*New for this year, vehicle demos of actual vehicles from Daimler and V-Charge in the parking lot.
|09:00-09:05||Opening notes from Workshop Organizers|
|09:05-09:45||Invited Speaker: Vision for Low-Cost Autonomy with the Oxford University RobotCar, Will Maddern, Oxford University, United Kingdom.|
|09:45-10:45||Contributed works (15min per talk)|
|09:45-10:00||Ten Years of Pedestrian Detection, What Have We Learned?, Rodrigo Benenson, Mohamed Omran, Jan Hosang, Bernt Schiele|
|10:00-10:15||Fast 3-D Urban Object Detection on Streaming Point Clouds, Attila Börcs, Balázs Nagy, Csaba Benedek|
|10:15-10:30||Relative Pose Estimation and Fusion of Omnidirectional and Lidar Cameras, Levente Tamas, Robert Frohlich, Zoltan Kato|
|10:30-10:45||Good Edgels To Track: Beating The Aperture Problem With Epipolar Geometry,Tommaso Piccini, Mikael Persson, klas Nordberg, Michael Felsberg, Rudolf Mester|
|11:15-11:55||Invited Speaker: Localization in Urban Canyons using Cadastral 3D City Models, Srikumar Ramalingam, MERL, USA.|
|11:55-12:15||Demo talk: Stixmantics: Real-time semantic segmentation of street scenes, Uwe Franke / Timo Scharwächter, Daimler, Germany.|
|14:00-14:40||Invited Speaker: Is the self-driving car around the corner? Mobileye's work on Computer Vision centric approach to self-driving at consumer level cost, Amnon Shashua, MobilEye, Israel.|
|14:40-15:00||Demo talk: Multi-Camera Systems in the V-Charge Project: Fundamental Algorithms, Self Calibration, and Long-Term Localization, Paul Furgale, Vcharge ETH Zurich, Switzerland.|
|15:00-15:40||Invited Speaker: Intelligent Drive & Pedestrian Safety 2.0, Dariu Gavrila, Daimler, Germany.|
|16:00-18:00||Posters / Demos of actual vehicles from Daimler and V-Charge in the parking lot.|
Invited Speakers / Demos
- Uwe Franke / Timo Scharwächter, Daimler, Germany Stixmantics: Real-time semantic segmentation of street scenes
- Paul Furgale, Vcharge ETH Zurich, Switzerland Multi-Camera Systems in the V-Charge Project: Fundamental Algorithms, Self Calibration, and Long-Term Localization
- Prof. Dr. D. M. Gavrila. Daimler, Germany Intelligent Drive & Pedestrian Safety 2.0
- Will Maddern, Oxford University, United Kingdom Vision for Low-Cost Autonomy with the Oxford University RobotCar
- Srikumar Ramalingam, MERL, USA Localization in Urban Canyons using Cadastral 3D City Models
- Amnon Shashua, MobilEye, Israel Is the self-driving car around the corner? Mobileye's work on Computer Vision centric approach to self-driving at consumer level cost Bio: Prof. Amnon Shashua holds the Sachs chair in computer science at the Hebrew University of Jerusalem. His field of expertise is computer vision and machine learning. For his academic achievements he received the MARR prize Honorable Mention in 2001, the Kaye innovation award in 2004, and the Landau award in exact sciences in 2005. He is the co-founder of Mobileye, an Israeli company developing systems-on-chip and computer vision algorithms for detecting pedestrians, vehicles, and traffic signs for driving assistance systems. He is the cofounder of OrCam, an Israeli company who recently launched an assistive product for the visually impaired based on advanced computerized visual interpretation capabilities.
We will provide a live demonstration of multi-class semantic scene segmentation of street scenes. Images are captured using a front-facing vehicle-mounted stereo camera system. The demonstrated system discriminates vehicles, pedestrians, buildings, ground surface and sky, we will be able to offer short rides to workshop attendees.Bios: Timo Scharwächter, graduated from RWTH Aachen in 2012 with a focus on machine learning and computer vision and currently works as a PhD student in Dr. Franke's Image Understanding group at Daimler R&D, Sindelfingen, Germany. Uwe Franke received the Ph.D. degree in electrical engineering from the Technical University of Aachen, Germany in 1988. Since 1989 he has been with Daimler Research and Development and has been constantly working on the development of vision based driver assistance systems. Since 2000 he has been head of Daimler's Image Understanding Group and is a well known expert in real-time stereo vision and image understanding. Recent work is on optimal fusion of stereo and motion, called 6D-Vision. The stereo technology developed by his group is the basis for the stereo camera system of the new Mercedes S- and E-class vehicles introduced in 2013. Besides fully autonomous emergency breaking these cars offer autonomous driving in traffic jams.
