming mired down in endless maintenance & complexity.

you should re-evaluate your ask, to ask for something that, will, in the end, not becoming limiting & ossified. open source ought best avoid the pretense commercial software competes on, of being a complete and final thing, of being everything to everyone. open source ought be more humble, and for this, it is better. ask, instead of a solution, after what systems of software might help one accomplish the home security systems they might want to build.

the best piece of open source home security cameras that I've seen is Frigate, which has masking & less interesting to me but probably interesting to many, object detection. designed for home assistant but it has other uses. much assembly required. good. solve your problem how you want to solve it: not how everyone else also has to.

i'd point out that home assistant itself is regarded by many as somewhat of an abomination, too big, unwieldy. core has 1.4k issues open and almost 300 PRs open. it's a shit show. it does way too much, it's way too monolithic, the project is basically doomed to stay where it is. it's a solution, and one that will rot & be no better in 3, 30, or 300 years: there's no open future here. it's only upside is that it is a plugin architecture, that it is an interoperation layer, that projects like Frigate can contribute & add capabilities too. home assistant is hugely popular, but basically it's core redeeming feature is that it allows small pieces, loosely coupled, to do something. home assistant itself is an anti-solution. frigate is an anti-solution. they are both parts, pieces, and in that is the strength. and hopefully, someday, we can kind a better system of pieces such that we can get rid of bloody ugly home assistant.

https://github.com/blakeblackshear/frigate

Chezck out the Frigate project. It uses the Coral tpu accelerator which could be used with the rpi.

https://github.com/blakeblackshear/frigate

I've just started down this path using a plain RTSP-serving camera and a low end box using the Coral EdgeTPU to process the frames. It looks like there are a variety of solutions available. https://github.com/blakeblackshear/frigate https://docs.ambianic.ai/users/configure/ etc