News

Center for Bio-Image Informatics (CBI) works with the Marine Science Institute at UCSB, NOAA and Greenpeace to analyze a large collection of underwater videos

April 3rd, 2012

CBI’s research featured in Washington Post article “Fishery managers could call for a review of massive underwater canyons”
photo of deep sea fish

In 2007 a research team surveyed the sea floor for three weeks using DeepWorker submersibles — small, single-pilot submarines equipped with high-definition video cameras, indexing lasers and robotic sampling arms — along with a remotely operated vehicle. In addition to identifying 15 species of coral and collecting 20 sponge species, scientists documented 13 instances of fishing impacts at depths ranging from 328 feet to 3,280 feet below sea level.

Researchers from UC-Santa Barbara’s Center for Bio-Image Informatics (CBI) later used software and algorithms to analyze 884GB of high definition video covering more than 23 hours acquired during the 16 deep water dives. Video analysis software then automatically selected video frames to estimate the sea surface area and capture the distribution of various visible objects. The Digital Notebook software, developed by Dmitry Fedorov at the Center, was then used to detect and classify the object/event types in the image frames into 56 different classes (for example, fish, corals, fishing disturbance, etc.).

The data was later used to extract co-occurrence statistics published in the PLoS ONE journal research article, “Structure-Forming Corals and Sponges and Their Use as Fish Habitat in Bering Sea Submarine Canyons”. The complete dataset of annotated frames is freely available through CBI’s Bisque image database.

UCSB Press Release

Center for Bio-image Informatics