Ultimate Tracker is a simple and intuitive tracking software that measures animal's movements in open field as well as in other experimental setups such as the 3-chamber or dark/light room, where it measures the time spent in each area and the number of entries and exits. Ultimate Tracker also analyzes the animal's interactions with objects.
Videos are played using the OpenCV library. Supported formats are .avi, .mp4, and .mkv.
The name of the video is displayed next to the [Load video] button, and information is displayed on the screen when you hover over this text field.
! It is important that the frame size corresponds to a standard ratio, otherwise the image will be displayed incorrectly: be careful with cropping!
The frames are loaded by a thread into a “buffer” list, from which they are read for display. (The size of the buffer is calculated based on the amount of RAM available on the machine).
Actions on the video:
- Move around the display window with the left (or middle) mouse button.
- Zoom with the mouse wheel.
- Play/pause, fast forward with the cursor, advance frame by frame with the [] buttons.
- Play every other frame or the value displayed in the “each n frame” field.
- The frame size can be reduced on display according to the value indicated in the “video scale” field.
Several analysis modes are available, allowing the user to use presets and to define how the data is analyzed.
- Open field: this involves defining an area in which the target of interest is tracked to measure the amount of movement
- Open field centered: a rectangular area is automatically defined in the center of the area of interest so that the “central” and “peripheral” areas are the same area. The distance traveled is measured separately in each central and peripheral area
- Open field quadrant: The main area is divided into four identical areas by its center.
- Open field objects: possibility of defining areas of interaction with objects; for color images, the contrasts of the scene are attenuated for maximum contrast with the animal (conversion of the image to grayscale by maximum projection of intensities).
- Open field interaction
- Dark Light box (red)
- Three chambers
*For the “centered, quadrant, objects” modes, the analysis method is the same; it is the predefined rules or the way the image is converted to grayscale that change.
- CPP 4 boxes
- Cuttlefish (multi)
Once the type of experiment has been selected, the main frame for selecting the analysis area is displayed on the image.
Two tracking methods are available:
- Threshold: this involves applying a threshold to the image, whether pre-processed or not (this is the purpose of the options below), to binarize the image and extract the target of interest.
The threshold can be set manually and modified by the user in the “Settings” tab.
The threshold can be automatic and varies depending on the image content.
The threshold can be determined automatically locally based on the target's immediate environment.
- Feature tracker (CSRT): the target is tracked using a learning algorithm based on a history that is updated during analysis. This method is less accurate due to the inaccuracies in centering the tracking box on the center of the target, but it adapts more easily to the target's movements and changes in shape.
An essential step in segmentation is converting the color image to grayscale.
At present, it is not possible to select a specific color for tracking, except in “CSRT feature detection” mode.
For color images, in order to perform effective threshold segmentation, it is necessary to obtain the best contrast between the target and the background. It is therefore possible to select one of the channels composing the image in the Lab space.
L: only brightness values.
a: red values are lighter and green values are darker.
b: blue values are lighter and yellow values are darker.
brightest: the three channels are projected in maximum mode, and the lightest values are retained.
For grayscale images, the L option is the only option available.
To further improve the quality of the segmentation, it is possible to perform a background subtraction to isolate the target, which is mobile. To do this, you must either load an image of the scene that does not include the target to be tracked and is the same size as the video frames, or generate a background from images extracted from the video. A median image is then calculated to eliminate moving animals.
The blob-detector option is only used for multi-animal tracking. Validate the selected options with the [Method validation] button.
(Analysis tab)
To draw a region of interest, activate the corresponding button, select the shape of the area (Rectangle or Circle), and left-click on the image to define a starting point. Hold and move the mouse to the diagonally opposite end to draw a frame encompassing the area of interest. Release the mouse button to finish.
ROI exclusion: this involves masking an area that would interfere with image analysis.
Object ROI: if the study focuses on the animal's interaction with an object or another animal (contained within a restricted area), this option allows you to define a frame around which an interaction area can be extended. By checking the “Inside mask” box, a mask is applied to the center of the ROI.
Target ROI: this is the frame in which the target is identified. This area of the image is cut out and only this area is analyzed to define the position of the target, whose coordinates are repositioned in the original full image.
Analysis ROI: only used in multi-track mode. Each analysis area is defined by an analysis ROI. (*The main ROI acts as the Target ROI in which the image is analyzed using the blob detector.)