Cameras will play an important role in the commercialization of autonomous driving; they provide rich information about the three-dimensional structure and appearance of the environment at the fraction of the cost of laser scanners or radar. While there has been excellent work using single cameras or stereo pairs in driving assistance systems, the use of multi-camera systems has been limited because of the lack of available standard tools. This talk describes our efforts to close this gap in three fundamental areas. First, we discuss our development of algorithms for treating data from arbitrary multi-camera systems, including several novel minimal and N-point methods for both relative and absolute pose, as well as OpenGV, an open source library that implements these methods. Second, we describe CamOdoCal, open-source software for self calibration for autonomous cars. The library includes both full self-calibration as well as a novel low-computational-cost unsupervised calibration method that can potentially scale to fleets of autonomous vehicles. Finally, we show how map data can be managed over longer time scales, updated with new data, and used for localization over a wide range of lighting and weather conditions.Bio: Paul Furgale is the Deputy Director of the Autonomous Systems Lab at the Swiss Federal Institute of Technology in Zurich (ETH Zurich). His current research is focused on long-term autonomy for mobile robotic systems, including perception, mapping, localization, and planning over long timescales and in highly dynamic environments. He is the scientific coordinator for V-Charge, a European project and industry/academic collaboration that seeks to develop automated valet parking and charging of electric vehicles in mixed traffic. He received a PhD (2011) from the University of Toronto Institute for Aerospace Studies (UTIAS) where he developed algorithms to support over-the-horizon sample return for planetary exploration rovers as part of the Autonomous Space Robotics Lab. His PhD work was field tested in the Canadian High Arctic and subsequently integrated into several Canadian Space Agency rover prototypes.
Daimler has introduced an advanced set of driver assistance functions in its novel Mercedes-Benz 2013-2014 S-, E-, and C-Class models, termed Intelligent DriveŁ, using stereo vision. It includes a pedestrian safety component which facilitates fully automatic emergency braking - the system works day and night. I first provide an overview of Intelligent Drive, covering the stereo vision-based functions in particular. I thereafter focus on our research towards developing the next-generation active pedestrian safety systems. These systems extract higher-level visual cues and use more sophisticated motion models for path prediction. They have the potential to react earlier in dangerous traffic situations, without increasing the false alarms. I describe a prototype system that takes into account pedestrian situational awareness and traffic scene context. Interested talk attendees will be offered a liveŁ vehicle demo in a separate time slot.Bio: Dariu M. Gavrila received the PhD degree in computer science from the University of Maryland at College Park, USA, in 1996. Since 1997, he has been with Daimler R&D in Ulm, Germany, where he is currently a Principal Scientist. In 2003, he was further appointed professor at the University of Amsterdam, chairing the area of Intelligent Perception Systems (part time). Over the past 15 years, Prof. Gavrila has focused on visual systems for detecting humans and their activity, with application to intelligent vehicles, smart surveillance and social robotics. He led the multi-year pedestrian detection research effort at Daimler, which was incorporated in the Mercedes-Benz S-, E-, and C-Class models (2013-2014). He is frequently cited in the scientific literature and he received the I/O 2007 Award from the Netherlands Organization for Scientific Research (NWO) as well as several conference paper awards.
Robust and reliable localisation and perception at any time of day is an essential component towards widespread adoption of autonomous road vehicles. However, typical solutions to this problem involve active 3D sensors (such as Velodyne LIDARs) which dramatically increases the cost of the sensor suite. In this talk I will provide an overview of methods developed at the Oxford Mobile Robotics Group to perform calibration, localisation, mapping, navigation and obstacle detection using low-cost cameras and 2D LIDARs towards the goal of low-cost autonomy for road vehicles. In particular I will cover our demonstration of autonomous driving on UK roads using a £5000 sensor suite, and our recent experiments towards vision-only autonomous driving on public roads.Bio: Will Maddern is a postdoctoral researcher in the Oxford Mobile Robotics Group and flagship leader for the Oxford University RobotCar project (www.robotcar.org.uk). He manages the technical aspects of autonomous driving; primarily localisation, mapping and navigation using vision and laser, along with path planning, control and obstacle perception and avoidance. Prior to joining Oxford, Will completed a PhD in appearance-based localisation and mapping at QUT, Australia.
In recent years, cadastral 3D models have been created for many cities around the world. We show a method to register images captured by a car-mounted camera with the cadastral 3D city models. This allows us to get accurate position in GPS challenged urban canyons. Our proposed solution takes a sequence of upward facing omnidirectional images and coarse 3D models of cities to compute the geo-trajectory. The camera is oriented upwards to capture images of the immediate skylines, which is generally unique and serves as a fingerprint for a specific location in a city. Our goal is to estimate position by matching skylines extracted from omni-directional images to skyline segments from coarse 3D city models. Under day-time, we propose a sky-segmentation algorithm using graph cuts for estimating the geo-location. In cases where the skyline gets affected by partial fog, night-time and occlusions from trees, we propose a shortest path algorithm that computes the location without prior sky detection. I will also briefly highlight some of our other results in pose/motion estimation, dense 3D reconstruction and obstacle detection.Bio: Srikumar Ramalingam is a Principal Research Scientist at Mitsubishi Electric Research Lab (MERL) since 2008. He received his B.E from Anna University in India and his M.S from University of California (Santa Cruz) in USA. He received a Marie Curie Fellowship from European Union to pursue his studies at INRIA Rhone Alpes (France) and he obtained his PhD in 2007. His thesis on generic imaging models received INPG best thesis prize and AFRIF thesis prize (honorable mention) from the French Association for Pattern Recognition. He has published numerous papers in flagship conferences such as CVPR, ICCV, SIGGRAPH ASIA and ECCV. He has coauthored books, given tutorials and organized workshops on topics such as multi-view geometry and graphical models. His research interests are in computer vision, machine learning, robotics and autonomous driving.
