报告题目：深度学习在点云3D目标检测中的应用
主讲人： 史少帅 德国马普信息所博士后研究员
报告时间：2022年06月15日（星期三）14:30-16:30
报告链接：腾讯会议542-591-827
报告摘要：
自动驾驶和机器人领域的快速发展促进了点云场景3D目标检测算法的不断迭代更新，也衍生了一系列3D目标检测相关的研究方向。本次报告将围绕点云3D目标检测方向介绍若干相关工作，包括如何利用长序列连续帧的点云有效提升自动驾驶场景下的3D目标检测的性能，知识蒸馏在提升3D目标检测模型效率中的应用，以及在室内场景下的3D目标检测算法框架等内容。最后希望与大家分享讨论3D目标检测未来的研究方向与落地场景。
The rapid development of autopilot and robotics has promoted the continuous iterative updating of point cloud scene 3D target detection algorithm, and also derived a series of research directions related to 3D target detection. This report will introduce some related work around the direction of point cloud 3D target detection, including how to effectively improve the performance of 3D target detection in autopilot scene by using point cloud with long sequence of continuous frames, the application of knowledge distillation in improving the efficiency of 3D target detection model, and the 3D target detection algorithm framework in indoor scene. Finally, I hope to share and discuss the future research direction and landing scene of 3D target detection with you.
报告人简介：
史少帅，德国马普信息所博士后研究员，博士毕业于香港中文大学多媒体实验室，师从王晓刚教授和李鸿升教授。主要研究方向是三维场景的感知和理解及其在自动驾驶场景中的应用。在CVPR/ICCV/ECCV/ICLR/TPAMI等顶级会议和期刊上发表多篇论文。谷歌学术引用量2500+，单篇引用量990+。读博期间曾获香港政府奖学金、谷歌博士生奖学金、WAIC云帆奖明日之星等荣誉。
Shishaoshuai is a postdoctoral researcher of Max Planck Institute of information in Germany. He graduated from the Multimedia Laboratory of the Chinese University of Hong Kong and studied under professors wangxiaogang and lihongsheng. The main research direction is the perception and understanding of 3D scene and its application in automatic driving scene. He has published many papers on top conferences and journals such as cvpr/iccv/eccv/iclr/tpami. Google has 2500+ academic citations and 990+ citations per article. During his doctoral study, he won the Hong Kong government scholarship, Google doctoral scholarship, waic cloud sail award, tomorrow star and other honors.
邀请人：人工智能系 温程璐副教授