ESP32-CAM Video Streaming Camera that works with Node-RED and Home Assistant

In this video I will show you the ESP32-CAM. This is the cheapest WIFI camera that I've ever found on the market. Also This tiny Camera can stream video via web server and can work with Node-RED & Home Assistant and I will show you how.

To make this camera to work as I want I will flash it will Arduino sketch which will actually start a mini Web Server inside it that can be accessed from any web browser and of course Node-Red and Home-Assistant.

If you prefer to read check the article - https://peyanski.com/esp32-cam-video-streaming-camera-that-works-with-node-red-and-home-assistant/

I really hope that you find this information useful and you are going to make your ESP32-CAM to work with Node-RED, Home Assistant and as video streaming web-server

Thank you for watching, stay safe and see you next time.

Cheers,
Kiril

6 Likes

Thanks for posting this. I got a couple when they first came out to make the "free shipping" threshold when I got my first Coral TPU.

I've done nothing with them other than run the demo as I've put all my effort in the python code for an "AI person detector". But from what I see, they would work with my AI system.

That sounds very interesting. Can I see this in action somewhere ?

Here as a video clip of some local TV news footage of a home invasion that I downloaded and ran through my AI when I first got started:

An "alert" when the first "green box" is drawn could be a game changer!

Need "wake you up in the middle of the night" and "drop what you are doing" reliability. This means false positives are evil, false negatives are of near zero consequence if the frame rate and camera coverage is decent as a real positive is all that is required a few frames later.

I'm working on an update to my github, will be the third version. The second version is here:

The new version using the Coral TPU and a better AI is a major improvement.

Impressive, really good job. Keep it up!

I'm now considering to buy a coral USB accelerator for my raspberry after your answer :slight_smile: Are you using such (or some other kind) and you said first is that means that you have several?

I started with the Intel Movidius NCS, got an NCS2 as soon as they were available, and moved to the Coral TPU shortly afterwards.

I've kept support for the NCS/NCS2 and OpenVINO CPU AI (if you have an i3-4025 or better) but if you are buying something for the first time get the TPU at present. An NCS3 is supposed to be on the way in 2020.

The Pi3 basically can't decode rtsp streams. The Pi4 can do 4 or more HD streams, but struggles with 4K UHD streams -- about 2 4K cameras is it. Edit 1 is it, with two the latency makes it useless.

Your EPS32 cam works like Onvif snapshots. so a Pi3 should easily handle 5 or 6 of them although the lack of USB3 will hold back the frame rate you can get with the TPU.

Right now I'd advise a Coral TPU and the Jetson Nano for a compact AI "person detection" system. It can either use IP "netcams" directly or use the streams from a security DVR.

I recommend a security DVR if you want 24/7 or scheduled recording -- they've "solved" that problem at reasonable prices, but all the "consumer priced" ones I've played with have terrible "scrubbing" and their "motion detection" is so bad you might as well record 24/7 as you'll still miss stuff if you don't, and have mostly bogus motion events unless your cameras are all indoors. The AI detection events give you the timestamp of where to look in the video to see the entire event so it adds a lot of value.

For a do-it-youself security dvr look at motioneye or motioneyeOS.The Movidius NCS has been integrated into a branch of it: https://github.com/jasaw/motion/tree/movidius

A lot of discussion has happened here: https://github.com/ccrisan/motioneyeos/issues/1505

WOW, thank you for the detailed explanation. I have a homework now to get more informed :slight_smile: Thank you one more time and keep up the pace

This topic was automatically closed after 60 days. New replies are no longer allowed.