Topics of Interest
Analyzing road scenes using cameras could have a crucial impact in many domains, such as autonomous driving, advanced driver assistance systems (ADAS), personal navigation, mapping of large scale environments, and road maintenance. For instance, vehicle infrastructure, signage, and rules of the road have been designed to be interpreted fully by visual inspection. As the field of computer vision becomes increasingly mature, practical solutions to many of these tasks are now within reach. Nonetheless, there still seems to exist a wide gap between what is needed by the automotive industry and what is currently possible using computer vision techniques. The goal of this workshop is to allow researchers in the fields of road scene understanding and autonomous driving to present their progress and discuss novel ideas that will shape the future of this area. In particular, we would like this workshop to bridge the large gap between the community that develops novel theoretical approaches for road scene understanding and the community that builds working real-life systems performing in real-world conditions. To this end, we encourage submissions of original and unpublished work in the area of vision-based road scene understanding. The topics of interest include (but are not limited to):
- Prediction and modeling of road scenes and scenarios
- Semantic labeling, object detection and recognition in road scenes
- Dynamic 3D reconstruction, SLAM and ego-motion estimation
- Visual feature extraction, classification and tracking
- Processing for prosthetic (bionic) vision and low-vision assistive devices
- Design and development of robust and real-time architectures
- Use of emerging sensors (e.g., multispectral, RGB-D, LIDAR and LADAR)
- Fusion of RGB imagery with other sensing modalities
- Interdisciplinary contributions across computer vision, optics, robotics and other related fields.
We encourage researchers to submit not only theoretical contributions, but also work more focused on applications. Each paper will receive 3 double blind reviews, which will be moderated by the workshop chairs.
- Submission Deadline: July 5 (Extended!).
- Notification of Acceptance: July 20.
- Camera-ready Deadline: July 25.
- Workshop: September 7.
- Bart Nabbe (Toyota, USA)
- Mathieu Salzmann (NICTA, Australia)
- Lars Petersson (NICTA, Australia)
- Jose Alvarez (NICTA, Australia)
- Raquel Urtasun (University of Toronto, Canada)
- Fatih Porikli (NICTA, Australia)
- Gary Overett (NICTA, Australia)
- Nick Barnes (NICTA, Australia)
- H. Badino, National Robotics Engineering Center, USA
- R. Benenson, MPI Saarbruecken, Germany
- P. Borges, CSIRO Brisbane, Australia
- M. Brubaker, TTI Chicago, USA
- S. Fidler, TTI Chicago, USA
- M. Fritz, MPI Saarbruecken, Germany
- A. Geiger, MPI Tuebingen, Germany
- S. Gould, Australian National University, Australia
- X. He, NICTA, Australia
- H. Kong, MIT, USA
- A. Lapedriza, Universitat Oberta de Catalunya, Spain
- B. Leibe, RWTH Aachen University, Germany
- P. Lenz, Karlsruhe Institute of Technology, Germany
- Y. Li, NICTA, Australia
- S. Narasimhan, Carnegie Mellon University, USA
- U. Nunes, University of Coimbra, Portugal
- C. Shen, University of Adelaide, Australia
- M.A. Sotelo, University of Alcala, Spain
- G. Zen, University of Trento, Italy
- G. Dedeoglu, PercepTonic, USA
- U. Franke, Daimler, Germany
- J. Fritsch, Honda Research Institute, Germany
- R. Hammoud, BAE Systems, USA
- B. Heisele, Honda Research Institute, USA
- A. Khiat, Nissan, Japan
- A. Krainov, Yandex, Russia
- D. Langer, ASCar, USA
- D. Levi, General Motors, USA
Papers should describe original and unpublished work about the above or closely related topics. Each paper will receive double blind reviews, moderated by the workshop chairs. Authors should take into account the following:
- All papers must be written in English and submitted in PDF format.
- Papers must be submitted online through the CMT submission system. The submission site is: https://cmt2.research.microsoft.com/CVRSUAD2014.
- The maximum paper length is 12 pages. Note that shorter submissions are also welcome. The workshop paper format guidelines are the same as the Main Conference papers.
- Submissions will be rejected without review if they: contain more than 12 pages, violate the double-blind policy or violate the dual-submission policy. The author kit provides a LaTeX2e template for submissions, and an example paper to demonstrate the format. Please refer to this example for detailed formatting instructions.
- A paper ID will be allocated to you during submission. Please replace the asterisks in the example paper with your paper's own ID before uploading your file. More detailed instructions can be found at the main conference website.