As part of the Video Annotation and Reference System (VARS) project, a software tool, called VARS Query, was developed to give users access to information in MBARI’s video annotation database. MBARI uses high-resolution video equipment to record hundreds of operated vehicle (ROV) dives per year. As of 2015, there are over 21,000 hours of video recorded and over 4.5 million annotations about those video observations. With the VARS Query, users can locate video sequences or access other information recorded during ROV research dives. Complex queries can be made by constraining temporal, spatial, or physical parameters (for example, season, location, or depth). The VARS Query references a knowledge base of thousands of biological, geological, and technical terms, providing consistent hierarchical information about objects and associated descriptions for annotating video at sea or on shore.

VARS has been tested and runs on Windows 7, Mac OS X, and Linux. It should run on any platform with Java Runtime Environment, version 1.8 or later.

Download links and instructions for using VARS are accessible within the sidebar links on the left.

 

Yes, you should click on Start. VARS needs access to your local machine and network to store query results and configurations on your hard drive and connect to the query server and frame grab archive. Java’s security model warns you about this security issue so that you don’t accidentally install malicious programs without your knowledge. The VARS Query is not a malicious program. Depending upon your computer operating system and JAVA version, there may be additional security settings that may need to be configured for VARS to run correctly. These are addressed in the installation section of each VARS component.

The video annotation database contains over four million annotations. The query application cannot retrieve all of them at once. Try hitting Cancel and constraining your search to a few days, months, or years (depending on how frequently your chosen concept appears in ROV dive video). If you’re a new user who simply wants to try out the VARS Query, do your first search on something rare (in MBARI ROV dive videos) like pacific-viperfish, vampire-squid, basaltic-lava, or grimpoteuthis.

Nil represents a wild card and tells VARS to include everything in that field. Running a search on the NIL concept means you want to see records of all concepts in the database. You’ll get every annotation record except for what is filtered out by your specified Constraints. To avoid getting tied up by an overwhelming number of hits, we recommend users only do NIL concept searches on highly constrained datasets like individual dives, specific dates, or narrowly defined depth or geographic ranges.

The query opens with some default settings for columns and no constraints. After specifying a basic search, click on the Advanced tab, make selections with the returns and constrain checkboxes, and fill in the details/limits. See instructions on the Query page.

If the return checkbox under Miscellaneous-Image on the Advanced tab is checked (a default selection), an Image column will be returned in the Results table. An entry in the Image column indicates that a frame grab was taken with that annotation. Double-click on the image name and the image will appear in a pop-up window. It may take several seconds to load the image. See the VARS Query Results page for instructions on how to save single or multiple images.

VARS allows you to export the query results data table as a tab-delimited text file or as a KML, which can be used for visualizing the observations spatially within Google Earth, ArcGIS, and other geospatial analysis applications. You also have the option of saving all of the images that are available for this query.