GAMES Webinar 2022 – 215期(智能三维内容生成：重建与创造) | Kangxue Yin (NVIDIA Toronto AI Lab)，Dai Bo (S-Lab, Nanyang Technological University)
【GAMES Webinar 2022-215期】(视觉专题：智能三维内容生成：重建与创造)
报告嘉宾1：Kangxue Yin (NVIDIA Toronto AI Lab)
报告题目：3D content creation and stylization with AI
3D content creation is essential in 3D industries such as gaming, film, VR/AR. However, traditional 3D content creation or modeling tools usually require tedious user interactions and professional skills from 3D artists. We hope to simplify this process, and thus democratize 3D content creation, via AI techniques. In this talk, I will introduce some research work we have done in this direction.
Kangxue Yin is a research scientist at NVIDIA. He works on computer graphics and computer vision, with a special interest in research problems related to 3D content creation. Before joining NVIDIA, he received his Ph.D. degree in the year 2020 from Simon Fraser University. Prior to joining SFU in 2015, he worked at Shenzhen Institutes of Advanced Technology (SIAT), Chinese Academy of Sciences (CAS). He received his B.Eng. degree from Chang’an University in the year 2012.
报告嘉宾2：Dai Bo(S-Lab, Nanyang Technological University)
报告题目：Learning Shapes from unposed 2D images using GANs
Modern GANs such as StyleGAN have made remarkable progress in capturing the natural distribution of 2D images, from which we can synthesis photo-realistic images and support various image manipulations. While some manipulation effects including changing pose require 3D geometric understanding, we are interested in whether GANs trained on 2D images can learn 3D shapes. In this talk, I will present our early attempts towards answering this question. I will start by introducing a study showing that existing GANs already possess 3D semantics to some extent, followed by two works that advance the architectures of GANs to better learn 3D shapes.
Dr.Dai Bo is a Research Assistant Professor with S-Lab, Nanyang Technological University, Singapore. He obtained his PhD and bachelor’s degree respectively from MMLAB, The Chinese University of Hong Kong and the ACM class of Shanghai Jiao Tong University. He was a postdoctoral Research Fellow at The Chinese University of Hong Kong and a visiting scholar at the University of Toronto. His interests include generative models, video analysis and cross-modality analysis. He has published over 20 papers in top AI conferences. He has also served as Area Chair for BMVC 2021 and AAAI 2022.
韩晓光博士，现任香港中文大学(深圳)助理教授，校长青年学者，2009年本科于南京航空航天大学毕业，2011年获得浙江大学应用数学系硕士学位，2011年至2013年于香港城市大学创意媒体学院任研究助理，之后于2017年获得香港大学计算机科学专业博士学位。其研究方向包括计算机视觉和计算机图形学等，在该方向著名国际期刊和会议发表论文近50余篇，包括顶级会议和期刊SIGGRAPH(Asia),CVPR,ICCV,ECCV,NeurIPS,ACM TOG,IEEE TPAMI等。他的工作DeepFashion3D曾获得CCF图形开源数据集奖，计算机图形学顶级会议Siggraph Asia 2013新兴技术最佳演示奖，2019年和2020年连续两年计算机视觉顶级会议CVPR最佳论文列表（入选率分别为0.8%和0.4%），入选2021腾讯AI Lab犀牛鸟专项研究计划，IEEE VR 2021 最佳论文荣誉提名。更多细节详见https://mypage.cuhk.edu.cn/academics/hanxiaoguang/。
GAMES主页的“使用教程”中有 “如何观看GAMES Webinar直播？”及“如何加入GAMES微信群？”的信息；