AAII Day 2020 Workshop

Visual dialogue is an extension of a technology called visual question answering that really revolutionised the way we think about human-agent interaction. What is interesting about visual dialogue is that you can have an ongoing conversation with an agent about an image, or other visual information. Instead of answering one question, as with visual question answering, you are able to have an ongoing conversation. This is a significant development because it means that the dialogue agent can actively contribute to a conversation and it may even be able to help the human reach their objective, even if the human doesn’t initially know what their objective is, or how to arrive at it. It is a significant step towards real active collaboration between humans and agents. This talk will cover some of the developments, and issues, associated with visual question answering, and visual dialogue.

Anton van den Hengel is the founding Director of The Australian Institute for Machine Learning, Australia’s largest machine learning research group, and on many measures its most successful.  He is also a Chief Investigator of the Australian Centre of Excellence in Robotic Vision, and a Professor of Computer Science at the University of Adelaide. Professor van den Hengel has been a CI on over $60m in research funding from sources including Google, Facebook, Canon, BHP Billiton and the ARC.

Prof van den Hengel has won a number of awards, including the Pearcey Foundation Entrepreneur Award, the SA Science Excellence Award for Research Collaboration, and the CVPR Best Paper prize in 2010. He has authored over 300 publications, has an h-index of 49, has had 8 patents commercialised, formed 2 start-ups, and has recently had a medical technology achieve first-in-class FDA approval. Current research interests include leep learning, vison and language problems, interactive image-based modelling, large-scale video surveillance, and medical machine learning.

Prof van den Hengel and his team have developed world leading methods in a range of areas within Computer Vision and Machine learning, including methods which have placed first on a variety of international leaderboards such as: PASCAL VOC (2015 & 2016), CityScapes (2016 & 2017), Virginia Tech VQA (2016 & 2017), and the Microsoft COCO Captioning Challenge (2016). Professor van den Hengel’s team placed 4th in the ImageNet detection challenge in 2015 ahead of Google, Intel, Oxford, CMU and Baidu, and 2nd in ImageNet Scene Parsing in 2016. ImageNet is one of the most hotly contested challenges in Computer Vision.