12th Conference on Computer and Robot Vision
Halifax, Nova Scotia. June 3-5, 2015.
Welcome to the home page for CRV 2015 which will be held at Dalhousie
and at Saint Mary's Universities.
CRV is an annual conference hosted in Canada, and co-located with Graphics Interface (GI)
and Artificial Intelligence (AI). A single registration covers attendance in all three conferences.
Please see the AI/GI/CRV general
conference website for more information.
The CRV proceedings are published through the Conference Publishing Services (CPS). Accepted papers will be submitted to Xplore. Xplore has published CRV accepted papers since 2004. See archive for links to previous years' papers.
Important Dates:
| Paper submission | March 3, 2015 | |
| Acceptance/rejection notification | March 27, 2015 | |
| Revised camera-ready papers | April 10, 2015 | |
| Early registration | April 20, 2015 (Registration Website) | |
| Conference | June 3-5, 2015 |
Conference History
In 2004, the 17th Vision Interface conference was renamed the 1st Canadian Conference on Computer and Robot Vision.
In 2011, the name was shortened to Conference on Computer and Robot Vision.
CRV is sponsored by the
Canadian Image Processing and Pattern Recognition Society (CIPPRS).
CRV 2015 Program
The higher-level view of the joint conference program which also includes the AI and GI meetings is available here.
DAY 1: Wednesday, June 3, 2015 (Saint Mary's University)
8:30-9:00 AM -- Joint Conference Welcome
9:00-10:00 AM -- Symposium: Novel Imaging Techniques
- Marcus Brubaker (University of Toronto)
Efficient 3D Molecular Structure Estimation with Electron Cryomicroscopy - Sebastien Roy (Université de Montréal)
Immersive Stereo Capture: the Omnipolar Camera
10:00-10:30 PM -- Oral Presentations
session chair: Marcus Brubaker- Development of a Low Cost Gamma-Ray Imaging System Using Handheld Scintillation Detectors for Visual Surveying of
Radiation Fields with Robots
Alex Miller, and Rachid Machrafi - Deep Learning Architectures for Soil Property Prediction
Matthew Veres, Griffin Lacey, and Graham W. Taylor
10:30-11:00 AM -- Break
11:00-12:00 PM -- KEYNOTE
-
Kostas Daniilidis (University of Pennsylvania)
3D Object Detection, Pose, and Shape Estimation
12:00-12:30 PM -- Oral Presentations
session chair: Michael Langer- Drifter Sensor Network for Environmental Monitoring
Daniel Boydstun, Matthew Farich, John McCarthy III, Silas Rubinson, Zachary Smith, and Ioannis Rekleitis - The Battle for Filter Supremacy: A Comparative Study of the Multi-State Constraint Kalman Filter
and the Sliding Window Filter
Lee Clement, Valentin Peretroukhin, Jacob Lambert, and Jonathan Kelly
12:30-2:00 PM -- Lunch
2:00-3:00 PM -- Symposium: LIDAR
- Ruisheng Wang (University of Calgary) Scene Parsing Using Graph Matching on Street View Data
- Gunho Sohn (York University) 3D Infrastructure Scene Reconstruction Using Laser Point Clouds
3:00-3:30 PM -- Oral Presentations
session chair: Yiannis Rekleitis- Registration of Noisy Point Clouds using Virtual Interest Points
Mirza Tahir Ahmed, Mustafa Mohamad, Joshua A. Marshall, and Michael Greenspan - Simultaneous Scene Reconstruction and Auto-calibration using Constrained Iterative
Closest Point for 3D Depth Sensor Array
Meng Xi Zhu, Christian Scharfenberger, Alexander Wong, and David A. Clausi
3:30-4:00 PM -- Break
4:00-5:30 PM -- POSTER SESSION (see bottom of page for list of posters)
6:00-9:00 PM -- Reception Alexander Keith Brewery
DAY 2: Thursday, June 4, 2015 (Saint Mary's University)
9:00-10:00 AM -- Symposium: Autonomous Robots
- Tim Barfoot (University of Toronto) Long-Term Visual Route Following for Mobile Robots
- Howard Li (University of New Brunswick) Perception, Navigation and Target Localization for Autonomous UAVs
10:00-10:30 AM -- Oral Presentations
session chair: Jonathan Kelly-
Eyes in the Back of of Your Head: Robust Visual Teach & Repeat Using Multiple Stereo Cameras
