ViiHM First Workshop – Landscaping exercise – Theme reports Group colours Grey, Black and Cyan Theme name Vision for action and active vision Members Nicola Bellotto, Andy Foulkes, David Gibson, Iain Gilchrist, Paul Graham, Frank Guerin, Nick Hockings, Alan Johnston, Dmitry Kit, Casmir Ludwig, Andrew Philippides, Simon Rushton, Jeremy Wyatt, Johannes Zanker Please address the following questions on a flip chart putting your group colours and theme name on the chart. Add notes here also. Current hot topics: What’s currently exciting in this area - what is generating a buzz? (2) (3) (4) (5) (6) (1) Real time vision Radically different bio-inspired architectures for robotics (hardware and software) The role of other sensing modalities (including e.g. touch for robotics) The vision for action and action for vision loop How to specify and schedule the task for visually guided action and behaviour The problem of manipulation in robotics Overlaps: What overlaps between biological and computer vision are taking place? Can you identify any potential overlap areas? 1. Vision as a mobile platform 2. Active Vision Sampling + structureing vision Active Simultaneous Location and Mapping (SLAM) – Computer Vision Info Max – Biological 3. Vision for Action Goal directed Navigation Manipulation Aims: What would your group expect or like to see happen within the theme in the next 5 years? Can you identify a single ‘challenge’ topic with an overlap between biological and machine vision that might be addressed in the next 5 years? Challenges: Is there anything restricting collaboration in this area? What challenges might old back progress in this area? ViiHM First Workshop – Landscaping exercise – Theme reports Group colours Red and Yellow - 1 Theme name Plasticity and modelling Members Wendy Adams, Roland Fleming, Ross Goutcher, Sophie Wurger, David Foster, Aaro Sloman, Sepehr Jalali, Ahmed Al-Baldi, N V Kartheek Medatathi, Miles Hansard, Andrew Parker, Kun Giu Please address the following questions on a flip chart putting your group colours and theme name on the chart. Add notes here also. Current hot topics: What’s currently exciting in this area - what is generating a buzz? Perceptual Learning / Plasticity – Equivalents in machine systems Level of Models Overlaps: What overlaps between biological and computer vision are taking place? Can you identify any potential overlap areas? Aims: What would your group expect or like to see happen within the theme in the next 5 years? Can you identify a single ‘challenge’ topic with an overlap between biological and machine vision that might be addressed in the next 5 years? Targeted review for counterpart in “other camp”. Importance of being honest about what we do and do not know. Utilitarian / behaviour oriented concepts of vision. Challenges: Is there anything restricting collaboration in this area? What challenges might old back progress in this area? Frustrations: “Biological Motivation” is shallow and often simply and excuse for poor performance, mediocrity. Question: Are neurons per se important for computation? Where to publish. “Falling between the chairs” Special issues with joint reviews – consensus between reviewers from different fields. ViiHM First Workshop – Landscaping exercise – Theme reports Group colours Red and Yellow - 2 Theme name Objects, 3D, Recognition Members Max Di Luca, Andrew Glennester, Peter Scarfe, David Hunter, Paul Hibbard, Alex Murry, Sonya Coleman, Dermot Kerr, Arnlod Wilkins, Andrew Welchman, Simon Watt, Paul Hand, Toby Breckon, Shufan Yang, George Lovell Please address the following questions on a flip chart putting your group colours and theme name on the chart. Add notes here also. Current hot topics: What’s currently exciting in this area - what is generating a buzz? Deep Learning, Object recognition – Converging towards “difficult” tasks (i.e., high regularity in the space of object recognition), Bayesian implementation in neurons. Overlaps: What overlaps between biological and computer vision are taking place? Can you identify any potential overlap areas? Parallelism in processing information Natural scene statistics Aims: What would your group expect or like to see happen within the theme in the next 5 years? Can you identify a single ‘challenge’ topic with an overlap between biological and machine vision that might be addressed in the next 5 years? - Improve speed in computer vision systems using biological mechanisms. I.e. for autonomous vehicles - Efficiency; how is it defined in the two areas, how is it measured. -Make labs form different backgrounds work on same topic. I.e., to identify some specific tasks that can be tackled from the two directions. Challenges: Is there anything restricting collaboration in this area? What challenges might old back progress in this area? - Feedback from computer vision to biological models and viceversa. - How to identify the biological models to implement in large-scale computer projects (HBP, Blue brain). - Computer vision solves “low level” problems and biological vision solves “large scale” problems but it’s challenging to find overlaps. -Openness of data bases and results / open science. -Different goals in