Hi Folks,
First of all, i'd like to thank everybody involved in this project, nice work you're doing here!
As stated in the subject, i'm thinking of using the pi as the "heart" for a cheap autonomous surveillance system, which will roughly include:
* Raspberry Pi
* webcam
* WLAN / UMTS stick
* solar module and lead-acid battery for powering the stuff
This leads to a few questions, hopefully someone can answer them so that i know if i can continue my thoughts on this project or forget it ;)
here we go:
* does the Pi support some "floating" input voltage between 12V and 14VDC?
* has anyone tried running a HD (720p) webcam on the board?
* do you think the CPU will be strong enough to run motion detection on videos (i.e. "motion" for linux)
* will the final board have microphone input or onboard mic?
Thanks in advance,
greetings from austria,
Juergen
Raspberry pi as autonomous surveillance system
4 posts
- Posts: 1
- Joined: Sat Sep 10, 2011 3:55 pm
Search the forums each point exceot the HD webcam has been discussed, and it should run one fine but it would severly reduce traffic on your USB ports.
- Posts: 167
- Joined: Fri Jul 29, 2011 4:20 pm
How well motion will run on the Raspberry Pi will depend largely on the resolution and frame rate that you want to achieve with your camera. An idea of the needed CPU power can be found from this page
http://elinux.org/BeagleBoardUbuntu
which says "Using a 960x720 resolution webcam with 15 fps rate under the UVC driver the Rev C BeagleBoard under Xubuntu reports ~60% CPU utilisation." The BeagleBoard has a 1 GHz Cortex-A8, while the Raspberry Pi has a 700 MHz ARMv6. We will have to wait until the release to see what the performance for this application will be unless some alpha board owner does some motion testing.
http://elinux.org/BeagleBoardUbuntu
which says "Using a 960x720 resolution webcam with 15 fps rate under the UVC driver the Rev C BeagleBoard under Xubuntu reports ~60% CPU utilisation." The BeagleBoard has a 1 GHz Cortex-A8, while the Raspberry Pi has a 700 MHz ARMv6. We will have to wait until the release to see what the performance for this application will be unless some alpha board owner does some motion testing.
- Posts: 66
- Joined: Fri Aug 26, 2011 7:31 am
What you really need is a camera attached to the GPU. We hope to have something like this next year sometime, once we find a supply of cameras. You could then get h264 out of the GPU with no impact on the Arm at all. Not sure about motion detection, but you may be able to tap in to the H264 stream and get the course motion vectors which would be moving around if there is motion present, which would save you decoding it all. I'm not H264 expert though, and all the codec people here are on a course, so cannot ask them!
- Moderator
- Posts: 6468
- Joined: Sat Jul 30, 2011 7:41 pm