Michael Paton, Francois Pomerleau, and Timothy D. Barfoot - Being in Two Places at Once: Smooth Visual Path Following on Globally Inconsistent Pose Graphs
Sebastian Kai van Es and Timothy D. Barfoot
10:30-11:00 AM -- Break
11:00-12:10 PM -- Doctoral Dissertation Award Session
session chair: Michael Langer- Zahra Karimaghaloo (2014 CIPPRS Dissertation Award Winner)
Dept. of Electrical and Computer Engineering & Center for Intelligent Machines, McGill University
"Hierarchical Adaptive Voxel and Textural Conditional Random Field for Enhanced Pathology Segmentation"
, - Yogesh Girdhar (2014 CIPPRS Dissertation Award Honourable Mention)
School of Computer Science & Center for Intelligent Machines, McGill University
"Unsupervised Semantic Perception, Summarization, and Autonomous Exploration for Robots in Unstructured Environments"
12:10-12:30 PM -- Invited Talk
- Gregory Dudek (McGill U)
Emerging Challenges in Field Robotics
12:30-2:00 PM -- Lunch
2:00-2:30 PM -- Oral Presentations
session chair: Jean-Francois Lalonde-
At All Costs: A Comparison of Robust Cost Functions for Camera Correspondence Outliers
Kirk MacTavish and Timothy D. Barfoot -
RKLT: 8 DOF real-time robust video tracking combing coarse RANSAC features and accurate fast template registration
Xi Zhang, Abhineet Singh, and Martin Jagersand
2:30-3:30 PM -- Symposium: Vision for Graphics
- Jean Francois Lalonde (Laval U) Understanding outdoor lighting in vision and graphics
- Minglun Gong (Memorial University) Modeling and analyzing 3D shapes using clues from 2D images
3:30-4:00 PM -- Break
4:00-5:00 PM -- Symposium: Intelligent Vehicles
- Steve Beauchemin (Western U) Vehicular Instrumentation for the Study of Driver Intent and Related Applications
- Raquel Urtasun (University of Toronto) Towards Affordable Self Driving Cars
5:00-5:30 PM -- Oral Presentations
session chair: Steven Beauchemin-
Detection and Segmentation of Quasi-Planar Surfaces Through Expectation Maximization under a Planar Homography Constraint
Christopher Herbon, Gabriel Schumann, Klaus-Dietz Tönnies, and Bernd Stock -
Dense Depth Map Reconstruction from Sparse Measurements Using a Multilayer Conditional Random Field Model
Francis Li, Edward Li, Mohammad Javad Shafiee, Alexander Wong, and John Zelek
5:30-6:30 PM -- CIPPRS Annual General Meeting
7:00-10:00 PM -- Awards Banquet
DAY 3: Friday, June 5, 2015 (Dalhousie University)
9:00-10:00 AM -- Symposium: Vision and Learning
- Graham Taylor (University of Guelph) Learning Representations with Multiplicative Interactions
- Yang Wang (University of Manitoba) Recognizing and Localizing Novel Objects
10:00-10:30 AM -- Oral Presentations
session chair: Graham Taylor- Zero-Shot Object Recognition Using Semantic Label Vectors
Shujon Naha and Yang Wang - Fire Detection in Videos of Violent Crowds Acquired with Handheld Devices
Kawthar Moria, Alexandra Branzan Albu, and Kui Wu
10:30-11:00 AM -- Break
11:00-12:00 PM -- Object Detection Symposium
- Sven Dickinson (University of Toronto) Detecting Symmetric Parts in Cluttered Scenes
- Sanja Fidler (University of Toronto) Understanding Complex Scenes and People That Talk about Them
12:00-12:30 PM -- Oral Presentations
session chair: Sven Dickinson- Feature Ranking in Dynamic Texture Clustering
Thanh Minh Nguyen, Jonathan Wu, and Dibyendu Mukherjee - CPS: 3D Compositional Parts Segmentation through Grasping
Safoura Rezapour Lakani, Mirela Popa, Antonio J. Rodriguez-Sanchez, and Justus Piater
12:30-2:00 PM -- Lunch
2:00-3:00 PM -- KEYNOTE
- Demetri Terzopoulos, UCLA
3:00-3:30 Oral Presentations
session chair: Faisal Qureshi- Automated Localization of Brain Tumors in MRI Using Potential-K-means Clustering Algorithm
Ivan Cabria and Iker Gondra - Lung Nodule Classification Using Deep Features in CT Images
Devinder Kumar, Alexander Wong, and David A. Clausi
3:30-4:00 PM -- Break
