Generative Machine Perception: Learning to See Visually Subtle Signals
#AI
Abstract:
Humans rely heavily on visual signals sensed by our eyes, forming the basis of our visual perception system. To build human-like AI agents, computer vision techniques have advanced significantly as a core component. Despite these advancements, all vision systems, including our eyes, have fundamental limitations in seeing things that are small, occluded, or in the dark. In this talk, I present my recent journey toward building data-efficient, versatile, and generalizable machines by developing the next generation of machine perception, which is generative and multi-modal. The core idea is to exploit other multi-modal signals that describe the world around us, including sound and language, and pose them as generative cross-modal translation problems to fill in missing visual information beyond sight. I present deep learning systems that learn to perceive and visualize subtle signals, and what they sense about our world.
Date and Time
Location
Hosts
Registration
- Date: 04 Jun 2025
- Time: 06:00 PM UTC to 08:00 PM UTC
-
Add Event to Calendar
- Gachon University
- Seongnam, Gyeonggi-do
- South Korea 13120
- Building: AI Hall
- Room Number: 402
- Contact Event Hosts
- Co-sponsored by IEEE Seoul Section Sensors Council Chapter
Speakers
Topic:
Generative Machine Perception: Learning to See Visually Subtle Signals
Abstract:
Humans rely heavily on visual signals sensed by our eyes, forming the basis of our visual perception system. To build human-like AI agents, computer vision techniques have advanced significantly as a core component. Despite these advancements, all vision systems, including our eyes, have fundamental limitations in seeing things that are small, occluded, or in the dark. In this talk, I present my recent journey toward building data-efficient, versatile, and generalizable machines by developing the next generation of machine perception, which is generative and multi-modal. The core idea is to exploit other multi-modal signals that describe the world around us, including sound and language, and pose them as generative cross-modal translation problems to fill in missing visual information beyond sight. I present deep learning systems that learn to perceive and visualize subtle signals, and what they sense about our world.
Biography:
About the Speaker
I am an Associate Professor at the School of Computing, KAIST, Daejeon, Korea since 2025. Prior to KAIST, I joined Dept. of Elec. Eng. (adjunct with Grad. school of AI and Dept. of Creative IT Eng.), POSTECH, Korea as an Assistant Professor in 2020 and was appointed as an Associate Professor since 2023. I was also jointly affiliated with OpenLab, POSCO-RIST, Pohang, Korea from 2021 to 2023 as Research Director.
I received the B.E. degree (Valedictorian of the class) in Computer Engineering from Kwang-Woon University, South Korea in '10, and the M.S. and Ph.D. degrees in Electrical Engineering from KAIST, South Korea in '12 and '17, respectively.
I was with Facebook AI Research working with prof. Lorenzo Torresani. I was a postdoctoral associate in Computational Fabrication Group (PI: prof. Wojciech Matusik) at MIT CSAIL, Cambridge, MA, US. I completed my M.S. and Ph.D. at Dept. EE, KAIST, Korea (co-advised by profs. In So Kweon and Jinwoo Shin, and former co-advisor Dr. Yu-Wing Tai). I finished my B.E. degree (the first-class honor) from Dept. Computer Engineering at Kwang-Woon University (KWU), Korea.
I was a research intern in Visual Computing Group, Microsoft Research Asia (MSRA), Beijing, China from 2014 to 2015, where I was working with Drs. Yasuyuki Matsushita and David Wipf. I visited Cognitive Group, Microsoft Research, Redmond, WA, US in 2016 as a research intern, and worked with Drs. Sing Bing Kang, Neel Joshi and Baoyuan Wang on computational photography.
I am a recipient of Best Poster Award at BMVC2024, Microsoft Research Asia Fellowship, Gold prize of Samsung HumanTech thesis award, Qualcomm Innovation awards, Excellent research achievement award by Hyundai Motor Co., CVPR outstanding reviewer award 2020 and top research achievement awards from KAIST. I was also selected as an outstanding reviewer in CVPR'20 and ICLR'22.
I have served as area chairs for CVPR, ICCV, ECCV, NeurIPS, ICML, and ICLR, senior area chairs or senior program chairs for ICCV’25 and AAAI'22, and an associate editor for the International Journal of Computer Vision (IJCV) and Visual Computer journal (TVCJ), and ICRA'23-'24.
Agenda
Title: Generative Machine Perception: Learning to See Visually Subtle Signals
Date & Time: 3:00pm, June 5, Thursday
Speaker: Tae-Hyun Oh, Associate Professor at School of Computing, KAIST