Scientists Make Facebook For Whales

Facebook has helped millions of Americans waste time at work, share drunken photos, and stalk old high school crushes, but now scientists are hoping to use the software to save endangered whales.

Using Facebook like facial recognition software, scientists are creating a comprehensive photo database of the world’s 500 North Atlantic right whales in an effort to save the species.

Marine biologists teamed with computer programmers to create algorithms capable of identifying individual right-whales from photos taken during overhead flights, according to The Atlantic.

The whales even have names: whale 1611 is Clover, whale 1006 is Quasimodo and whale 1250 is Herb.

The Atlantic magazine chronicled the right-whale recognition project in a recent article by Elizabeth Preston.

The project started when Christin Khan from the National Oceanic and Atmospheric Administration (NOAA) teamed with the online data-science competition platform Kaggle.

Kaggle has already helped scientists at Cornell University create software to detect right-whales in audio recordings.

Khan’s goal for the project was for programmers to create software that would immediately identify individual right-whales from overhead photos thereby freeing up the biologists to spend their time in more useful ways.

Ordinarily, researchers would fly over the whales in a small utility aircraft and lean out the back to take photos. Then, the scientists would return to the office and page through a massive online photo index maintained by the New England Aquarium to identify the whales in question, a project that could take hours.

Khan hoped to create software that could easily and quickly recognize individual whales, which would cut research time and help biologists track the whales. If researchers found they were tracking the same whale over and over, they could move on and leave that animal alone.

With a $10,000 prize from MathWorks up for grabs, 470 players joined 364 teams during a five month competition starting in August to create whale recognition software that would recognize the animals based on the whitish markings on their heads called callosities.

A team from the company DeepSense.io in Warsaw, Poland, won the competition this month with software that was capable of recognizing 87 percent of photographed whales.

The first step was to create the whale’s passport photo by aligning the blowhole on one side and the bonnet or head on the other. Then, using a neural network computer learning program, the team trained the software to search for patterns.

Now, marine biologists can use the software to track the right-whales and monitor their movements, assist in medical diagnosis and allow scientists to spend their time in more useful ways.

It will also help volunteers save whales entangled in fishing lines and nets by identifying the animal and using that information to help determine how much gear they’re dragging and how it’s attached.

The next step will be to evaluate the software in the marine biologist community and decide if the software should be sold as-is. There’s also the possibility of packaging it with other programs that would allow researchers to identify the right-whales from pictures taken alongside the animals in research boats.

Right-whales are slow moving creatures that skim the surface for food and swim along the coast; habits that made them excellent targets for whale hunters who drove the marine mammals to the brink of extinction.

They are among the most endangered creatures in the world.

[Photo by Cameron Spencer/Getty Images]