Title: How and why to learn a 3D model of the human body in motion
Michael J. Black, MPI for Intelligent Systems, Tübingen, Germany
Abstract: The human body is complex and deformable. For many applications in computer vision, graphics, fashion, and medicine, having a realistic, low-dimensional, 3D model of the body is useful. Getting a good one, however, is difficult. This talk with review the history of our work on learning 3D models of the human body from 3D scans. It will answer “what” is a body model, “why” it is useful, and “how” to build one. It will summarize how to accurately align 3D meshes of bodies in arbitrary poses and how to build a statistical model of body shape and non-rigid pose variation. It will also describe how to use a body model in computer vision by fitting it to data including 3D scans, range images, mocap markers and video. The talk will give a quick survey of applications of body shape in graphics, skin-cancer detection, body fat analysis, and body perception studies. Finally it will describe how the field is developing to capture and model 4D data of soft tissue and clothing in motion.
Bio: Michael Black received his B.Sc. from the University of British Columbia (1985), his M.S. from Stanford (1989), and his Ph.D. in computer science from Yale University (1992). After post-doctoral research at the University of Toronto, he joined the Xerox Palo Alto Research Center in 1993 where he later managed the Image Understanding area and founded the Digital Video Analysis group. From 2000 to 2010 he was on the faculty of Brown University in the Department of Computer Science (Assoc. Prof. 2000-2004, Prof. 2004-2010). He is presently one of the founding directors at the Max Planck Institute for Intelligent Systems in Tübingen, Germany, where he is the Managing Director and leads the Perceiving Systems department. The MPI studies intelligent systems from molecules to machines. He is an Honorarprofessor at University of Tübingen in Computer Science and an Adjunct Professor (Research) in Computer Science at Brown University.
He is a recipient of the 2010 Koenderink Prize for Fundamental Contributions in Computer Vision and the 2013 Helmholtz Prize for work that has stood the test of time. His work has won several paper awards including the IEEE Computer Society Outstanding Paper Award for his work with P. Anandan on robust optical flow estimation (CVPR’91). His work received Honorable Mention for the Marr Prize in 1999 (with David Fleet) and 2005 (with Stefan Roth). For his work on forensic video analysis he received the Commendation and Chief’s Award from the Henrico County Division of Police in Virginia. He is a Senior Member of the IEEE and an associate of the Canadian Institute for Advanced Research. He serves on the advisory boards of Videosurf and Willow Garage. He is also a co-founder and member of the board of directors of Body Labs Inc., which focuses on applications of 3D human body shape.
Prof. Black’s research interests in machine vision include optical flow estimation, human shape and motion analysis and probabilistic models of the visual world. In computational neuroscience his work focuses on probabilistic models of the neural code and applications of neural decoding in neural prosthetics.
Title: Solving the Person Re-identification Problem
Shaogang Gong, Queen Mary University of London, United Kingdom.
Abstract: For making sense of the vast quantity of visual data generated by the rapid expansion of large scale distributed multi-camera systems in crowded urban spaces, automatic person re-identification is essential. However, it poses a significant challenge to computer vision systems. The majority of current methods for person re-identification is focused primarily on ‘closed-world’ benchmark datasets of limited scope and size. Very little work has been done to address the real-world challenge of person re-id in ‘open-world’ environments typically exhibited in large distributed and disjoint public spaces. In this talk, I will describe recent progress on person re-identification against benchmark datasets, discuss their assumptions and limitations before presenting studies on addressing the more difficult challenge of open-world person re-identification, by exploring context information, user information, and domain-transfer information from open-source data such as the internet for facilitating person re-identification as Big Data Search in crowded urban spaces.
Bio: Shaogang Gong is Professor of Visual Computation (since 2001) and the Director of the Computer Vision Group (since 1995) at Queen Mary University of London. He received his DPhil in computer vision from Keble College, Oxford University (1989) with a thesis on computing optic flow by higher-order geometric analysis, worked as a Research Fellow on the EU ESPRIT project VIEWS for developing Bayesian graphical model-based learning systems for visual surveillance of wide-area scenes (1989-1992), and founded the Queen Mary Computer Vision Laboratory in 1993. He is elected a Fellow of the Institution of Electrical Engineers (now IET), a Fellow of the British Computer Society, a Member of the UK Computing Research Committee, and served on the Steering Panel of the UK Government Chief Scientific Adviser (GCSA) to the Prime Minister’s Science Review.
His research interests include computer vision, machine learning and video analysis. Prof. Gong has published over 300 academic papers; 2 research monographs on “Visual Analysis of Behaviour: From Pixels to Semantics” (2011) and “Dynamic Vision: From Images to Face Recognition” (2000); and 5 edited books on topics ranging from Person Re-Identification (2014), Video Analytics for Business Intelligence (2012), to Face and Gesture Recognition (2003, 2005, 2007).
He twice won the Best Science Prize (1999, 2001) of the British Machine Vision Conferences, the Best Paper Award (2001) of the IEEE International Workshops on Analysis and Modelling of Faces and Gestures, and the Best Paper Award (2005) of the IEEE International Conferences on Imaging for Crime Detection and Prevention. He is a recipient of a Queen’s Research Scientist Award (1987), a Royal Society Research Fellow (1987, 1988), a GEC-Oxford Fellow (1989), a Senior Visiting Scientist at Microsoft Research (2001) and Samsung Electronics (2003). He has founded a number of companies and is the Chief Scientist of three start-ups.
Title: The empirical examination of superior face recognisers
Josh P. Davis, University of Greenwich, United Kingdom.
Abstract: Recent research has demonstrated that there are large individual differences in the ability to recognise faces. Possibly a consequence of a natural normal distribution in ability, some people appear to possess a memory for faces vastly superior to most of the rest of the population. Although they may utilise other cues (e.g., body shape, clothing, gait etc), some London police officers possessing this ability who have identified multiple suspects from CCTV footage, have probably assisted in the detection of more crimes than any police officers in history. The results of cognitive and neuroscientific tests on some of these officers, as well as other people possessing superior face recognition ability will be discussed.
Bio: Dr. Josh P. Davis, PhD MSc BSc is a Senior Lecturer in the Applied Psychology Research Group at the University of Greenwich. His PhD was on the “Forensic Identification of Unfamiliar Faces in CCTV Images” (2007) and he has since published research on human face recognition and eyewitness identification ability, as well as methods used by expert witnesses to provide evidence of identification in court. His research has a strong applied focus in terms of its impact on the criminal justice system. His first co-edited book “Forensic Facial Identification: Theory and Practice of Identification from Eyewitnesses, Composites and CCTV” (Wiley Blackwell) is due to be published in 2015. He has recently received funding from the European Commission to advice on the management of CCTV images to ensure the Metropolitan Police Service optimally utilise staff possessing superior face recognition ability.