I always wonder why author of such software will not add some screenshot or gif showing UI and how it does work? In features there is mention of web interface.
How does this handle motion detection? Anyone tried it?
I have a bunch of Nest cameras and some cheaper Tapo cameras.
The "motion detection" is night and day different. The Tapo does some basic frame-diffing and it's awful. Shadows? DING DING DiNG MOTION DETECTED! gust of wind made some blades of grass move? Motion! Spider? Motion motion motion! Turn down the sensitivity a notch or two and it won't notice a human walk across the frame 2 meters away. The Tapo ones claim to be smart but twigs and leaves still trigger pet/person/motion alerts. It makes them essentially useless.
The Nest cameras are so much better and their "human" detected is usually zero-false-negatives at the cost of one or two false-positives perhaps once every 3 or 4 months, and their app is superior (tapo one frequently needs to be false-killed to load clips). Yes I am aware that the nest ones are streaming back to google 24/7.
Tldr: naive frame-diffing sucks for this sort of thing if used outside. An open source implementation that has accurate and reliable "human detection" would be amazing. Doesn't need to be "AI" - I would hope that there is some sort of computationally reasonable OpenCV way of doing person detection. Perhaps wait for frame-diffing to flag motion then feed it to a more expensive algorithm?
Seriously… how hard is it to setup an RPi with a camera and offload the detection to some open source software? Last time I checked there projects performing animal, face and even license plates reckon, extraction and db recording of “who”, “when”, “what” using cheap cameras and an RPi.
What is the difference between this and Frigate? Before clicking the link, I thought this would be a hardware solution (build you own camera from open components).
For those (like me) who don't know what "motion" refers to here, it's Motion Project, an open-source program to monitor multiple cameras and trigger real-time hooks based on motion detection, etc.:
My question about this project and the numerous similar projects: where do I get a usb camera suitable for this? Webcams typically have unsuitable enclosures, focus behavior, and low light performance.
Motion is a great software and it overlaps with CheapSecurity. However this software pretend to be:
1. a single-camera-per-instance CCTV with a web dashboard and Telegram bot (motion is a full-featured NVR/motion daemon: multi-camera, masks, timelapse, SQL logging, external triggers, etc...)
2. One independent process per camera; multi-camera is done by running multiple instances. (motion is multi-camera support in one process)
3. Small Flask dashboard with live stream, settings toggles, and bulk actions on recordings. (motion is functional stream/status pages)
4. Specifically optimized for low-power ARM boards like the Odroid XU4
5. Last but not least small Python/Flask/OpenCV codebase, easy to hack on...
Correct! But I have to be honest... I do not like the night vision since with a consumer webcam is not working very well... I will test with a cheap camera that support by default night vision... I will keep you posted!
19 comments
[ 0.27 ms ] story [ 36.6 ms ] threadI have a bunch of Nest cameras and some cheaper Tapo cameras.
The "motion detection" is night and day different. The Tapo does some basic frame-diffing and it's awful. Shadows? DING DING DiNG MOTION DETECTED! gust of wind made some blades of grass move? Motion! Spider? Motion motion motion! Turn down the sensitivity a notch or two and it won't notice a human walk across the frame 2 meters away. The Tapo ones claim to be smart but twigs and leaves still trigger pet/person/motion alerts. It makes them essentially useless.
The Nest cameras are so much better and their "human" detected is usually zero-false-negatives at the cost of one or two false-positives perhaps once every 3 or 4 months, and their app is superior (tapo one frequently needs to be false-killed to load clips). Yes I am aware that the nest ones are streaming back to google 24/7.
Tldr: naive frame-diffing sucks for this sort of thing if used outside. An open source implementation that has accurate and reliable "human detection" would be amazing. Doesn't need to be "AI" - I would hope that there is some sort of computationally reasonable OpenCV way of doing person detection. Perhaps wait for frame-diffing to flag motion then feed it to a more expensive algorithm?
Nest does it in the cloud. Obviously it's more capable but it also means Google grabs all your video: https://support.google.com/googlehome/answer/16594324?hl=en#...
https://motion-project.github.io/index.html
That's it