In 1974, a grocery item with the now universal black and white barcode was scanned for the first time. This invention streamlined the checkout process and has forever transformed how we shop.
Now the development of an innovative image analysis program that uses an animal’s distinctive markings as a natural barcode is transforming data collection for wildlife conservation efforts.
The Image Based Ecological Information System (IBEIS) was developed by researchers at Princeton University, Rensselaer Polytechnic Institute, the University of Illinois-Chicago, and the non-profit Wild Me. Using computer vision algorithms similar to those used in facial recognition software, IBEIS identifies animals based on “hot spots” created by their markings or the shape of their fin or fluke.
The system was initially designed to identify zebras based on their unique stripe patterns. It has since been further developed to identify any species of animal with distinctive markings, wrinkles or notches. For example, manta rays have unique patterns of spots on their bellies and whale’s tails contain unique notches.
In an e-mail to Mongabay-Wildtech, Dr Tanya Berger-Wolf, a researcher from the University of Illinois-Chicago and one of the developers of IBEIS, explained how this “hot spot” technology works.
“IBEIS uses local variations in the pixel values of an image of an individual to calculate ‘hot spots’ of high change in those values, [which means there is a pattern of dark and light]. The hot spots are unique to each animal and become the signature of that animal. The hot spots of a new picture of an animal are matched with the hot spots of existing images, and a similarity score is computed based on the goodness of that match.”
All matches made using IBEIS are then reviewed by a trained individual. Additional information, such as age, sex and when/where the animal was photographed, can then be added to the animal’s profile in an open-source software framework called Wildbook.
Wildbook was developed by Wild Me in 2003 to help track individual whale sharks. Now, with IBEIS as its image analysis engine, Wildbook is used by research institutions around the world to study animals ranging from nurse sharks and manta rays to cetaceans and polar bears.
Once an animal has been identified using IBEIS, the Wildbook platform allows researchers and other interested people to follow the life of the animal (ie where it likes to hang out and what individuals it likes to hang out with), just as social media networks like Facebook give us insight into our friends’ social lives. According to Berger-Wolf, Wildbook “allows non-invasive, scalable, cheap, wide coverage tracking and counting of animals, as well as getting information about their intra-species [social] and inter-species [other species] interactions.”
This ability to easily identify individual animals has proven to be an especially useful tool for carrying out comprehensive population censuses. An accurate estimate of the number of individuals in a population is key to determining the health of the population and, subsequently, the IUCN conservation status for that species. For threatened or endangered species, these population censuses help inform management decisions to minimize the risk of extinction.
Collecting data from images is non-invasive and requires no special technology beyond a digital camera. However, carrying out a comprehensive population survey with IBEIS requires taking thousands of images, sometimes over a huge area. Early on, researchers using IBEIS realized that this obstacle could be overcome by recruiting the help of citizen scientists.
The power of citizen science
Two population census events carried out in Kenya have helped to prove that, equipped with only digital cameras, ordinary citizens of all ages can contribute substantively to the collection of images to be used as data for scientific studies.
In January 2016, IBEIS was used to carry out a full census of the Grevy’s zebra, one of Kenya’s most endangered species. The event engaged the help of hundreds of volunteers, including school children, tribal chiefs and the US ambassador to Kenya.
Over two days, these volunteers used GPS-enabled digital cameras to capture over 40,000 images of the zebras at 45 locations in Kenya’s Rift Valley. Using IBEIS, researchers identified which individual zebras were resighted on the second day of the survey. Based on this Sight-Resight census, they estimated that the Grevy’s population was stable but contained just 2,350 individuals – a sharp decline from the more than 15,000 found in Kenya in the 1970s.
In March 2016, a similar event called the Kid’s Twiga Tally was organized by the NGO WildlifeDirect, the Mpala Research Centre, the Laikipia Wildlife Forum and scientists from Princeton and Columbia Universities. Twiga is the Swahili word for giraffe. In Laikipia county, Kenya, children aged 10 to 13 from eight primary schools spent a day capturing over 1,300 images of reticulated giraffes. Their photo data provided scientists with valuable information on the structure of giraffe populations in differently managed areas.
According to Berger-Wolf, two advantages IBEIS has over other data collection methods are its ability to use images to identify individual animals and the ease with which citizen scientists can collect the data. She added, “the system allows wide participation of citizen scientists, casual nature lovers and curious people in the process of science and conservation just by taking a picture.”
Wild Me encourages contributions by supporting citizen science projects that could benefit from Wildbook’s image analysis capabilities and user-friendly interface. Nevertheless, in an e-mail to Mongabay-Wildtech, Berger-Wolf acknowledged that developing Wildbook for different species comes at a cost.
“There is a huge need and demand,” she explained. “We have requests from more than 100 organizations to create a Wildbook system for them, but each new species needs both training, the algorithms, customizing the data architecture, creating an instance of the system on the cloud, and training the people who will use it.
Just like the first genome, it is expensive. The price will go down, but we need the funding to do the first several. Conservation organizations are not used to funding tech tools and have few sources of funding for something like this.”
Another obstacle identified by Berger-Wolf for using Wildbook in field studies is infrastructure. The team will need to develop new offline technology to provide access to the platform in remote areas with poor internet access. For endangered species that are valuable to poachers, security is also a major concern. “The information from Wildbook in the wrong hands can lead to the death of the animals,” said Berger-Wolf.
A growing social network
For citizen scientists, Wildbook provides the opportunity to contribute to scientific studies and, ultimately, policy and management decisions. The Wildbook software also allows any curious person to take an image of animal, in the wild or in captivity, and learn more about that animal’s background.
Currently, various projects use the Wildbook platform to facilitate citizen science contributions. The platform provides a uniform and user-friendly process to contribute to each of these projects.
All a citizen scientist has to do is capture a clear photo of the target area for that species (i.e. manta ray’s belly or shark’s flank) and submit the photo to that species’ Wildbook page, along with any additional information such as the sex of the animal, where it was found and what it was doing at the time.
This information is then reviewed by a researcher, and the IBEIS program is used to determine whether the image matches an individual already found in the database or represents a new sighting.
Wildbook provides a free online platform for researchers to store and organize data on individual animals gathered using IBEIS and/or biological and chemical information obtained from tissue samples. Users can use these data to study species distributions and movement patterns, as well as social and behavioral ecology questions.
They can also upload other data sets, such as climate patterns, land-use change or agricultural development. Combining such data types allows researchers to delve into more complex questions such as how a population is affected by climate change or how harmful human-wildlife interactions can be minimized in an area.
Berger-Wolf envisions a social network that will continue to harness the power of citizen science to help conserve wildlife at a global scale.
“Ideally, Wildbook will be the system of choice, like Facebook, for scientists and conservation managers, as well as the general public, to study and protect entire species or regions, at the planetary scale, but at the resolution of an individual animal. A system like Wildbook allows anybody to engage with wildlife, creating a sense of ownership, making it personal.
By taking a picture you can learn about the giraffe you are seeing either in the wild or at the zoo. Or follow an animal’s story through the images that others have taken. A living breathing social data fabric with an animal at the center”.
Mongabay seeks to raise interest in and appreciation of nature and wildlife, while examining the impact of emerging trends in climate, technology, economics, and finance on conservation and development.