Guest Speakers

The visualization community should not try to reinvent psychological science as it is represented by organizations like APS, Psychonomics, Cognitive Science, etc. There is over a century of experimental psychology in the West and a millennium of less structured investigation in the East (e.g Ibn Al Haytham, 1024 AD) that we must build upon. We risk spending another century reinventing the wheel. Neither can we count on off the shelf psychology research to solve our problems, most psychologists are neither aware of nor interested in the special characteristics of the experiences that we create with our interactive visualization systems. The nature of those environments differs in fundamental ways from the natural environment that experimental psychologists implicitly gear their research and their laboratory and field experiment methods. 

The solution to this problem lies in the creation of a translational cognitive science of interaction, AKA a cyberpsychology. This begins with studies of ”cognition in the wild”, continues with laboratory studies of the ways in which the novel perceptual and interactive environments that are generated by complex visualization environments affect human perception, cognition, action and collaboration and ends with translational field experiments that integrate theory and methods from lab studies into semi-realistic tasks and settings to inform visualization design choices and evaluation metrics.

Cameras are now used for many purposes beyond taking photographs. Example applications include remote medical diagnosis, crop monitoring, 3d reconstruction, document recognition, and many more. For such applications, it is desirable to have a camera act as a sensor that directly measures scene light.   The problem, however, is that most commodity cameras apply a number of camera-specific processing steps to the captured image in order to produce visually pleasing photos. As a result, different cameras produce noticeably different colors when imaging the exact same scene.   This is problematic for applications relying on color because algorithms developed using images from one camera often will not work with images captured on another camera due to color differences.  In this talk, I’ll discuss the current state of affairs for color on commodity cameras, common incorrect assumptions made in the scientific literature regarding image color, and recent developments that are helping to improve the situation.

This presentation walks through the development of a geographic information system (GIS)-based campus model for York University. The detailed workflow and data manipulation steps are explained to produce an accurate 3D model of the campus buildings, terrain, roads, pathways, lighting, and vegetation patches and trees. The model contains buildings derived from various sources including full building information models (BIM) at level of detail (LOD) 5 with interiors and in some cases furniture also modelled. The ensuing campus model, which now will be in a perpetual state of renovation and modification as the campus changes and evolves, will form a basis for planned uses in probabilistic crowd simulations for emergency preparedness both within and between buildings, other applications focused on aspects of GeoDesign principles, and multi-modal transportation-based analysis within the extended campus area. Some of these uses are discussed but work is too preliminary at this stage to present details of any end use.

In today’s highly connected world, the global population has become increasingly vulnerable to infectious disease and the health, economic, and security impacts of epidemics. In this talk, we will look at some of the ways geospatial data visualizations have contributed to preventing, detecting, and mitigating a number of today’s most pressing global health issues by supporting both insight-generating analytics and knowledge translation.

The canvases today’s designers are working with are very different from the canvases of the past and as a result the design process itself is changing. One of the most fundamental differences that The Globe and Mail is working to better understand is what happens when our designs collide with data. For a newspaper where more than 170 years of finely honed judgement and experience has traditionally provided the foundation for the decisions we make, trying to make the transition to a design paradigm where our assumptions are held up and questioned openly to the sort of scrutiny that the data now lets us bring to bear has been both a struggle and a challenge that’s proven very worthwhile. In this talk, we will walk you through some of those challenges and hopefully open your eyes to just how much of a difference emerging capabilities can make in and around the data available though mechanisms like testing and advanced analytics as well as machine learning techniques like predictive systems, classifiers and recommenders. We’ll also talk a bit about where artificial intelligence might take us when it comes to design.

This presentation will describe the experiences of QoC Health in designing healthcare solutions through co-design with patients and providers. It will include a focus on engaging patients and providers in ideation, in development of objectives and measures, through to co-design, testing, implementation and evaluation. These processes will be illustrated through examples that highlight the unique ways in which co-design can influence healthcare technology solutions.

Initiatives related to mapping of the human genome and human brain have advanced our understanding of genetic markers for disease and the neural pathways that underlie aspects of function and behaviour. Less advanced is our knowledge of the structures that comprise the human musculoskeletal and nervous systems, which are fundamental to understanding our physical interactions with the external world. To address this gap in anatomical knowledge, we are creating data-driven, computer-based representations to quantify and visualize the spatial and geometric relationships amongst anatomical structures (e.g., nerves, muscles, bones) and systems. By combining spatial data and symbolic information, we are working to create digital representations that computationally encode human anatomical variation, and that can facilitate modelling and simulation for accelerating design and healthcare innovation.

