BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Towards 4D Foundational Models for Dynamic World Reconstruction - 
 Zeren Jiang\, University of Oxford
DTSTART:20260316T150000Z
DTEND:20260316T160000Z
UID:TALK245764@talks.cam.ac.uk
CONTACT:Elliott Wu
DESCRIPTION:Abstract: In this talk\, I will present a line of work on diff
 usion-based models for 4D reconstruction and tracking of dynamic scenes. I
  begin with Geo4D and Track4D\, where we explore how video diffusion prior
 s can be leveraged to recover the dynamic 4D structure of the world from v
 ideo observations. These models demonstrate how generative priors learned 
 from large-scale video data can significantly improve the reconstruction a
 nd tracking of complex dynamic scenes. Next\, I focus on Mesh4D\, an objec
 t-centric reconstruction framework that models dynamic objects as temporal
 ly coherent meshes. Leveraging diffusion models\, Mesh4D can plausibly inf
 er and hallucinate unseen regions of objects during motion. To further sta
 bilize training and improve reconstruction quality\, we incorporate skelet
 on priors as privileged knowledge within the diffusion reconstruction pipe
 line. Finally\, I will introduce Syn4D\, a large-scale synthetic dataset d
 esigned to support a wide range of 4D vision tasks. Syn4D enables research
  on geometry-aware novel view synthesis\, 4D reconstruction and tracking\,
  and human pose estimation\, providing a scalable platform for training an
 d evaluating future 4D models. Together\, these works represent steps towa
 rd 4D foundational models capable of reconstructing the dynamic physical w
 orld.\n\n\nBio: Zeren Jiang is a DPhil student in the Visual Geometry Grou
 p (VGG) at the University of Oxford. He received a dual degree in Software
  Engineering and Mathematics & Applied Mathematics from Beihang University
  in Beijing\, and later earned an MSc in Computer Science with distinction
  from ETH Zurich. His research lies at the intersection of computer vision
  and computer graphics\, with the goal of building systems that can percei
 ve and understand the dynamic physical world in real time\, and ultimately
  learn generative models capable of creating immersive and physically plau
 sible virtual environments. Zeren has published five peer-reviewed papers 
 as first or co-first author at top-tier venues. His work has received the 
 Best Paper Award at ACM Multimedia 2021 and the Best Video Award at IJCAI 
 2021.\n\nZoom link: https://cam-ac-uk.zoom.us/j/84730633222?pwd=C4HZnh8F5O
 NlVJEa77asYXZ6WCNYD6.1
LOCATION:Lecture Room 3B\, Department of Engineering (Trumpington Street)
END:VEVENT
END:VCALENDAR