Post provided by Violette Chiara and Sin-Yeon Kim
There are many tools available to track and analyse animal movement. In this post, Violette Chiara and Sin-Yeon Kim share insights on how AnimalTA fills the niche of an accessible, user-friendly software for video analysis.
Why AnimalTA?
AnimalTA is a new tool to track and analyse animal movement in digital videos. When we began to develop AnimalTA, most of our colleagues were surprised and did not understand why – because a lot of good animal-tracking programs already exist. What more does our program offer that others don’t? The truth is that AnimalTA does not propose any new technology or algorithm, and cannot be considered an improvement in our knowledge of computer vision. However, we believe that this program offers something equally important: its accessibility.
While a new program usually proposes something more advanced and complex than its predecessors, we noticed this can exclude important parts of the scientific community who do not have time or resources to invest in accessing these new tools. Nowadays, the majority of the tracking programs – even the ones claiming to be “user-friendly”- are difficult for people with little or no experience in programming (or programming environments) to operate.
Some may argue the scientific community is much more familiar with programming than it was a few years ago, mainly due to the popularization of tools such as R or MATLAB. Although this is true, we know through experience that many researchers still get frustrated when they spend lots of time searching for a free video tracking program, spend even more time learning how to install and use the software, only to find that it does not work with their videos.
In fact, many researchers give up midway through the process, and either end up using expensive alternatives (if they have resources) or analyse their videos manually without using a video tracking program. This is the reason we believe that a simple, flexible and user-friendly video tracking program like AnimalTA is needed.
Addressing research hurdles
We designed AnimalTA for the purpose of creating a video tracking program that prioritizes assisting scientists in their daily work, over introducing flashy new technology. Throughout its development, we aimed to address common difficulties and frustrations that users may encounter when tracking their videos. Therefore, AnimalTA offers more than just a simple video tracking program: it combines tools for project management, video editing, video improvements, tracking correction, and analyses of movement trajectories.
We identified problems that researchers often encounter while recording videos in their experiments and using them for video-tracking, and then proposed solutions. For instance, all the raw videos can be loaded into the program directly. There is no need for a prior preparation, such as conversion, cropping or editing: you can do these directly in AnimalTA. An unlimited number of videos can be stored in the same project, allowing you to have a single file for your whole experiment and to easily apply the same tracking parameters for all your videos without the need for repetitive operations.
We also thought about common technical difficulties of recording video. For instance, it is not uncommon to experience camera tremors, which would either disrupt tracking with most other programs or bias the tracking results. AnimalTA provides a stabilization tool that will suppress such tremors. Changes in lighting conditions, due to the presence of shadows or changes in the sun light, are another commonly encountered problem that AnimalTA can fix with a lighting correction tool.
In some experiments, it may be impossible to place the camera above the experimental arenas, and this problem induces an effect of perspective and biases the analyses of the trajectories. A similar problem may occur when there is image deformation in video records. These problems can be corrected with perspective correction function provided by AnimalTA. The heterogenous backgrounds are also not a problem with our program as it proposes a system of background subtraction. Each video can be (re)scaled, allowing you to work in international units, instead of pixels, from the beginning and obtain more understandable data. Furthermore, AnimalTA provides multiple tools for tracking correction and data analyses.
Correcting the trajectories
Despite all precautions and video correction options, some tracking errors can still occur. Correcting the tracking data is often difficult and time consuming. We provided our users with a wide set of correction tools, which can fix different kinds of tracking errors in fast and efficient ways.
First, the visualisation of tracked animals is easy and straightforward, allowing you to quickly check the quality of tracking. It is possible, for instance, to play videos at an accelerated pace with the tracked position of the targets, to show the entire trajectory of the target, or to jump directly to the frames for which the targets were lost.
Once you identified portions of videos with tracking errors, you have various options to correct them. You can correct by hand the position of the targets by clicking directly on the video. You can also interpolate the trajectory between two frames, in case of straight-line movements, or select the problematic portion of the video and re-run a tracking with different parameters for this portion. This last option is, to our knowledge, a novelty in video-tracking programs, which will highly facilitate the work of our users.
Extracting data
AnimalTA also integrates multiple tools for data extraction. We tried to provide useful tools for researchers, especially in the fields of ethology or behavioural ecology. One of our objectives was to make the program accessible and understandable for everyone.
Thanks to direct visual feedback, the effect of any change in the analysis parameters will be immediately illustrated. We believe that this feature allows the users to see different analysis options and enhances the program’s practicality. AnimalTA allows analyses of movements (speed, meander, distance travelled, proportion of time moving, etc.), analyses of exploration, analyses of social interactions, and analyses of interactions between the targets and elements/areas of interest. You can, for instance, draw a circle in the middle of an arena to know how much time your target spent in this circle, or draw a segment to know how many times this segment was crossed by the target. Once you perform all the analyses you are interested in, AnimalTA will provide you with summary tables that can be opened in Excel, Libre office or text editors.
Brief overview of some analysis proposed by AnimalTA. From left to right: i) fishes in two-way choice tests, the program will measure the time spent in red coloured arenas and the number of times purple segments are crossed. ii) Ants in food choice experiment, the program will measure time spent in food patches (yellow-coloured shapes) and close to the arenas’ borders. iii) A vacuum robot cleaning a kitchen, the program will calculate the proportion of the kitchen that has been explored by the robot.
AnimalTA for everyone
Because we wanted our program to be as accessible as possible, especially for students, AnimalTA can be used in 5 different languages, including Mandarin, English and Spanish.
The program allows creation of videos tracking process with multiple options. For example, showing the identities of the targets and their trajectories in different colours, and the user can choose a specific step of tracking to be shown in the videos. We believe this will help researchers or other users to present their results in scientific conferences or lectures.
You can read more in the full article:
Methods Blog 