报告题目：Human Vision, Machine Vision, and the Future of Visual Intelligence

报告人：虞晶怡  教授（上海科技大学）

主持人：何高奇  教授

报告时间：2023年10月31日（星期二）15:15-16:10

报告地点：线下：必赢BWIN普陀校区，办公楼小礼堂

                 线上：腾讯会议，会议号：316-509-387

报告摘要：

Human visual systems employ eye fixations, foveal vision, stereo, etc., to organize, identify, and interpret visual sensory information. It plays a critical role in helping us humans to represent and understand the 3D environment we live in. In this talk, I show that theories based on human vision have been deployed in art creation, psychophysics, and for AI researchers, camera designs in machine vision. For example, the lens, aperture, and sensor used in a camera resemble the cornea, pupil, and retina, respectively. At the same time, I demonstrate that machine vision and human vision exhibit significant discrepancies in mechanism and results, contrary to the traditional view that they are nearly identical. Such discrepancies can potentially benefit, by nature or by design, various visual computing tasks. More importantly, they are critical in assessing whether “AI” algorithms will have human-level visual intelligence in the foreseeable future.

报告人简介：

       虞晶怡，OSA Fellow，IEEE Fellow，ACM杰出科学家，智能感知与人机协同教育部重点实验室主任。他于2000年获美国加州理工学院（Caltech）双学士学位，2005年获美国麻省理工学院（MIT）博士学位。现任上海科技大学副教务长、信息科学与技术学院教授、执行经理。虞教授长期从事计算机视觉、计算成像、计算机图形学、生物信息学等领域的研究工作，并先后获得美国国家科学基金杰出青年奖（NSF CAREER Award）和美国空军研究院杰出青年奖 (AFOSR YIP Award)。在智能光场研究上，他拥有十余项国际PCT专利，已广泛应用于智慧城市、数字人、人机交互等场景。他同时担任IEEE TPAMI、IEEE TIP等多个顶级期刊编委和多个国际人工智能会议（ICCP 2016、ICPR 2020、WACV 2021、CVPR 2021、ICCV 2025）的程序主席。他是达沃斯世界经济论坛（WEF）“全球议程理事会”理事，并担任其Metaverse方向的Curator。