We are an inter-disciplinary team of researchers working in visual computing, in particular, computer graphics and computer vision. Current areas of focus include 3D and robotic vision, 3D printing and content creation, animation, AR/VR, geometric and image-based modelling, machine learning, natural phenomenon, and shape analysis. Our research works frequently appear in top venues such as SIGGRAPH, CVPR, and ICCV (we rank #11 in the world in terms of top publications in visual computing, as of 7/2020) and we collaborate widely with the industry and academia (e.g., Adobe Research, Google, MSRA, Princeton, Stanford, and Washington). Our faculty and students have won numerous honours and awards, including FRSC, Alain Fournier Best Thesis Award, Google Faculty Award, TR35@Singapore, NSERC Discovery Accelerator, and several best paper awards from ECCV, SCA, SGP, etc. Gruvi alumni went on to take up faculty positions in Canada, the US, and Asia, while others now work at companies including Apple, EA, Facebook, Google, IBM, and Microsoft.
October 10, 2020
Richard Zhang will deliver one of the keynote talks (virtually) at ChinaGraph 2020 held in Xiamen, China, on October 23. ChinaGraph is a bi-annual conference on computer graphics in China, the most important gathering of graphics researchers, students, and industries in the country. In another news, 3D-FRONT, a large-scale 3D indoor scene dataset published earlier this year by Richard and colleagues from Alibaba and the Chinese Academy of Sciences, has won the inaugural ChinaGraph Best Dataset Award.
October 10, 2020
Richard Zhang will deliver one of the keynote talks (virtually) at ChinaGraph 2020 held in Xiamen, China, on October 23. ChinaGraph is a bi-annual conference on computer graphics in China, the most important gathering of graphics researchers, students, and industries in the country. In another news, 3D-FRONT, a large-scale 3D indoor scene dataset published earlier this year by Richard and colleagues from Alibaba and the Chinese Academy of Sciences, has won the inaugural ChinaGraph Best Dataset Award.
September 25, 2020
Congratulations to Angel Xuan Chang, Yasutaka Furukawa and Manolis Savva for receiving fundings form Canada Foundation for Innovation (CFI). This funding will allow our researchers to take their transformative discoveries to the next level. More information about the funding and the projects can be found here here.
September 25, 2020
Congratulations to Angel Xuan Chang, Yasutaka Furukawa and Manolis Savva for receiving fundings form Canada Foundation for Innovation (CFI). This funding will allow our researchers to take their transformative discoveries to the next level. More information about the funding and the projects can be found here here.
September 23, 2020
Title: Category-Level Object Perception for Physical Interaction
Time: 1:30 - 2:30, Wednesday, September 23
Abstract: Deep neural networks have shown great success both in semantic perception tasks, e.g. object recognition and semantic segmentation, and in end-to-end perception for reinforcement learning and robotic tasks. However, it is still unclear how to bridge these two perception paradigms to gain a deep semantic and interaction-driven understanding of physical interaction.
In this talk, I will focus on how to explore categorical actionable information for the sake of perceiving and understanding physical interactions. First, I will show that learning high-level semantic actionable information, e.g. object state, can help with action planning. Second, I will introduce the problem of estimating category-level 6D pose and 3D size for rigid objects. This category pose can be seen as low-level actionable information and can benefit object manipulation tasks. Lastly, I will present my recent works on curating an articulated object dataset and estimating category-level articulated object pose.
Bio: He Wang is a fifth-year PhD student at Stanford University under the supervision of Prof. Leonidas Guibas. His research interests span across computer vision, geometric computing, and robotics. In his PhD he contributed to generative modeling of human object interactions, opened up a new direction in estimating category-level pose and size for rigid and articulated objects. He receives Eurographics 2019 best paper honorable mention award and three of his works are accepted as CVPR oral presentations. Prior to his PhD he obtained his bachelor in Microelectronics from Tsinghua University.
