RESOURCES

All tools are beta versions, tested extensively but with no guarentee for flawlessness.

Contact us for more details, comments or suggestions!

SignT

An easy-to-use GUI to perform contour tracking.

Uses movies from any common video file format to perform contour tracking, either RGB or grey levels (e.g. IR).

- Globally based on thresholding and simple mask operations as morphological opening/closing, that allow to deal with common problems such as cables hanging between the and the subject. Parameters can be adjusted manually while checking the result on the tracking performance, by replaying the movie or manually navigating through the frames

 - Allows to choose a specific channel in an RGB movie (useful to deal with certain reflections)

- Allows to process a background picture offline, either from automatically or manually selected frames (helpful when the animals remain a long time in the same spot)

- Allows to prepare tracking for several files, and then launch batch to process them together

   

Output: a single .mat file with the parameters used for the analysis, and for each frame:

- XY coordinates for the contour

- XY coordinates for the center of gravity

- A value for the motion of the animal: % of pixel changes from the previous frame (between the corresponding masks)

   

The current version has not been tested for compatibility and dependencies; in particular, it might not run under MATLAB versions older than R2018a.

   

Future implementations/changes:

- Expand the manual and explain the different functionalities

- Take into account potential differences in screen resolutions/sizes

- Use parallel processing for a single movie to increase speed

- Auto-detect if parallel computing toolbox is available and adapt batch-processing accordingly; same for MATLAB releases.

- Calibration (pixels <-> cm) to have absolute values

- Process speed as well (either absolute if calibration or pixel-based)

- Add "online" processing (display tracking while also saving the results)

- Add progress bar and/or infos in the command window

- Auto-resize plots to accomodate different ratios

- Adjust some parameters to get a smooth browsing (some movies have ridiculous FPS and resolution that are too heavy to handle like this)

- Add the possibility to compute a "moving" background

SignX

An easy-to-use GUI to extract immobility and freezing episodes

Uses the motion output from the tracking GUI and the original movie to extract freezing episodes and display the results. Parameters can then be adjusted to give the best results. The same parameters are usually usable for different animals in the same conditions.

 

- Uses a motion threshold to detect freezing episodes

- Allows to merge episodes closer than a defined gap

- Defines a minimum freezing duration

- Allows manual editing/removal of single episodes (e.g. sometimes to remove grooming detected as freezing)

 

  Output: a new variable in the .mat tracking file with the start/end times of each freezing episode

 

The current version has not been tested for compatibility and dependencies; in particular, it might not run under MATLAB versions older than R2018a. Also, unlike the tracking GUI, it needs the frames timestamps, that are currently either retrieved from the tracking file or a file generated by our acquisition system. This will soon be expanded to accommodate different systems.

 

Future implementations/changes:

- Expand the manual

- Switch from function to class for more robustness and ease of use on successive files

- Make the overall code less dependent on the recording system

- Use the calibration option from the tracking GUI to get an even more absolute freezing threshold (here it is influenced by the real pixel size, that depends on the camera distance)

- Take into account potential differences in screen resolutions/sizes

- Add inputs checking

- Adjust some parameters to get a smooth browsing (some movies have ridiculous FPS and resolution that are too heavy to handle like this)

- Add the possibility to compute a "moving" background

  • Facebook Clean Grey
  • Twitter Clean Grey
  • LinkedIn Clean Grey

© 2017 by Philip Tovote, last update 22/11/2019