4:00-5:00 PM -- Symposium: Human-Robot Interaction / Assistive Tech
- Dana Kulic (University of Waterloo) Human Motion Analysis for Rehabilitation
- Babak Taati (Toronto Rehabilitation Institute, University of Toronto) Computer vision systems in dementia care
5:00-5:30 PM -- Oral Presentations
session chair: Faisal Qureshi-
3D vs. 2D: On the Importance of Registration for Hallucinating Faces under Unconstrained Poses
Chengchao Qu, Christian Herrmann, Eduardo Monari, Tobias Schuchert, and Jürgen Beyerer - Reconstruction of 3-D Density Functions from Few Projections: Structural Assumptions for Graceful
Degradation
Michael Cormier, Daniel J. Lizotte, and Richard Mann
List of Posters (Wed. June 3, 4:00-5:30 PM)
Improved Threshold Selection by using Calibrated Probabilities for Random Forest ClassifiersFlorian Baumann Jinghui Chen, Karsten Vogt, and Bodo Rosenhahn
An Online Unsupervised Feature Selection and Its Application for Background Suppression
Thanh Minh Nguyen, Q. M. Jonathan Wu, and Dibyendu Mukherjee
Image Sensor Modeling: Noise and Linear Transformation Impacts on the Color Gamut
Mehdi Rezagholizadeh and James J. Clark
Mobile 3D Gaze Tracking Calibration
Christian Scheel and Oliver Staadt
A Perceptual Depth Shape-based CRF Model for Deformable Surface Labeling
Gang Hu, Derek Reilly, Qigang Gao,Arthur Bastos, and Nhu loan Truong
Face Retrieval on Large-Scale Video Data
Christian Herrmann and Jürgen Beyerer
Latent SVM for Object Localization in Weakly Labeled Videos
Mrigank Rochan and Yang Wang
A Fingerprint Indexing Approach Using Multiple Similarity Measures and Spectral Clustering
Ntethelelo A. Mngenge, Linda Mthembu, Fulufhelo V. Nelwamondo, and Cynthia Ngejane
A method for global nonrigid registration of multiple thin structures
Mark Brophy, Ayan Chaudhury and Steven S. Beauchemin, and John L. Barron
Uncertainty Reduction via Heuristic Search Planning on Hybrid Metric/Topological Map
Qiwen Zhang, Ioannis Rekleitis, and Gregory Dudek
A Solution to Face-to-Face Contact in Tele-presence Systems
Pierre Boulanger and Xiaozhou Zhou
Safe Close-Proximity and Physical Human-Robot Interaction Using Industrial Robots
Danial Nakhaeinia, Pascal Laferrière, Pierre Payeur, and Robert Laganière
On Visual Servoing to Improve Performance of Robotic Grasping
Mona Gridseth, Katharina Hertkorn, and Martin Jagersand
Evolution of Programs for Segmentation of Microscopic Images
Nawwaf Kharma, Mohammad Ebne-Alian, and Louis Charbonneau
Preprocessing Realistic Video for Contactless Heart Rate Monitoring Using Video Magnification
Ahmed Alzahrani and Anthony Whitehead
A Hidden Markov Model for Vehicle Detection and Counting
Nicholas Miller, Mohan A. Thomas, Justin A. Eichel, and Akshaya Mishra
An Integrated System for Mapping Red Clover Ground Cover Using Unmanned Aerial Vehicles, A Case Study in Precision Agriculture
Ammar M. Abuleil, Graham W. Taylor, and Medhat Moussa
Shrink Wrapping Small Objects
Sricharana Rajagopal and Kaleem Siddiqi
Computer Vision Based Autonomous Robotic System for 3D Plant Growth Measurement
Ayan Chaudhury, Christopher Ward, Ali Talasaz, Alexander G. Ivanov, Norman P.A. Huner, Bernard Grodzinski, Rajni V. Patel, and John L. Barron
Vision-based Collision Avoidance for Personal Aerial Vehicles using Dynamic Potential Fields
Faizan Rehmatullah and Jonathan Kelly
Orbital SLAM
Corinne Vassallo, Wennie Tabib, and Kevin Peterson
Out-of-Core Surface Reconstruction from Large Point Sets for Infrastructure Inspection
Chen Xu, Simon Fréchet, Denis Laurendeau, and François Mirallès
Situational Awareness for Manufacturing Applications
Olivier St-Martin Cormier, Andrew Phan, and Frank P. Ferrie
Improving Segmentation Boundaries with Nonparametric Image Parsing
Hong Pan and Jochen Lang
CRV 2015 Co-Chairs
- Michael Langer, McGill University
- Faisal Qureshi, University of Ontario Institute of Technology (UOIT)
CRV 2015 Program Committee
