最新消息

2017年12月4日学术报告:Deep Learning for Fast, Quality Image Processing


添加时间:2017-12-01 14:32:48



报告题目:Deep Learning for Fast, Quality Image Processing
报告时间:2017年12月4日(星期一) 下午1点30
报告时间:研究生楼103
报告人:杨蛟龙
 
 
Abstract:
In this talk, I‘ll present our latest research on employing deep learning for various image processing tasks. Our novel deep neural network based methods can produce state-of-the-art results of edge-preserving smoothing, image filter approximation, reflection removal, texture removal, intrinsic image decomposition, detail magnification, bokeh effect generation etc. The proposed networks run an order of magnitude faster than traditional methods with a GPU, and it is expected that the network structures can be further optimized for live image processing on smartphones. Our contributions include innovative network designs and training schemes, both tailored for image processing tasks.

 

Biography:
Dr. Jiaolong Yang is currently a Researcher at the Visual Computing Group of Microsoft Research Asia (MSRA). He received dual PhD degrees in Computer Science and Engineering from Australian National University (ANU) and Beijing Institute of Technology (BIT) in Sep 2016, and the bachelor’s degree in Computer Science from BIT in 2010. His research interests include computer vision, pattern recognition, and image processing. He has published 10+ papers in prestigious journals and conferences including TPAMI/CVPR/ICCV/ECCV, and been serving as reviewer/program committee member for them. The proposed new techniques have been transferred into several Microsoft products. He received the China Society of Image and Graphics (CSIG) Excellent PhD Thesis Award in 2017.

 



作者:姜玮