September 23, 2020
Title: Category-Level Object Perception for Physical InteractionTime: 1:30 - 2:30, Wednesday, September 23Abstract: Deep neural networks have shown great success both in semantic perception tasks, e.g. object recognition and semantic segmentation, and in end-to-end perception for reinforcement learning and robotic tasks. However, it is still unclear how to bridge these two perception paradigms to gain a deep semantic and interaction-driven understanding of physical interaction.In this talk, I will focus on how to explore categorical actionable information for the sake of perceiving and understanding physical interactions. First, I will show that learning high-level semantic actionable information, e.g. object state, can help with action planning. Second, I will introduce the problem of estimating category-level 6D pose and 3D size for rigid objects. This category pose can be seen as low-level actionable information and can benefit object manipulation tasks. Lastly, I will present my recent works on curating an articulated object dataset and estimating category-level articulated object pose.Bio: He Wang is a fifth-year PhD student at Stanford University under the supervision of Prof. Leonidas Guibas. His research interests span across computer vision, geometric computing, and robotics. In his PhD he contributed to generative modeling of human object interactions, opened up a new direction in estimating category-level pose and size for rigid and articulated objects. He receives Eurographics 2019 best paper honorable mention award and three of his works are accepted as CVPR oral presentations. Prior to his PhD he obtained his bachelor in Microelectronics from Tsinghua University.
September 13, 2020
Kevin (Kai) Xu, a former gruvier from 2008 to 2010, when he worked as a visiting Ph.D. student under the supervision of Richard Zhang, is now resuming his affiliation with GrUVi after being appointed as an Adjunct Professor at SFU CS for a three-year term. Welcome back to GrUVi, Kevin!
Check his website for more information about his recent works in computer graphics and computer vision!
September 13, 2020
Kevin (Kai) Xu, a former gruvier from 2008 to 2010, when he worked as a visiting Ph.D. student under the supervision of Richard Zhang, is now resuming his affiliation with GrUVi after being appointed as an Adjunct Professor at SFU CS for a three-year term. Welcome back to GrUVi, Kevin!Check his website for more information about his recent works in computer graphics and computer vision!
July 8, 2020
Congratulations to Angel Xuan Chang and Manolis Savva, whose work ScanNet: Richly-annotated 3D Reconstructions of Indoor Scenes won the 2020 SGP dataset award. ScanNet is an RGB-D video dataset containing 2.5 million views in more than 1500 scans, annotated with 3D camera poses, surface reconstructions, and instance-level semantic segmentations. More information can be found in the paper here.
Jun 23, 2020
Visual computing researchers from SFU received multiple awards at the annual Conference on Computer Vision and Pattern Recognition (CVPR) this past week. CVPR is the premier conference in computer vision with the highest impact factor among all conferences in computer science and was held virtually for the first time this year from June 14-19. Computing science professor Greg Mori served as one of four program chairs at the conference.
Zhiqin Chen and Richard Zhang, together with GrUVi alumnus Andrea Tagliasacchi, won the Best Student Paper Award with "BSP-Net: Generating Compact Meshes via Binary Space Partitioning" introduces a deep neural network which applies a classical graphics technique to learn compact shape representations.
Yasutaka Furukawa won the PAMI Longuet-Higgins Prize for his CVPR 2007 milestone paper on “multi-view stereo reconstruction”, which has been cited more than 3,000 times!
Also, Akshay Gadi Patil won the Best Paper Award at the CVPR Workshop on Text and Documents in the Deep Learning Era for his work "READ: Recursive Autoencoders for Document Layout Generation".
Read more here. Well done, gruviers!
Jun 09, 2020
CVPR, the premier conference on computer vision, will be held virtually next week (June 16-20). SFU Computing Science professor Greg Mori is front-n-center as a Program Chair of the major event! GrUVi lab will have an incredible showing at CVPR, with 11 technical papers (5 orals), 3 invited talks, and 4 co-organized workshops!
