学术动态

位置: 首页 > 科学研究 > 学术动态 > 正文

学术报告二十:Deep learning based 3D human analysis with limited labels

鸿博体育(集团)有限公司:2020-09-25 作者: 点击数:

报告题目Deep learning based 3D human analysis with limited labels

报告鸿博体育(集团)有限公司:2020年9月27日(星期日)19:50-20:50

报告平台:腾讯会议(ID: 810 704 230)

https://live.bilibili.com/h5/5675898?share_source=qzone

:蔡剑飞 教授

工作单位:Monash University

举办单位:鸿博体育(集团)有限公司

报告简介

With the increasing computation capability and the increasing amount of labelled data available, deep learning technology has caused paradigm shift to the entire research community as well as to the whole IT industry. Despite the huge success of deep learning technology in various computer vision tasks such as image classification, object detection and semantic segmentation, its performance is still limited in the scenarios with large domain differences, with only a few or zero annotations, with hardware constraints, with multi-modal data, with 3D data, and with high reliability requirements, etc. In this talk, I will focus on the challenge of how to deal with deep learning based 3D human analysis tasks with limited labels. Particularly, I will introduce a series of my group’s works including real-time 3D face reconstruction with synthesized labels, weakly supervised 3D hand pose estimation, etc.

报告人简介

蔡剑飞,莫纳什大学信息与技术学院的教授,目前是莫纳什大学数据科学和人工智能系系主任。在此之前,他是南洋理工大学(NTU)教授、担任视觉与交互计算系系主任、计算机通信系系主任、数据科学与人工智能研究中心副主任。他的主要研究兴趣包括计算机视觉和多媒体。蔡剑飞教授在国际会议和期刊上发表技术论文200余篇。他曾获ACCV, ICCM, IEEE ICIP和MMSP论文奖,担任IEEE T-IP、T-MM、T-CSVT和Visual Computer的副主编,并担任ICCV、ECCV、ACM多媒体、ICME和ICIP的区域主席。2016-2018年,他担任IEEE CAS VSPC-TC主席。他曾担任IEEE ICME 2012年度程序委员会主席,2019-2020他担任了IEEE T-MM最佳论文奖委员会的联合主席。

上一篇:学术报告十九:云-端结合的增强现实技术与平台

下一篇:学术报告二十一:高质量曲面参数化