报告题目：Deep Domain Adaptation for Object Detection and Semantic Segmentation
时间：12月19日（周三）上午10:00-11:30
地点：计算所421会议室
报告摘要：
Computer vision tasks like object detection and semantic segmentation typically assume that training and test data are drawn from an identical distribution, which, however, does not always hold in practice. Such a distribution mismatch will lead to a significant performance drop. In this talk, I will introduce our recent works on object detection and semantic segmentation using deep domain adaptation approaches. For object detection, we propose a new domain adaptive Faster R-CNN model, which demonstrates excellent cross domain object detection performance in different scenarios, including cross-dataset object detection, cross-weather object detection, and synthetic to real object detection. For semantic segmentation, we focus on the synthetic to real urban scene segmentation task, and propose several effective ways to align both the image styles and data distributions.
报告人简介：
李文，目前任瑞士苏黎世联邦理工学院计算机视觉实验室博士后研究员，合作导师为知名计算机专家Luc Van Gool教授。他于2015年在新加坡南洋理工大学取得博士学位。他的主要研究方向为计算机视觉与迁移学习，重点研究视觉应用中的标注数据有限、数据分布差异等问题，在T-PAMI、IJCV、CVPR、ICCV、ECCV等重要国际期刊和会议上发表30多篇学术论文，谷歌学术引用1300余次。个人主页：http://www.vision.ee.ethz.ch/~liwenw/
附件下载：