Workshop co-organization
GrUViers will co-organize 4 workshops featuring state-of-the-art research:
Invited workshop talks
Yasutaka Furukawa will give a talk at the aforementioned “ScanNet Indoor Scene Understanding Challenge” as well as the “3D Scene Understanding for Vision, Graphics, and Robotics” workshop , while Manolis Savva will participate as an invited speaker at the “Fair, Data, Efficient and Trusted Computer Vision” workshop.
Technical Papers and GrUVi authors
Congratulations to all authors, especially to Dr. Ping Tan, who got five accepted papers! The full list of papers featured on CVPR 2020 can be accessed here. In particular, GrUVi papers covered different topics:
Jun 07, 2020
SFU researchers helped to develop an AI system capable of assisting resident and less experienced doctors look over a data set and make a quick diagnosis before a senior doctor can step in. This is accoding to Yağız Aksoy, a gruvier and also a member of the team that proposed the diagnosis tool, which is currently in the validation phase at St. Paul’s Hospital in Vancouver, Canada. Read more about it here. Thank you for your hard work, Dr. Aksoy!
Oct 24, 2019
Congratulations to Kangxue, whose work LOGAN: Unpaired Shape Transform in Latent Overcomplete Space will be featured for press release at SIGGRAPH Asia 2019!
LOGAN is able to learn what shape features to preserve during shape translation, either local or non-local, whether content or style, depending solely on the input domains for training, which was an improvement over the state of the art.
LOGAN is an abbreviation for Latent Overcomplete GAN. Richard Zhang, who is also the last author, created the fun acronym.
Jul 23, 2019
The International Geometry Summit (IGS) 2019, organized by Richard Zhang and Ali Mahdavi-Amiri, took place from 17th June to the 21th June 2019 at SFU Harbour Center in Vancouver, Canada. IGS 2019 included four conferences: the Symposium on Solid and Physical Modelling, Shape Modeling International, SIAM Conference on Computational Geometric Design, and International Conference on Geometric Modelling and Processing. IGS 2019 featured state of the art researches in its technical paper sessions along with 12 keynote speeches from prominent researchers in the area of geometry.
Congratulations to Richard Zhang and Ali Mahdavi-Amiri for the successful organization and thanks to all GrUVi students that volunteered to make IGS'19 a success!
Jul 22, 2019
Richard Zhang was featured in the Eurographics Symposium on Geometry Processing (SGP) as a keynote speaker! His speech was titled “Can Machines Learn to Generate 3D Shapes?”.
In his talk, Richard highlighted the representation, data, and output challenges that researchers must tackle and how his research has shaped itself to address these challenges.
One of the key issues in this domain is not generate shapes that look right; instead, they need to serve their intended function while displaying the right part connections, arrangements, and geometry.
Jul 9, 2019
Dr. Xiaoming Liu visited SFU today following an invitation from Dr. Yasutaka Furukawa! His talk titled "Inverse Graphics for 3D Modeling and Reconstruction: from Face to Generic Objects" has the following abstract:
"Reconstructing the detailed and complete 3D surface of an object from a single 2D image is a long standing computer vision problem. In this talk, we present an inverse graphics-based framework to learn a 3D face model from a large set of in-the-wild 2D face images, without the need of capturing 3D scans. We also extend this framework to high-fidelity face modeling, as well as generic object modeling and 3D reconstruction. In the end, we will briefly overview other research efforts in the Computer Vision Lab at Michigan State University, including face anti-spoofing, explainable recognition, early sensor fusion of LiDAR and RGB, 2D/3D object detection from urban driving videos, etc."
Jul 3, 2019
GrUVi is glad to announce that Yagiz Aksoy will join our team this August. He is a PhD student of Marc Pollefeys at ETH Zurich. During his PhD, he spent a year at MIT CSAIL working with Wojciech Matusik and three years at Disney Research Zurich.
His research is at the intersection of computer vision and computer graphics, focusing on analyzing images to allow for realistic manipulation of photographs. His research has resulted in publications at venues such as SIGGRAPH, CVPR, and ACM Transactions on Graphics.
Welcome to the GrUVi team, Yagiz!
