Appendices
IR lights:
Outdoor IR Illuminator, 850nm, 30 degree beam, 12VDC or 24VAC (Reytec imaging; part # IR-LT30)
IR light power supply:
24 VAC power supply (Reytec imaging; part # C24AC-40)
camera:
Basler A622FM 2/3" CMOS FireWire.A Monchrome Camera (Edmund Optics; part # NT58-297)
Lens:
16 mm Fixed Focal Length Lens with Manual Focus and Iris, C-mount, 2/3" format, F-stop: 1.4,
IR-pass filter:
Hoya RM72 Infrared filter (B&H Photo; HOIR7252), passes above 720nm
Hair bleach:
Nice 'n Easy Born Blonde Maxi
Git:
tortoisegit.net
NX client:
http://www.nomachine.com/download.php
First download:
Once you've installed git* on Windows, when you right click you have a drop down menu that includes some git commands. To download the data from github, right click to display the menu and select "Git Clone...". This will pop up a window with that says "Clone existing repository" at the top. Enter https://github.com/motr/motr for the URL and directory on your computer where you'd like to store the motr files for Directory (e.g. c:\USERS\egnor\motr) and then click OK.
*at least this is true for TortoiseGit
Subsequent downloads (updates):
To stay current as improvements are made to motr, you can update using git. If you'd like to update the files you already have, just right click on the motr directory and select TortoiseGit, and then select
Pull from the drop down menu.
In general, with the exception of the *_tracks files, all the files and directories created by motr while running can be deleted once you are happy with the *_tracks files. Below is a more detailed description so you can decide what to keep for yourself.
Jobs directory
The Jobs directory contains a sub-directory per tracked video file. Inside, there are many files (JobInXXX.mat, each for a submitted job). Those files contain the initial conditions (where the mice are), and which frames should be tracked. Once final results are available, those files can be discarded.
Results/[individual video file] directories
These directories contain files called "JobOut". Those have the tracked position of the mice in the short interval of that job (along with classifier's data about head/tail and ID). These can be discarded once you have the final results.
Results/Tracks directory
This is where the [videofilename]_tracks.mat files are. These
track files
contain the trajectories of the individual mice.
Tuning directory
The Tuning directory contains Detection.mat and Identities.mat files, and a subdirectory for each single-mouse video.
The Detection.mat file contains information about how to do the segmentation. It can be discarded once the track file is finished.
The Identites.mat file contains the classifier that discriminates one mouse from another, and a classifier that determines which end of the mouse is the head and which is the tail. It also contains the example HOG (Histogram of Oriented Gradients) features for each frame of each single-mouse video. In addition it contains a struct called strctIdentityClassifier, with fields: m_astrctClassifiers, m_astrctClassifiersNegClass, and m_a3fRepImages.
The m_a3fRepImages field has representative images of the tracked mice, which can be useful in determining which track is which mouse, without having to compare the tracks to the original video (e.g. track 1 is the mouse with the spots, track 2 is the mouse with the horizontal stripes, etc).
In each single-mouse subdirectory, there is another file called Identities.mat. (Yes, it's confusing.) This Identities.mat file contains the output of motr on the single-mouse videos. In particular, it contains the position of the mouse in each frame, small image patches extracted at those locations, and the HOG features needed to build a multi-mouse classifier.