Speaker:
Matthias Tangemann, Postdoctoral Researcher, University of Toronto
Talk Title:
Linking Motion and Objectness in Humans and Machines
Date and Location:
Thursday, January 8, 2026
3–4 p.m.
BA 5166 (DGP seminar room) and online. Zoom registration link for virtual attendance.
Reception to follow
There is no registration required to attend this event in person. However, seating is limited, so arriving early is recommended.
Abstract:
In recent years, deep neural networks have rapidly approached human visual capabilities through end-to-end training on semantic tasks such as object recognition and vision-language alignment. However, research at the intersection of deep learning and psychophysics has also revealed striking differences to human visual perception, such as a lack of robustness and nonhuman-like errors. My research aims to improve the alignment of human and machine visual perception with a focus on mid-level vision, and in particular perceptual organization: How do humans and machines organize the signals from millions of photoreceptors and pixels, respectively, into a coherent, object-centered scene representation? In this talk, I will present recent studies centered around the Gestalt principle of "common fate" and discuss how the close link between motion and objectness can be modeled using DNNs. These studies not only reveal further differences between humans and DNNs in motion perception, but also offer a perspective on how insights from vision science can help improve the alignment of human and machine vision.
Biography:
Matthias is a researcher working at the intersection of computer vision and human vision. He received a B.Sc. in computer science from the Karlsruhe Institute of Technology, a B.Sc. in cognitive science and a M.Sc. in neural information processing from the University of Tübingen. He completed a PhD with Matthias Bethge at the University of Tübingen, studying the role of motion information for perceptual organization in humans and machines. In September 2025, he started a postdoc with Sven Dickinson and Dirk Bernhardt-Walther at the University of Toronto and the Vector Institute, Kaleem Siddiqi at McGill University and Zygmunt Pizlo at UC Irvine. His current research focus is on aligning mid-level visual perception in humans and machines.