Skip to content

Latest commit

 

History

History
95 lines (48 loc) · 8.8 KB

advanced_labelling.md

File metadata and controls

95 lines (48 loc) · 8.8 KB
Raw

SimBA behavioral Annotator GUI (ADVANCED LABELLING)

The SimBA behavioural annotator GUI is used to label (annotate) frames in videos that contain behaviors of interest. SimBA allows three different ways of annotating behaviors as present or absent in individual frames. The difference between these methods is how non user-annoated frames are treated

  • (1) LABEL BEHAVIOR: When selecting a new video to annotate, SimBA assumes that the behavior is absent in any given frame unless indicated by the user. In other words, the default annotation is that the behavior(s) are not present.
  • (2) PSEUDO-LABELLING: When selecting a new video to annotate, SimBA uses machine classifications the default annotation. Thus, any frame with a classification probability above the user-specified threshold will have behavior present as the default value.
  • (3) ADVANCED LABEL BEHAVIOR. When selecting a new video to annotate, SimBA has no default annotatation for any frame. In other words, the user is required annotate each frame as either behavior-absent or behavior-present. Only annotated frames will be used when creating the machine learning model(s).

This tutoral details how to use the ADVANCED LABEL BEHAVIOR annotator in SimBA. To use advanced labelling in SimBA, we recommend first reading the tutorial on how to use the standard LABEL BEHAVIOR / PSUDO-LABELLING. The standard LABEL BEHAVIOR / PSUDO-LABELLING tutoral documents the functions of the buttons in the SimBA annotator GUI, which are identical when using advanced labelling in SimBA.

Note: For information on how to append annotations created in alternative third-party software outside of SimBA, click HERE

Step 1. Loading project_config file

In the main SimBA menu, click on File > Load Project > Load Project.ini > Browse File and select the config file (project_config.ini) representing your SimBA project. This step must be done before proceeding to the next step.

Step 2. Opening the labelling behavior interface

Once your project is loaded, click on the [Label Behavior] tab and you should see the below four sub-menus (Note: I'm writing this document on a Mac, if you're running SimBA on a PC or Linux, the aestetics might be slightly different):

These four different sub-menus represent four different ways of getting human annotations appended to your video data. In this tutorial, we will use the buttons in sub-menu number 3. We will click on Select video (create new video annotation) in the ADVANCED BEHAVIOR LABEL submenu. This will bring up a file selection dialog menu. We navigate to the project_folder/videos directory and select a video we wich to annotate in SimBA. In this tutorial, I am selecting Together_1.mp4 and click Open:

Step 3. Using the advanced labelling behavior interface

Once I've selected my video file, the annotation interface will pop open, looking something like the below image. The different buttons have identical functions to when using standard LABEL BEHAVIOR annotator or PSEUDO-LABELLING. For detailed instructions of their functions, check out the standard label behavior tutorial.

In this tutorial I am annotating a video within a SimBA project that has two classifiers. We can see this in the Check behavior submenu, which is populated with two rows for the two classifiers (Attack and Sniffing) and two columns for annotating if the behavior happening, or not happening, in the frame (PRESENT and ABSENT).

In the below gif, I am viewing frame number 28 of a video called Together_1 (Note: you can see the name of the video currently beeing annotated by looking in the GUI annotator title window). In frame number 28, I annotate Sniffing as beeing present, and Attack as being absent. My annotations are saved into SimBA working memory when I click to navigate to a different frame.

As opposed to PSEDO-LABELLING or standard LABELING BEHAVIOR, within advanced labelling, it is possible to omit a frame from being annotated and omitted from ant downstream machine learning processes. This means that we score the behavior(s) as neither PRESENT or ABSENT in the frame (or a range of frames). For example, if we do not tick any behaviors as either absent of present (leave all classifier checkboxes unticked) for a frame (or range of frames), that means that the specific frame (or range of frames) will be omitted as annotated examples for any downstream processes involving machine learning predictive classifiers. In the gif below, I use the label frame range tool in the annotator to label frames 50-100 as containing no suitable examples of Sniffing and Attack as being present or absent.

IMPORTANT: If you are working on annotating a video for multiple classifiers (just as in the current tutorial, where we are annotating both Attack and Sniffing), then we cannot annotate some behaviors as PRESENT or ABSENT, while we annotate other behaviors as neither PRESENT or ABSENT in the same frame. For example, in the below gif I annotate Sniffing as present in frame number 35 and I don't tick Attack as either PRESENT or ABSENT. In this scenario I will be presented with an error message, and asked to either (i) annotate all behaviors as neither PRESENT or ABSENT (remove all ticks from all tick-boxes), or (ii) provide ticks in tick-boxes representing all behaviors. In this below gif example, I go for the latter solution, and tick Attack as absent.

Once you have completed your annotations, we need to save all your annotations into your SimBA project. We do this by clicking the Save Annotations button, and this is a very important button. This buttons saves your annotations into your SimBA project, which you are required to do in order to use the annotations for creating machine learning models. Clicking this buttons saves a data file inside your project_folder/csv/targets_inserted directory. This file will contain all of the body-part coordinates and features in seperate columns, plus a few additional columns at the end (one for each behavior that you are annotating) with the headers that represent the behavior names. Hence, clicking this button, in this tutoral, will generate a file inside the project_folder/csv/targets_inserted directory called Together_1 (which is the name of the video I am annotating). The last two columns in this file will be named Attack and Sniffing. These two column will be filed with 1 and 0s - a 1 for every frame where I noted the behavior to be present, and a 0 for every frame where I note the behavior to be absent. This file will contain as many rows as the number of frames where I annotated the behaviors as PRESENT or ABSENT`.

CONTINOUING PREVIOUSLY STARTED ANNOTATIONS

After clicking Save Annotations and closing the SimBA annotation GUI, you may want to come back and continue annotating the same video using the advanced SimBA annotation GUI. To do this, click the Select video (continue existing video annotation) button in the ADVANCED LABEL BEHAVIOR sub-menu:

SimBA will then look for the file representing your video inside the project_folder/csv/targets_inserted directory. SimBA will also look for the file representing your video inside the project_folder/csv/features_extracted directory in order to de-code how many frames you have provided neither PRESENT or ABSENT annotations for. It will next bring up the SimBA advanced annotation GUI at the frame number where you last clicked Save Annotations.

Author Simon N