学术报告第201869期

报告题目: Deep learning and beyond for image understanding

报 告 人：Prof. Chunhua Shen（University of Adelaide, Australia）

报告时间：2018年12月24日9：30

报告地点：南一楼中311

摘要: Dense per-pixel prediction provides an estimate for each pixel given an image, offering much richer information than conventional sparse prediction models. Thus the Computer Vision community have been increasingly shifting the research focus to per-pixel prediction. In the first part of my talk, I will introduce my team’s recent work on this topic. In particular, my team developed a few new methods including: 1) deep structured methods for per-pixel prediction that combine deep learning and graphical models such as conditional random fields. I show how to improve depth estimation from single images and semantic segmentation with the use of contextual information in the context of deep structured learning; 2) a new encoder-decoder network for dense prediction; 3) structured output learning with adversarial learning, which exploits structure information during the course of training, and removes time-consuming inference during testing. Recent advances in computer vision and natural language processing (NLP) have led to new interesting applications. Two popular ones are automatically generating natural captions for images/video and answering questions relevant to a given image (i.e., visual question answering or VQA). In the second part of my talk, I will describe several work we recently published, which takes advantage of state-of-the-art computer vision and NLP techniques to produce promising results on both tasks of image captioning and VQA.

简历:

Chunhua Shen is a Professor and Director of Research at School of Computer Science, University of Adelaide. He held an ARC Future Fellowship from 2012 to 2016. He received a PhD degree at University of Adelaide; then worked at the NICTA (National ICT Australia) computer vision program for about six years. From 2006 to 2011, he held an adjunct position at College of Engineering & Computer Science, Australian National University. He moved back to University of Adelaide in 2011.

At Adelaide, he is currently supervising a team of 21 PhD students and postdoc researchers, with a few more joining soon. His research and teaching have been focusing on Statistical Machine Learning and Computer Vision. In the past a few years, his team spent the effort on Deep Learning. In particular, with tools from deep learning, his research contributes to understand the visual world around us by exploiting the large amounts of imaging data. Professor Shen has published 220 peer-reviewed papers, including 75 papers published in CVPR, ICCV, ECCV, ICML and NIPS. His work has received 10,400 citations with H-index of 52 (Google scholar).