Geuntaek Lim

Currently, I am working as a research intern under the guidance of Professor Jonghyun Choi at the Machine Perception and Reasoning Lab, Seoul National University.

My research focuses on multimodal AI, with a particular emphasis on video-centric AI modeling. My research focuses on empowering AI systems to aid humans in interpreting complex video content, thereby facilitating higher-level reasoning across various applications such as sports analytics, surveillance and media-content. I am deeply interested in exploring the following three key areas for a comprehensive understanding of videos:

1. Efficient Video Representation: The high computational demands of video data necessitate the use of efficient techniques like keyframe selection and tokenization.

2. Perception and Reasoning in Videos: Understanding temporal information in video, such as frame continuity, causality, and diversity, remains a challenging problem.

3. Multi-modal Learning: Audio, when encoded in video streams, introduces semantic information that complements but is distinct from visual semantics.