Log in

No account? Create an account

Previous Entry | Next Entry

[videorecording cricket behavior]

So, further progress, and very good progress, at that!

I've been trying to work on two different aspects of the video recording process. The part that involves mEncoder and mPlayer comes after I've collected some videos. In Texas, I used a camera that had a Windows GUI that would only collect video at 100 frames per second, with a fairly crappy image resolution. That was okay for that project, because I wasn't collecting all that much video - just 30-minute snapshots of activity at two times of day for about 30 crickets in total.

Once I'd collected those videos in Texas, I used mEncoder on a machine set up to dual-boot to Ubuntu to go through a couple of post-processing steps to condense down the files into something I could analyze quickly with Biotrack. My tracking needs aren't nearly as precise as the researchers trying to track large numbers of ants, so I would knock the analysis file from 21 minutes at 100 frames per second (=126,745 frames) down to 32 seconds at 10 frames per second (=3169 frames). That's the equivalent of what, about 2.5 frames per second grabbed from the original? Then I would knock that even lower for Biotrack's background-generating tool, to 317 frames.

So on that front, all I need to do here is figure out a reasonable method for running mEncoder on Windows. Now that I've got that all sorted out, it's on to the next challenge.

The next challenge is figuring out how to set things up to minimize the video capture file sizes. Since I'm interested in collecting cricket activity data across the whole 24-hour light-dark cycle, the videos are going to be much longer - 1440 minutes as opposed to 20 minutes, or 72 times as long. Those could be some enormous files. So first, I need to figure out what my options are in terms of how to set up the video capture. I found this guy's page, which has a great little script, and which caused me to seek out "wget," and, lo and behold, it's easy to use in Cygwin, especially after you figure out the appropriate IP address for your webcam. (The most annoying thing about Cygwin, by the way, was just figuring out navigating and where it stuck me to begin with. Now we're all cool.) "wget" should give me a few more options than the Zavio GUI for how I want to collect images/video, and is way more satisfying than the half-dozen IP camera GUI "apps" that want you to use them instead. For instance, if necessary, I should be able to set things up to only record for ___ minutes out of every hour.

So, now that I have a bit of preliminary material to work with, and a basic method, time to do some math and some thinking about how I want to schedule the recordings. Progress.



( 1 remark — Remark )
Apr. 21st, 2016 11:07 pm (UTC)
I am vastly amused that we're both doing stuff on Windows. The paid project uses Cygwin on Windows, too. I've been ssh'd in to that and then also remote-desktoping in as needed. wget is an absolute staple (or curl, take your pick).

I don't know exactly what the interface for that camera is, but mencoder should be able to compress on the fly. It could take on wget's role of pulling over the network via HTTP, dropping frames, and writing the output, to save having the intermediate, large files on your HD.
( 1 remark — Remark )

Latest Month

October 2018


Page Summary

Powered by LiveJournal.com
Designed by Naoto Kishi