10/30/2025 Mackenzie Wranovics
CS professor David Forsyth wants you to imagine a world where computers have a real understanding of what they see — they can recognize faces, interpret gestures, and even predict your next move. For Forsyth, that world is not science fiction; it is simply the future.
Written by Mackenzie Wranovics
Imagine a world where computers have a real understanding of what they see — they can recognize faces, interpret gestures, and even predict your next move. For David Forsyth, a computer science professor from The Grainger College of Engineering Siebel School of Computing and Data Science at the University of Illinois Urbana-Champaign, that world is not science fiction, it is simply the future.
Professor Forsyth has been named the recipient of the 2025 PAMI Distinguished Researcher Award from IEEE TCPAMI. Forsyth, the Fulton Watson Copp Chair in Computer Science, has been working with the foundation on and off since its inception and was awarded the 2024 PAMI Mark Everingham Prize for service, making him the first person to receive both the Mark Everingham and a distinguished researcher award.
“I’m unusual now in modern computer vision in that I’ve done things in all sorts of different areas of vision. I don’t really specialize, I kind of roam around and do things,” said Forsyth. “In the very early days, I was the guy who started the whole thing about describing images with words. It was cruder than it is now by a long, long way, but that’s where we started it.”
Recently, Forsyth has been studying the ways we can exploit and understand image generators. “There are a bunch of models out there that will make images out of nothing,” said Forsyth. “You can persuade them to tell you things about the images that they’re generating that are actually very hard to get other ways, you can make them change the lighting in the image, for example, which is kind of cool.”
Forsyth has also devoted his time to finding flaws in existing imaging programs. He explained the generators can be “fantastically stupid” when it comes to simple geometry. “We have more recent work that suggests that that’s in some way kind of profoundly woven into the fabric of the generator. We don’t know how or why.”
For Forsyth, the appeal of computer vision goes beyond technical innovation — it taps into the philosophical question of intelligence. “I think vision is much more interesting than anything else,” Forsyth explains. “There’s a philosophical case that can be made here as well.”
Vision is a gateway into understanding how the mind interprets and interacts with the world. It’s probably more trouble than anyone needs,” but it’s the complexities that fuel curiosity. I regularly annoy my colleagues by just saying vision is more interesting than anything else. Why would you bother doing anything else?
—David Forsyth
|
Photo Credit: Leo Luo |
After decades of research and progress, computer vision remains a field defined by curiosity. “The reserves of ignorance are huge,” Forsyth notes, not as discouragement but as an invitation for students and researchers to take these opportunities and learn from them. “There’s so much we don’t know, but we should find out and fix them.”
With so many minds working in the field, true exploration demands vision in every sense of the word. “You have to stick out; you cannot stick out by doing what everybody else is doing,” Forsyth said. Whether this means finding an unconventional angle or embracing ideas others have overlooked, it’s not easy. The base of knowledge is vast and quickly evolving. As Forsyth puts it, “Annoyingly, the stuff you have to learn keeps changing.”
It’s relentless curiosity that makes this field so appealing to Forsyth. By testing the boundaries of what machines can see and what humans can understand about seeing, Professor Forsyth is one of many computer scientists helping to shape the future of computer vision.