- Mohand Said Allili, Université du Québec en Outaoauis, Canada
- Robert Allison, York University, Canada
- Alexander Andreopoulos, IBM Research, Canada
- John Barron, University of Western Ontario, Canada
- Steven Beauchemin, University of Western Ontario, Canada
- Robert Bergevin, Université Laval, Canada
- Guillaume-Alexandre Bilodeau, École Polytechnique Montréal, Canada
- Pierre Boulanger, University of Alberta, Canada
- Jeffrey Boyd, University of Calgary, Canada
- Marcus Brubaker, University of Toronto, Canada
- Neil Bruce, University of Manitoba, Canada
- Gustavo Carneiro, University of Adelaide, Australia
- James Clark, McGill University, Canada
- David Clausi, University of Waterloo, Canada
- Dana Cobzas, University of Alberta, Canada
- Jack Collier, DRDC Suffield, Canada
- Kosta Derpanis, Ryerson University, Canada
- Gregory Dudek, McGill University, Canada
- James Elder, York University, Canada
- Mark Eramian, University of Saskatchewan, Canada
- Frank Ferrie, University of McGill, Canada
- Alexander Ferworn, Ryerson University, Canada
- Paul Fieguth, Waterloo, Canada
- Brian Funt, Simon Fraser University, Canada
- Philippe Giguère, Laval University, Canada
- Yogesh Girdhar, Woods Hole Oceanographic Institute, USA
- Minglun Gong, Memorial University of Newfoundland, Canada
- Michael Greenspan, Queens University, Canada
- Kamal Gupta, Simon Fraser University, Canada
- Wolfgang Heidrich, University of British Columbia
- Jessy Hoey, University of Waterloo, Canada
- Andrew Hogue, University of Ontario Institute of Technology, Canada
- Randy Hoover, South Dakota School of Mines and Technology, USA
- Martin Jagersand, University of Alberta, Canada
- Michael Jenkin, York University, Canada
- Allan Jepson, University of Toronto, Canada
- Hao Jiang, Boston College, USA
- Pierre-Marc Jodoin, Université de Sherbrooke, Canada
- Jonathan Kelly, University of Toronto, Canada
- Dana Kulic, University of Waterloo, Canada
- Robert Laganière, University of Ottawa, Canada
- Jean-Francois Lalonde, Laval University, Canada
- Jochen Lang, University of Ottawa, Canada
- Cathy Laporte, ETS Montreal, Canada
- Denis Laurendeau, Laval University, Canada
- Howard Li, University of New Brunswick, Canada
- Jim Little, University of British Columbia, Canada
- Shahzad Malik, University of Toronto, Canada
- Scott McCloskey, Honeywell Labs, USA
- David Meger, McGill University, Canada
- Jean Meunier, Universite de Montreal, Canada
- Max Mignotte, Universite de Montreal, Canada
- Gregor Miller, University of British Columbia, Canada
- Greg Mori, Simon Fraser University, Canada
- Christopher Pal, École Polytechnique Montréal, Canada
- Pierre Payeur, University of Ottawa, Canada
- Cédric Pradalier, Georgia Tech. Lorraine, France
- Yiannis Rekleitis, University of South Carolina, USA
- Junaed Sattar, Clarkson University, USA
- Christian Scharfenberger, University of Waterloo, Canada
- Angela Schoellig, University of Toronto, Canada
- Kaleem Siddiqi, McGill University, Canada
- Gunho Sohn, York University, Canada
- Minas Spetsakis, York University, Canada
- Uwe Stilla, Technische Universitaet Muenchen, Germany
- Graham Taylor, University of Guelph, Canada
- Lan Tian, Stanford University, USA
- Chi Hay Tong, University of Oxford, United Kingdom
- John Tsotsos, York University, Canada
- Olga Veksler, University of Western Ontario, Canada
- Ruisheng Wang, University of Calgary, Canada
- Yang Wang, University of Manitoba, Canada
- Steven Waslander, Waterloo University, Canada
- Alexander Wong, Waterloo University, Canada
- Robert Woodham, University of British Columbia
- Yijun Xiao, University of Edinburgh United Kingdom
- Herb Yang, University of Alberta, Canada
- Alper Yilmaz, Ohio State University, USA
- John Zelek, University of Waterloo Ontario, Canada
- Hong Zhang, University of Alberta, Canada
CIPPRS Executive
