From 3D CT Volume Reconstruction to 3D AI Content Generation
Speaker: Prof. Hongdong Li
DEIB - Alpha Room (Bld. 24)
October 4th, 2024 | 3.30 pm
Contacts: Proff. Federica Arrigoni, Giacomo Boracchi
DEIB - Alpha Room (Bld. 24)
October 4th, 2024 | 3.30 pm
Contacts: Proff. Federica Arrigoni, Giacomo Boracchi
Abstract
On October 4th, 2024 at 3.30 pm the seminar titled "From 3D CT Volume Reconstruction to 3D AI Content Generation" will take place at DEIB Alpha Room (Building 24).
Reconstructing the geometry of 3D shape from multi-view images or projections has been a fundamental task in Computer Vision and Medical Imaging. Recently, the emerging generative-AI technology has provided new insights and new capability to generate realistic, high quality 3D contents simply from the user's natural prompt input (such as natural language text, or a single image). In this talk, I will report two of our recent works on 3D visual reconstruction and 3D content generation. The first work is about how to reconstruct the 3D volumetric shape from X-ray projections, and the second one is on generative 3D content creation using pre-trained large-scale diffusion models such as Stable Diffusion.
For the latter, I will cover both 3D single object generation from text, and 3D scene generation from sketches/layout.
Reconstructing the geometry of 3D shape from multi-view images or projections has been a fundamental task in Computer Vision and Medical Imaging. Recently, the emerging generative-AI technology has provided new insights and new capability to generate realistic, high quality 3D contents simply from the user's natural prompt input (such as natural language text, or a single image). In this talk, I will report two of our recent works on 3D visual reconstruction and 3D content generation. The first work is about how to reconstruct the 3D volumetric shape from X-ray projections, and the second one is on generative 3D content creation using pre-trained large-scale diffusion models such as Stable Diffusion.
For the latter, I will cover both 3D single object generation from text, and 3D scene generation from sketches/layout.
This talk is part of the Imaging Seminar from EMJM in Imaging.
Short Bio
Hongdong Li is a Full Professor with the Australian National University (ANU). He is a leading researcher in the area of 3D Computer Vision. He was one of the founding CIs for Australian Centre for Robotic Vision - a Government funded research centre for Robotic Vision.
He has been serving on the Editorial Board for IEEE-TPAMI, and regularly serves as an Area Chair for CVPR, ICCV, and ECCV, and is a Senior Area Chair (SAC) for ICRA 2022, SAC for CVPR 2023, SAC for Neurlps 2024, and SAC for CVPR 2025. He was a visiting fellow with the UPMC Paris-6 University in 2009-2010, a visiting professor with the Robotic Institute of Carnegie Mellon University in 2017-2018. He was Tencent's Distinguished Scientist for its VR/AR program, leading five research labs globally in 2022.
He has been serving on the Editorial Board for IEEE-TPAMI, and regularly serves as an Area Chair for CVPR, ICCV, and ECCV, and is a Senior Area Chair (SAC) for ICRA 2022, SAC for CVPR 2023, SAC for Neurlps 2024, and SAC for CVPR 2025. He was a visiting fellow with the UPMC Paris-6 University in 2009-2010, a visiting professor with the Robotic Institute of Carnegie Mellon University in 2017-2018. He was Tencent's Distinguished Scientist for its VR/AR program, leading five research labs globally in 2022.
