1. Scene understanding from a moving camera
2. Visual tracking and activity recognition
Australian Centre for Visual Technologies
University of Adelaide, Australia
Duration: 3 years
The Australian Centre for Visual Technologies (ACVT) is one of the largest computer vision and image analysis research groups in Australia. Based in Adelaide, we currently number over 40 members including academics, research staff and students. We work on a mixture of fundamental and commercially oriented research projects in computer vision and machine learning, and we have a strong track record of publications in the top venues.
Prof Ian Reid recently joined the group and is seeking two post-doctoral researchers to expand the research breadth of the group. The researchers will participate in exciting and innovative projects that combine tracking, SLAM, object recognition and machine learning and real-time processing, for high-level scene understanding,
Post 1. Semantic SLAM
This post is funded from an Australian Research Council grant concerned with semantic visual SLAM. The Research Fellow will investigate representations and algorithms for extraction of high-level
semantic information to improve the mapping process in SLAM, in particular introducing object recognition, object tracking and scene recognition into visual SLAM.
Post 2. Structured learning for behaviour analysis
This project is concerned with analysis of human behaviour, ranging from tracking people in crowds, analysis of pose (particularly gaze), and activity and intention recognition. In particular the Research Fellow will develop structured learning methods to leverage huge, unlabelled (or weakly labelled) datasets for improved pose analysis and activity recognition.
Within the broad remit of either project there will be considerable room for developing the research directions with Prof Reid. There will also be potential opportunities for joint supervision of postgraduate and honours level students, collaboration with the Centre, and with visitors and with international partners.