Once video footage from a benthic video survey has been collected, it must be viewed and annotated. The annotation process consists of coding each second of raw video for types and percentage composition of substrate, species of fauna presence and their abundances, and species of flora presence and their percentage coverage.
There are two main approaches to video annotation - frame annotation and timeline annotation. Depending on the particular type of analyses that needs to be performed, Ocean Ecology uses both frame annotation (e.g., using VARS [Video Annotation and Reference System] developed by the Monterey Bay Aquarium Research Institute [MBARI]) and timeline annotation (e.g., using Anvil developed by Michael Kipp of the German Research Center for Artificial Intelligence [DFKI]).
Frame Annotation
In frame annotation, each frame of the video can be annotated individually. This generates a series of unique data points, one for each video frame. This is very useful for statistical analyses, but can be very tedious and repetitous from the annotation perspective, as often a single annotation event (e.g., the presence of a particular patch of algae or the movement of an individual fish) may span a number of frames.
Shown below is an example of a video being annotated using frame annotation.
Timeline Annotation
In timeline annotation, segments of the video timeline can be annotated for each "event". This generates a start and end time for each observed event. This type of annotation is much more rapid than frame annotation; however, the type of data generated by timeline annotation is much more difficult to analyze statistically, and must be transformed into point data for most statistical work.
Shown below is an example of a video being annotated using timeline annotation.
Annotation Analyses
All annotation data are entered into a relational database, either Microsoft Access or Apache Derby. The database can then be queried for specific data records, or used for further statistical analyses on the data. Data from the database can also be imported into GIS applications for mapping.