computer vision and biological vision research - What are the underlying representations? ViiHM First Workshop – Landscaping exercise – Theme reports Group colours Pink and Purple Theme name Better Together (Augmenting vision, learning and other topics) Members John Solomon, Daniel Baker, David Tolhurst, Manos Protonotarios, Craig Henderson, Mark van Rossum, Xiaohng Gao, Ruixuan Wangm Farzin Deravi, Nicolas Pugeault, Zyg Pizlo, Keith May, Thomas Tanay, Sander Keemink, Jonas Kubilius, Brian Barsky, Elena Gheorghiu, Zhaoping Li, Cristina Hilario Please address the following questions on a flip chart putting your group colours and theme name on the chart. Add notes here also. Current hot topics: What’s currently exciting in this area - what is generating a buzz? Multi-electrode recording Big data, Optogenetics, Materials perception Deep networks Learning linking to machine vision Fundamental differences between fovea and periphery Attention Future hot topic using computers to augment human vision Overlaps: What overlaps between biological and computer vision are taking place? Can you identify any potential overlap areas? Difference may be a good thing Augmentation of low vision Edge detection Segmentations Statistics of signals Aims: What would your group expect or like to see happen within the theme in the next 5 years? Can you identify a single ‘challenge’ topic with an overlap between biological and machine vision that might be addressed in the next 5 years? Challenges: Is there anything restricting collaboration in this area? What challenges might old back progress in this area? ViiHM First Workshop – Landscaping exercise – Theme reports Group colours Orange & Green Theme name Attention and scene understanding Members Darren Cosker, Eva Krumhuber, Dietmar Heinke, Anna Belardinelli, Jiangning Gao, Nick Costen, Jason Vanderventner, Isabelle Mareschal, Ales Leonardis, Tom Foulsham, Richard Bowden, Charles Leek Please address the following questions on a flip chart putting your group colours and theme name on the chart. Add notes here also. Current hot topics: What’s currently exciting in this area - what is generating a buzz? Context: whats important in a scene - How visual system selects / extracts - Describes which information to use to make decision about scene. - Top down / Bottom up - Egocentric capture - Type of stimulus / Scene, colour - How we extract information from complex scenes / simple stimuli Overlaps: What overlaps between biological and computer vision are taking place? Can you identify any potential overlap areas? - Create model of scene& synthesise new ones - Attention, where to look in a scene - How biological systems & machine systems help each other -Plausible model of biological system what canbe tested by machine vision. -Building intelligent models / realism -Priors. Aims: What would your group expect or like to see happen within the theme in the next 5 years? Can you identify a single ‘challenge’ topic with an overlap between biological and machine vision that might be addressed in the next 5 years? - Grand challeng (50 years) blade runner. -What do humans attend to? / How we understand interaction. -Understand basis of attention / cognitions / Navigation. -Go beyond very small defined areas – vision systems normally studies in very small constrained way (small areas). - How info is put together (humans are good at integrating information) subtle cues, ways of finding ground truth. Challenges: Is there anything restricting collaboration in this area? What challenges might old back progress in this area? -Black box: not most efficient system if too aim for perfect system. - What are rhe most fundamental components. - What’s the metric of performance (accuracy, speed)? - Scalability, how to model large number of categories. - Different types of methods, can work in incremental manner (functional level). ViiHM First Workshop – Landscaping exercise – Theme reports Group colours Blue and Brown Theme name Colour and appearance Members Stephen Pollard, Frederic Fol Leymarie, Han Gong, Jansa Matinovic, Bougslaw Obara, Chas Nelson, Kazda Xiao, Tushar Chauhan, Rafel Mantiuk Please address the following questions on a flip chart putting your group colours and theme name on the chart. Add notes here also. Current hot topics: What’s currently exciting in this area - what is generating a buzz? - 3D printing - > colour management. - Colour in 3D -> luminance role in colour perception. - Appearance as multisensory problem. - Influence of higher – Cognition, - Emotion, - Culture Overlaps: What overlaps between biological and computer vision are taking place? Can you identify any potential overlap areas? Conflict between: Simplify to measure / study Personalise to perform Aims: What would your group expect or like to see happen within the theme in the next 5 years? Can you identify a single ‘challenge’ topic with an overlap between biological and machine vision that might be addressed in the next 5 years? Challenges: Is there anything restricting collaboration in this area? What challenges might old back progress in this area? - Language not shared - Taking time to speak to colleagues just to understand other areas.