- President: Gregory Dudek, McGill University
- Treasurer: John Barron, Western University
- Secretary: Jim Little, University of British Columbia
Keynote Speakers
Wednesday, June 3, 2015, 11am
Kostas Daniilidis, GRASP Laboratory, University of Pennsylvania
Title: "3D Object Detection, Pose, and Shape Estimation"
Abstract:In this talk, we address the problem of detection, localization, and reconstruction of 3D objects in cluttered scenes. Object exemplars are given in terms of 3D models and view exemplars. We learn and select appearance-based discriminative parts which are mapped onto the 3D model from the training set through a facility location optimization. The training set of 3D models is summarized into a sparse set of shapes from which we can generalize by linear combination. The main challenge is how to combine geometry and appearance: how to select part hypotheses and compute the 3D pose and shape at the same time. To achieve this, we optimize a function that minimizes simultaneously the geometric reprojection error as well the appearance matching of the parts. We apply the alternating direction method of multipliers (ADMM) to minimize the resulting convex function. For specific classes of objects like surfaces of revolution, we prove how we can estimate 3D pose and shape from two views without any appearance information.
This is joint work with Menglong Zhu, Xiaowei Zhou, Matthieu Lecce, Cody Phillips, Kosta Derpanis, Nikolay Atanasov, Spyros Leonardos, and George Pappas.
Bio: Kostas Daniilidis is Professor of Computer and Information Science at the University of Pennsylvania where he has been faculty since 1998. He was the director of the interdisciplinary GRASP laboratory from 2008 to 2013. He obtained his undergraduate degree in Electrical Engineering from the National Technical University of Athens, 1986, and his PhD in Computer Science from the University of Karlsruhe, 1992. His research interests are on visual motion and navigation, image matching, and 3D object recognition and reconstruction. He was Associate Editor of IEEE Transactions on Pattern Analysis and Machine Intelligence from 2003 to 2007. He co-chaired with Pollefeys 3DPVT 2006, and he was Program co-chair of ECCV 2010. He is an IEEE Fellow.
Friday, June 5, 2015, 2 pm
Demetri Terzopoulos, UCLA
Title: "TBD"
Abstract: TBDBio: TBD
Symposia
CRV 2015 will feature eight exciting symposia on subtopics related to computer and robot vision.
Novel Imaging Techniques
Wed. June 3, 9 AM- 10:00 AM- Marcus Brubaker, Univ. of Toronto
"Efficient 3D Molecular Structure Estimation with Electron Cryomicroscopy"
Abstract:
Discovering the 3D structure of molecules such as proteins and viruses is a fundamental research problem in biology and medicine. Electron Cryomicroscopy (Cryo-EM) is a promising vision-based technique for structure estimation which attempts to reconstruct 3D structures from 2D images. This talk reviews the computational problems in Cryo-EM which are closely related to classical vision problems such as object detection, multiview reconstruction and computed tomography. Finally, a framework is introduced for reconstruction of 3D molecular structure which exploits modern methods for stochastic optimization and importance sampling. The result is a method which is efficient, robust to initialization and flexible. - Sebastien Roy, Univ. de Montreal
"Immersive Stereo Capture: the Omnipolar Camera"
Abstract:
With the recent advances in virtual reality headsets, there is a strong interest in reproducing the real environment using stereoscopic immersive (omnistereo) capture. This talk will introduce the problems of combining full immersive capture with stereo depth in the context of human stereo perception. A multi-camera setup, the "omnipolar camera", will also be presented. It relies on the epipolar geometry to reduce parallax artefacts and to simplify image stitching.
