The silence of the office is shattered by the clickety-clack of keys. In the corner of the room, computer monitors display a blur of code, data and images.
To a layperson, this might seem to be the work of an average computer programmer. But behind each keystroke and line of code is the beginning of a new process that could revolutionize conservation efforts for years to come.
The office belongs to BYU computer science professor Ryan Farrell. Both a professor and a researcher, Farrell’s work in visual categorization serves to combine technology and conservation into one easy-to-manage platform.
This platform contains an algorithm that can identify insect species using only photos, which cuts down on much of the confusion associated with species identification and has applications in several different fields.
“The great part about Professor Farrell’s research is the way it brings together disparate fields,” said Seth Bybee, Farrell’s research partner and BYU biology professor. “It is not always easy to rapidly identify species in my field, and a technology that has the ability to organize animals from mere pictures is very powerful.”
A BYU professor since 2013, Farrell teaches upper division computer science courses each semester. However, much of his time is dedicated to pursuing his passion for computer vision research.
“Something that I loved during my own studies was these ideas of computer vision and computer recognition,” Farrell said. “Then right after I got married, I got interested in birding as a hobby. Whenever I travel, I like to take some time and see what the local bird life is like, and during grad school, I realized I could actually combine that interest with birds with a very challenging computer vision problem.”
Farrell started developing algorithms that could recognize different species using photographs. This research evolved as Farrell continued his studies and became his emphasis as a professor.
Farrell’s research has been so innovative that he recently earned a Faculty Early Career Development (Career) Award — the National Science Foundation’s most prestigious award for junior faculty.
“Receiving this grant is a special thing,” Bybee said. “It really speaks to the value of this project and Farrell’s status as a special thinker.”
Farrell said he will continue developing his algorithms and image libraries using this grant.
His completed project will allow for smartphone cameras to identify the insect species. However, this is not the only application that Farrell’s research can have.
Farrell’s research has broad ecological implications because third-world countries can use this technology to document various species and plant life at a low cost, according to Bybee. He said species count is a major factor in securing resources for conservation efforts, so a low-cost method provides countries and organizations with an easy way to continue ecological work.
“I chose to focus on insects in part because this domain has not been explored in the computer vision community,” Farrell said. “But besides that, these techniques that I am going to be applying have a wide variety of other applications.”
Farrell has already applied his techniques to birds and insects, but he has also worked to identify specific tigers in India based on their stripes and even cars and different types of vehicles.
“The technology is far-reaching and can let us do things like track specific types of vehicles through street cameras,” Farrell said. “This would only be available to the military and law enforcement, but it just goes to show you how far-reaching this technology is.”
Farrell hopes this technology will be useful not only for governments but also for the average person.
“An app is one of the outcomes we hope to gain from this research,” Farrell said. “With the camera on your phone, you would be able to satisfy the curiosity that so many of us have. In addition to that, the app would be great if, for instance, you found a spider in your home. Using the image, you would be able to tell if you were dealing with a dangerous brown recluse or something harmless.”
Farrell’s colleagues are quick to praise his work and note he is one of the industry leaders in the field of computer vision.
“Ryan has a rare combination of deep knowledge of computer vision and machine learning with a personal passion for understanding the natural world around us,” said Cornell computer science professor Serge Belongie. “As visual search becomes an integral part of our computing experience alongside text-based search, new generations of users will increasingly demand fine-grained results, and I am confident that Ryan’s research will be instrumental in meeting this demand.”
