Log in

No account? Create an account

Previous Entry | Next Entry

If I whine about something, I should report progress on the problem, right right?

Here's what I figured out for running mPlayer or mEncoder from the command line. Apparently Windows 8 looks for relevant programs via what's specified at a global 'path' environment variable (see here for what I mean). It looks like there's supposedly a way to add programs to the "path" variable in a GUI fashion, at least according to this.

For whatever reason, at first when I queried this in the command line while I was in the directory that contains my video files (all you have to do is type in 'path'), it wasn't showing up. But apparently you can easily add in a new path directly in the command line anyway, so when I did that, it worked. And when I reopened the command line, the new path was present, so either changing it at the command line or changing it via the GUI did the trick. Huzzah. Now I can just type in 'mencoder [bunch of parameters] [video file name]' at the command line and my video will get processed (or played when I use mplayer).

I should note - basically, I want to do this because I'm anticipating wanting to process a large number of large cricket videos again. Our plan is to start measuring activity across the whole circadian cycle for crickets housed either by themselves or in groups. Another cool thing is that we should also be able to extract data on eating and drinking based on position. One of the grad students has nicknamed this project my "cricket voyeur" project because the setup allows me to spy on crickets inside of an incubator via an IP camera.

The annoying thing about most video analyses is that they tend to be enormously time-consuming, to the point where the amount of time it takes to analyze one video is equal to the length of the video multiplied by some ridiculousness factor. Consider, for example, videotaping a colony of 20 ants for 5 minutes and then going back to document what every ant is doing in that 5 minutes. I know people who have done this and it is tremendously tedious. I have found that crickets move slowly enough that it's possible to downsample videos from around 20 minutes to 30 seconds, making it feasible to process a crap-ton of videos from a crap-ton of crickets. [There tends to be a high level of variation in behavior, so large sample sizes are needed.] Assuming that I can get the tracking software to recognize crickets painted with a distinctive paint scheme designed to be visible in both incandescent and infrared light, ahem. That's the next step.

Regardless, the video quality this time around is WAY better than last time, which should help.


Latest Month

December 2018


Powered by LiveJournal.com
Designed by Naoto Kishi