LIDAR
Wed. June 3, 2:00-3:00 PM- Ruisheng Wang, Univ. of Calgary
"Scene Parsing Using Graph Matching on Street View Data"
Abstract:
In this talk, a street scene parsing scheme that takes advantages of images from perspective cameras and range data from LiDAR is presented. First, pre-processing on the image set is performed and the corresponding point cloud is segmented according to semantics and transformed into an image pose. A graph matching approach is introduced into our parsing framework, in order to identify similar sub-regions from training and test images in terms of both local appearance and spatial structure. By using the sub-graphs inherited from training images, as well as the cues obtained from point clouds, this approach can effectively interpret the street scene via a guided MRF inference. We further introduce low-rank regularization into the graph matching and reformulate the low-rank graph matching problem into a standard semidefinite proragmming problem, which is much easier to solve. The matching performance is enhanced and experimental results show a promising performance of our approach. - Gunho Sohn, York Univ.
"3D Infrastructure Scene Reconstruction Using Laser Point Clouds
Abstract:
LiDAR (Light Detection and Ranging) is an emerging remote sensing technology that directly measures the distance between the sensor and a target surface using the latest time-of-flight technology, thus providing massive and highly accurate three-dimensional point clouds. Over the last decade, LiDAR has been rapidly adopted as a primary sensor in Geomatics community for supporting a wide range of applications such as bathymetry, forestry, mining, ecology, topographic mapping, and engineering. One of primary research interests in Geomatics is to reconstruct “As-Built” infrastructure models, approximating the existing infrastructure conditions modelled with semantically rich primitives. Having such accurate model representation allows us to conduct high-precision risk analysis, inventory update and management. However, like many other vision tasks, automatically generating large-scale “As-Built” models still remains unresolved research problems. Thus, today’s practice used for the infrastructure management heavily relies on human-centric and time consuming process. This presentation will introduce the latest research activities at York University, studying image understanding and model reconstruction of building facades and rooftops, single trees, railways and power lines using LiDAR point clouds.
Autonomous Robots
Thurs. June 4, 9:00-10:00 AM- Tim Barfoot, Univ. of Toronto
"Long-Term Visual Route Following for Mobile Robots"
Abstract:
I will describe a particular approach to visual route following for mobile robots that we have developed, called Visual Teach & Repeat (VT&R), and what I think the next steps are to make this system usable in real-world applications. We can think of VT&R as a simple form of simultaneous localization and mapping (without the loop closures) along with a path-tracking controller; the idea is to pilot a robot manually along a route once and then be able to repeat the route (in its own tracks) autonomously many, many times using only visual feedback. VT&R is useful for such applications as load delivery (mining), sample return (space exploration), and perimeter patrol (security). Despite having demonstrated this technique for over 500 km of driving on several different robots, there are still many challenges we must meet before we can say this technique is ready for real-world applications. These include (i) visual scene changes such as lighting, (ii) physical scene changes such as path obstructions, and (iii) vehicle changes such as tire wear. I’ll discuss our progress to date in addressing these issues and the next steps moving forward. - Howard Li, Univ. of New Brunswick
"Perception, Navigation and Target Localization for Autonomous UAVs"
Abstract:
Unmanned Aerial Vehicles (UAVs) and robots usually are related to situations involving hazardous environments, repetitive and menial tasks. UAVs can be used in many areas, such as surveillance, forestry management, mine hunting, automatic inspection of power plants and refineries, and disposal of hazardous materials. In this talk, we will present our current research in UAVs and robotics. We will present the sensing, perception, navigation, and localization methods. Simultaneous Localization and Mapping algorithms will be introduced. Results of our current research in robotics and unmanned vehicles will be presented.
Vision for Graphics
Thurs. June 4, 2:30-3:30 PM- Jean-Francois Lalonde, Laval Univ.
