In addition to
automatically monitoring my TrailCam, the
garden based Raspberry Pi has continued to watch the
Blue Tits in
Bird Box III which have this year laid a clutch of seven eggs, averaging one a day.
 |  |
| No eggs | One week later, seven eggs |
Here are some of the intermediate images:
 |  |
| The first egg, recently laid | Two eggs |
 |  |
| Three eggs | Four eggs |
 |  |
| Feather delivery | Nesting material |
 | |
| Five eggs | Seven eggs |
 |  |
| Still seven eggs | Home delivery |
 |  |
| More insulation? | Another week later, no advance on seven eggs. |
These glimpses of the eggs are few and far between - most of the images are of one nesting
Blue Tit moving on the nest, with only a handful of images with both parents visible.
With the default
motion settings I was getting tens of thousands of images a day recorded - which was causing trouble filling up the Raspberry Pi's SD card, and occasionally failing with a read only file system. That required me to reboot and delete older images to make space. I've imposed a maximum of 5 frames a second for now...
According to the internet, the typical
Blue Tit egg incubation time is around two weeks, so these should hatch about the start of June.
Update:
The
Blue Tit eggs hatched at the end of May
Hi
ReplyDeleteI followed your posts before, and built myself a birdbox using a Pi, IR-removed XBox camera, POE etc - almost the same setup as you have.
It's worked flawlessy since setup.
I housed it all in a large "Dribox" - plenty of room with weatherproox inlets for cables etc.
I had blue tits nesting in ours; laid 5 eggs (quite a small amount for a blue tit) and they all hatched.
Unfortunately they've all died one by one, last one went at the weekend :(
Fingers crossed yours do better !
Anyway, a bit of techy info.
I was running the camera at 640x480, and found this eating CPU time with motion, even at around 2 or 3 fps.
I switched to the following setup:
- mjpeg-streamer - this captures the image from the camera in mjpeg format, and makes the stream available. We were then able to watch a live stream on the home network, which updated a bit faster than motion could manage.
- motion - used the stream output from mjpeg-streamer as input (ie not direct from the camera) and carried out motion detection, snapshotting etc.
Seemed to work better than letting motion do it all; motion doesn't seem too efficient at capturing from the camera.
Anyway, I've just ordered myself a RPi camera module to tinker with; has the potential to offer better framerates and resolution. I believe it's possible (although tricky) to remove the IR filter too.
Keep up the good work!
Thanks 'MuddyBoots'.
DeleteIts a shame about your Blue Tits, but as you say, fingers crossed for this clutch.
As to the technical side, as I'm running two cameras I've been limited to 320 x 240 pixels, and larger images would just fill up the SD card even faster. However your idea to use mjpeg-streamer is neat - you could have motion running back at the house on a separate computer, so that the Raspberry Pi never has to write loads of data to the SD card.
I've also ordered the RPi camera module, but haven't had a chance to open it yet - let alone read up on the IR modification.
Thanks!
Peter