In this presentation, we will describe the satellite Automatic Identification System (s-AIS)– a fundamental tool for the analysis and understanding of the mobility of ocean vessels. We will describe the AIS system, argue that AIS data  meets all the Big Data characteristics, and introduce the challenges it presents. We will then present the work at Dalhousie on the use of Machine Learning for several applications of AIS, especially in monitoring and understanding the global fishing activities. We will discuss the full application cycle: the AIS data management challenges and how we addressed them, the research and methodological issues in applying Machine Learning (both supervised and unsupervised) on AIS data, and the AIS visualization opportunities. We will present preliminary results for two fishing-related tasks using both classical Machine Learning and Deep Learning approaches, and discuss some of the challenges of either approach.  We will outline some of the exciting data science research opportunities arising from the growing use of AIS system and its data.

We propose an automatic method to infer high dynamic range illumination from a single, limited field-of-view, low dynamic range photograph of a scene. In contrast to previous work that relies on specialized image capture, user input, and/or simple scene models, we train an end-to-end deep neural network that directly regresses a limited field-of-view photo to HDR illumination, without strong assumptions on scene geometry, material properties, or lighting. This allows us to automatically recover high-quality HDR illumination estimates that significantly outperform previous state-of-the-art methods. Consequently, using our illumination estimates for applications like 3D object insertion, we can achieve results that are photo-realistic.

To visualize data one of the first steps is to develop a visual representation. This representation is a result of a mapping by which the data can be specified. Much has been said of about the power of these visual representations. Simon (1981) said that solving a problem is simply a matter of representing so as to make the solution transparent – implying that finding the right representation solves the problem. Card et al. (1998) said that interactive visual representations can amplify our cognition – can in effect make us smarter.  In spite of this, the small box in the visualization creation pipeline that signifies the development of the visual representation remains one of the least unpacked. Through examples from my own work and others’, I will discuss the power and potential of alternate visual representations.

Indoor location information is the basis for a growing number of applications, including navigation, emergency response, virtual assistants, and a range of location-based services. Despite the relative success of techniques such as WiFi fingerprinting and pedestrian dead reckoning, indoor positioning still faces a few major challenges. One challenge is the low accuracy (meter level) of these method, which is insufficient for applications such as augmented reality. Another challenge is the dependency on an infrastructure of beacons and transmitters, which may not always be relied upon, especially in emergency situations. In this talk, I discuss the visual approach to positioning based on imagery captured by smart devices – smartphones and smart glasses. I present preliminary results that demonstrate the potential of images to provide an accurate and universal solution to indoor positioning.

Many image processing and computer vision tasks, such as image segmentation, stylization, and abstraction, can be posed as image-to-image translation problems.  This talk presents two different image-to-image translation approaches, one is rule-base and the other is learning-based.

The rule-based algorithm is capable of stylizing an input face photo using a single exemplar image.  Since the numbers and varieties of patch samples are highly limited, special cares are put into sample selection to best preserve the identity and content of the input face.  A two-phase procedure is also designed, where colors are transferred first in a semantic-aware manner, followed by edge-preserving texture transfer.

The learning-based algorithm employs Conditional Generative Adversarial Networks (GANs) to perform general cross-domain image-to-image translation.  It requires a large set of training images, but unlike existing approaches, the images do not need to be labeled.  To train in an unsupervised manner, two GANs are constructed to translate images in opposite directions, forming a closed loop.  As a result, images from either domain can be translated to the other and then reconstructed, enabling a reconstruction error term for training.

Teledyne Optech has been making lidar systems for over 40 years.  A brief history will be discussed, highlighting the evolution to the current applications for airborne and mobile mapping.  In addition,  examples from more recent static and portable lidar systems will be presented.

In this talk, we will explore the role of co-design and emerging technologies in today’s healthcare ecosystem. We will also delve into the role of data analytics in the healthcare setting – where we’ve come from, where we are currently, and where we strive to go. We will review how QoC partners with patients and providers to design healthcare technology solutions and measure relevant data to demonstrate effectiveness, efficiency, and impact. These processes will be illustrated through examples that highlight the unique ways in which co-design can shape healthcare technology solutions.

Until recently, strobes used for signaling in fire and other emergencies situations were universally based upon xenon flash tubes. These tubes are bulky, delicate and consume a lot of power. The presentation describes the development of a low powered LED based strobe device from concept through prototyping through to production. The various performance issues and how they were resolved will be discussed. The design of this real product with performance and cost constraints plus details of how to take a product from an idea through to a real physical saleable, high volume, product will be presented. Special emphasis will be given to the visual performance criteria of this type of device for emergency signaling.

Esri Canada’s Education and Research department has been working with academic partners to model out a large portion of downtown Toronto. This “digital twin” is now being used a test sandbox to generate future development scenarios for Toronto’s built environment, as well as future street scenarios around the Waterfront and the Portlands area. This presentation will highlight how our 3D model was generated and visualized, and will discuss how scenario modelling is taking place through parametric design software, where street, sidewalk and open space components are being generated to create human-focused designs in urban study areas.