"Understanding outdoor lighting in vision and graphics"
Abstract:
Outdoor illumination creates challenges for computer vision and graphics alike. In vision, algorithms routinely get confused by strong shadows, highlights, and glare. In graphics, simulating the extreme dynamic range of outdoor lighting needs to be done accurately to realistically synthesize these effects. In this talk, I will present approaches that aim to improve our understanding of natural lighting with applications in both vision and graphics. First, I will briefly present approaches that rely on a physically-based illumination model to infer scene and illumination properties from time-lapse sequences and single images, by explicitly reasoning about the illumination conditions. Second, I will present recent work that relies on a data-driven model, trained on a novel dataset of 8,000+ HDR photographs of daytime skies. We leverage this new dataset to 1) automatically estimate the illumination conditions in image collections, which allows us to seamlessly insert virtual objects in the images, and 2) characterize the behavior of photometric stereo under natural lighting. - Minglun Gong, Memorial Univ.
"Modeling and analyzing 3D shapes using clues from 2D images"
Abstract:
An image worth a thousand words. From images, we humans are able to infer the 3D shape of an object and to decompose the object into semantically meaningful parts. Now, is it possible to teach computers to do these tasks? Two recent research projects that work along this direction will be presented in this talk. The first one investigates how the 3D modeling of flower head can be facilitated using a single photo of the flower. The core idea is that flower head typically consists of petals of similar 3D geometries, yet their observed shapes on 2D images vary due to differences in projecting directions. Exploiting this variation allows us to reconstruct the 3D geometry of the petals from a single image. The second project studies how to segment 3D models into semantically meaningful parts based on knowledge learned from labeled 2D images. Here the input 3D model is treated as a collection of 2D projections, which are labeled using training images of similar objects. The 3D model is then segmented by summarizing the labeling for its projections. Here the key is, for each query projection, how to retrieve objects with similar semantic parts and transfer their labels over.
Intelligent Vehicles
Thurs. June 4, 4 PM - 5 PM- Raquel Urtasun, Univ. of Toronto
"Towards Affordable Self Driving Cars"
Abstract:
Developing autonomous systems that are able to assist humans in everyday's tasks is one of the grand challenges in modern computer science. Notable examples are personal robotics for the elderly and people with disabilities, as well as autonomous driving systems which can help decrease fatalities caused by traffic accidents. In order to perform tasks such as navigation, recognition and manipulation of objects, these systems should be able to efficiently extract 3D knowledge of their environment. In this talk, I'll show how graphical models provide a great mathematical formalism to extract this knowledge. In particular, I'll focus on a few examples, including 3D reconstruction, 3D object and layout estimation and self-localization. - Steven Beauchemin, Western University
Title: "Vehicular Instrumentation for the Study of Driver Intent and Related Applications"
Abstract:
We describe a vehicular instrumentation for the study of driver intent. Our instrumented vehicle is capable of recording the 3D gaze of the driver and relating it to the frontal depth map obtained with a stereo system in real-time, including the sum of vehicular parameters actuator motion, speed, and other relevant driving parameters. Additionally, we describe other real-time algorithms that are implemented in the vehicle, such as a frontal vehicle recognition system, a free lane space estimation method, and a GPS position-correcting technique using lane recognition as land marks.
Vision and Learning
Fri. June 5, 9:00 AM- 10:00 AM- Graham Taylor, Univ. of Guelph
"Learning Representations with Multiplicative Interactions"
Abstract:
Representation learning algorithms are machine learning algorithms which involve the learning of features or explanatory factors. Deep learning techniques, which employ several layers of representation learning, have achieved much recent success in machine learning benchmarks and competitions, however, most of these successes have been achieved with purely supervised learning methods and have relied on large amounts of labeled data. In this talk, I will discuss a lesser-known but important class of representation learning algorithms that are capable of learning higher-order features from data. The main idea is to learn relations between pixel intensities rather than the pixel intensities themselves by structuring the model as a tri-partite graph which connects hidden units to pairs of images. If the images are different, the hidden units learn how the images transform. If the images are the same, the hidden units encode within-image pixel covariances. Learning such higher-order features can yield improved results on recognition and generative tasks. I will discuss recent work on applying these methods to structured prediction problems. - Yang Wang, Univ. of Manitoba
"Recognizing and Localizing Novel Objects"
Abstract:
A lot of progress has been made in object recognition in the last few years. Now we have reasonably accurate systems that can recognize thousands of object categories. We also have good object detectors for a handful of object classes. However, since the number of object is so big and new object classes might emerge over time, it is not clear whether the standard supervised learning approach is the final solution for object recognition. In this talk, I will discuss our recent work on transfer learning for recognizing and localizing objects for which we do not have training data.
