Identifying Dolphins with Technology
Source Newsroom: Dick Jones Communications
Newswise — Dolphins all look pretty similar. So it can be problematic when your job requires you to identify individual dolphins in order to study their behavioral and ecological patterns.
Photo-identification techniques – recognizing a particular dolphin by the nicks, scars and notches on its dorsal fin -- are useful, but tedious.
“Researchers photograph dolphins in their natural surroundings and compare new dorsal fin photographs against a catalogue of previously identified dolphins,” explains Kelly Debure, professor of computer science at Eckerd College in St. Petersburg, Florida. “These catalogs are often organized into categories based on either distinct fin shape or location of predominant damage. The manual photo-identification process, although effective, is extremely time consuming and visually stressful, particularly with large collections of known dolphins.”
It was time to bring dolphin identification into the digital age.
Debure, along with Eckerd students, developed DARWIN, or Digital Analysis and Recognition of Whale Images on a Network, a computer program that simplifies photo-identification of bottlenose dolphins by applying computer vision and signal processing techniques to automate much of the tedious manual photo-id process.
“DARWIN is a software system which has been developed to support the creation of reliable and intuitive image database queries using fin outlines,” she says. “It effectively performs registration of image data to compensate for the fact that the photographs are taken from different angles and distances and compares digital images of new dorsal fins with a database of previously identified fins.”
The software uses an automated process to create a tracing of the fin outline, which is then used to formulate a sketch-based query of the database. The system utilizes a variety of image processing and computer vision algorithms to perform the matching process that identifies those previously cataloged fins which most closely resemble the unknown fin. The program ranks catalog fin images from “most like” to “least like” the new unknown fin image and presents images for side by side comparison.
DARWIN is used by researchers at several academic institutions and by Eckerd College’s own Dolphin Project, a team of students who conduct population surveys of the bottlenose dolphin (Tursiops truncatus). Initiated in 1993, the project has trained dozens of students to take and analyze scientific data on dolphin populations to better understand their population dynamics and ecology in Tampa Bay. Such information can be used to help conserve dolphin populations.
The DARWIN software is free and available for download. Over 20 years, Debure and Eckerd students have continued to refine the software and are adapting the software's algorithms to make it appropriate for identification of other species.
“Although it was originally developed for use with bottlenose dolphins, it has been used by research groups on related species such as fin whales, indo-pacific humpback dolphins, spinner dolphins, and basking sharks,” she says.
With this technological help, researchers can spend more time doing their job.
“Answering the question, “Which dolphin is that?” is not that scientifically interesting,” says Debure. “With DARWIN, researchers can spend less time identifying animals and more time doing the real science.”