Save the whales! Really, using Big Data
December 03, 2015
With only about 500 North Atlantic right whales still on the planet, there's an urgent need to rescue the remaining population from the brink of extin...
With only about 500 North Atlantic right whales still on the planet, there’s an urgent need to rescue the remaining population from the brink of extinction. Marine biologists at the National Oceanic and Atmospheric Administration (NOAA) and their partners have been working to address the threats right whales face to preserve and grow the remaining population. To do so, they need great scientific information. Tagging right whales, tracking their migratory and feeding patterns, and keeping diligent health records all rely on the ability to identify individual whales.
NOAA has made great strides over the past 10 years, helping the right whale population increase by nearly 10 percent. However, the current identification process is labor intensive, costly, largely manual, and requires specialized training.
Using the current methodology to track right whales, scientists typically take photographs of the whales from a boat or plane. Once back on land, they manually catalogue each whale by the unique markings on its head, called callosities. While this process of tracking and identification has helped their efforts to restore a portion of the population, NOAA is hoping to automate the process, allowing them more time in the field.
NOAA realized that advances in image classification technology, through techniques such as machine learning, computer vision, and deep learning, would help automate the process of right whale identification and partnered with MathWorks and Kaggle for the Right Whale Recognition Contest – Creating a “Face Detector” for Whales.
The contest currently has 148 teams racing to develop a winning algorithm that will be used as the basis for automated image identification. The algorithm will help NOAA create a software application and build a database that scientists can use to track the health history, feeding, and migratory habits of individual whales in near-real time.
NOAA fisheries biologist Christin Khan is leading the effort to apply machine learning as a way to automate the right whale identification process. Khan is working with MathWorks, whose MATLAB numerical computing environment serves as a foundational analytics tool.
Khan says that by tapping the power of the Kaggle community and image processing, computer vision, and machine learning, they can focus on the conservation efforts. MATLAB and similar tools will help automate the identification process, saving critical research time and resources.
As the sponsor of the competition, MathWorks is giving participating teams free access to MATLAB, as well as several of its toolboxes: Statistics and Machine Learning Toolbox, Computer Vision System Toolbox, Neural Network Toolbox, Optimization Toolbox, Global Optimization Toolbox, Curve Fitting Toolbox, Econometrics Toolbox, Financial Toolbox, Image Processing Toolbox, and Parallel Computing Toolbox. In addition to the software, MathWorks is providing training and access to MathWorks engineering mentors and technical support.
Ultimately, the winning algorithm will help NOAA identify right whales faster and with greater accuracy, monitor sick or injured animals more closely, provide health checks, and assist with satellite tagging. If successfully applied to the right whale population, NOAA expects that it will be able to extend the identification technique and use of machine learning to protect other threatened and endangered marine species.
Paul Pilotte, technical marketing manager for MathWorks, focuses on MATLAB toolboxes for statistics, optimization, symbolic math, and computational finance. He has more than 20 years of experience in technical marketing and development in technical computing, security software, data communications, and test-equipment markets. He holds bachelors and masters degrees in electrical engineering from MIT and an MBA from Babson College.
Christin Khan received a Bachelor of Science degree in Biology from Northeastern University in 1998, and a Master of Science degree in Biology from San Francisco State University in 2004. She has extensive aerial survey experience, and has been identifying marine mammals and turtles from the air since 2005. She has held aerial survey management positions including Team Leader for Wildlife Trust, and Flight Coordinator for the Provincetown Center for Coastal Studies. She is currently a fishery biologist at NOAA Fisheries in Woods Hole working on a variety of projects including right whale aerial surveys, an analysis of right whale social behavior, an outreach campaign to install right whale awareness signs at marinas and ports, and using innovative approaches to tackling big data problems like right whale facial recognition.