Object Detection
Fri. June 5, 11:00 AM- 12:00 PM- Sven Dickinson, Univ. of Toronto
"Detecting Symmetric Parts in Cluttered Scenes"
Abstract:
Perceptual grouping played a prominent role in support of early object recognition systems, which typically took an input image and a database of shape models and identified which of the models was visible in the image. Using intermediate-level shape priors, causally related shape features were grouped into discriminative shape indices that were used to prune the database down to a few promising candidates that might account for the query. In recent years, however, the recognition (categorization) community has focused on the object detection problem, in which the input image is searched for a specific target object. Since indexing is not required to select the target model, perceptual grouping is not required to construct a discriminative shape index. As a result, perceptual grouping activity at our major conferences has diminished. However, there are clear signs that the recognition community is moving from appearance back to shape, and from detection back to multi-class object categorization. Shape-based perceptual grouping will play a critical role in facilitating this transition. One of the most powerful mid-level shape priors is symmetry, which forms the basis for many approaches to part-based object modeling and recognition. In this talk, I will review our recent progress on detecting symmetric parts in cluttered scenes. - Sanja Fidler, Univ. of Toronto
"Understanding Complex Scenes and People That Talk about Them"
Abstract:
Language is an important link between high level semantic concepts and more low level visual perception. A successful robotic platform needs to both, understand the visual world and the lingual instructions given by the human user in order to react appropriately. In this talk, I'll present our recent work on 3D understanding of indoor scenes, and show how natural sentential descriptions can be exploited to improve 3D visual parsing.
Human-Robot Interaction and Assistive Tech
Fri. June 5, 4:00 PM- 5:00 PM- Dana Kulic, Univ. of Waterloo
"Human Motion Analysis for Rehabilitation"
Abstract:
Mobility improvement for patients is one of the primary concerns of physiotherapy rehabilitation. Providing the physiotherapist and the patient with a quantified and objective measure of progress can be beneficial for monitoring the patient's performance and providing guidance and feedback. In this talk, we describe a system for data collection, on-line pose estimation, segmentation and user interface for patients undergoing lower body rehabilitation. An approach for quantifying patient performance is also introduced. Results from multiple studies evaluating the system with patients undergoing rehabilitation following joint replacement surgery will be presented. - Babak Taati, Toronto Rehabilitation Institute
"Computer vision systems in dementia care"
Abstract:
Computer vision systems can play a role in providing care to individuals living with dementia. In this talk, I will first briefly review vision-based systems to provide assistance to older adults with dementia and to assist with usability studies for this population. I will then present preliminary results on assessing the cognitive status of older adults by way of monitoring common activities of daily living. Early identification of dementia can potentially lead to improved quality of life both for older adults with dementia and their family and caregivers who can better plan informal/formal care in advance.
Links to Previous Conferences
This page archives aa historical content from past CRV meetings. A second source for some of this information is maintained at the CIPPRS website.
- CRV 2014: Montreal, Quebec, May 7-9, 2014
- CRV 2013: Regina, Saskatchewan, May 28-31, 2013
- CRV 2012: Toronto, Ontario, May-28-30, 2012
- CRV 2011: St-John's, Newfoundland and Labrador, May-25-27, 2011
- CRV 2010: Ottawa, Ontario, 31 May-2 June 2010
- CRV 2009: Kelowna, British Columbia, 25-27 May 2009
- CRV 2008: Windsor, Ontario, 28-30 May 2008
- CRV 2007: Montreal, Quebec, 28-30 May 2007
- CRV 2006: Quebec City, Quebec, 7-9 May 2006
- CRV 2005: Victoria, British Columbia, 9-11 May 2005
- CRV 2004: London, Ontario, 17-19 May 2004.