On the morning of March 15th at 10 o'clock, Long Xiaoxiao, a postdoctoral fellow at the University of Hong Kong, was invited by Associate Professor Yao Yao of our university to give a presentation on 3D Reconstruction and Generation for Complex Needs from Objects to Scenes to students at the School of Intelligent Science.
Abstract:
With the rapid development of computer vision, machine learning, and graphics processing technology, 3D reconstruction and generation technology has become a focus in the field of scientific research, especially in how to accurately model objects and scenes from multiple or individual images, and how to generate new 3D content based on images or text. This report first reviews the latest progress in 3D reconstruction technology, with a particular focus on the application of deep learning methods and neural rendering techniques in complex scene 3D reconstruction. We will provide a detailed introduction to the surface reconstruction techniques of neural rendering, as well as a series of research achievements in the challenges of generalization, non ideal scattering surfaces, and under calibrated multi camera systems.
With the popularity of generation models such as ChatGPT in OpenAI and Midjournal in Bunsen diagrams, the field of 3D generation has also ushered in unprecedented development opportunities. The report will delve into the history, current situation, and future development trends of the field of 3D generation, with a focus on analyzing a panoramic view from technological progress to commercial applications. This report will focus on the key technologies in the current field of 3D generation, including deep learning based model generation methods, strategies for creating 3D content through text or images, and how these technologies drive innovation in virtual reality, game development, and other commercial fields.
Reported by:
Long Xiaoxiao obtained a doctoral degree in Computer Science from the University of Hong Kong and is currently engaged in postdoctoral research at the University of Hong Kong. Studied under Professor Wang Wenping from IEEE Fellow&ACM Fellow. Engaged in long-term research in areas such as 3D reconstruction, neural rendering, and 3D generation. Currently, I have published 10 works and corresponding papers at top international conferences such as CVPR/ICCV/ECCV/ICLR.
