Sameh Khamis
![Photo Photo](photo.jpg)
I am currently a Principal Research Scientist at Nvidia, mainly working on machine learning-based 3D content creation.
I was a Senior Research Scientist in Google AR/VR. I was also the tech lead for machine learning in ARCore [Video], where I worked on enabling AR on smartphones. I built the infrastructure for cross-dataset training and model distillation and compression. I also worked on efficient models for depth estimation from passive and active stereo, from motion, and from a single image, and I worked on human pose estimation, neural rendering, and image matting.
I was a Senior Scientist and a Founding Team Member at perceptiveIO. I built a state-of-the-art 360 performance capture system for real-time reconstruction of arbitrary non-rigid scenes. We were acquired by Google.
I was a Researcher in the Interactive 3D Technologies (I3D) group at Microsoft Research, where I worked on Holoportation [Video]. This work was covered by ZDNet, Gizmodo, VentureBeat, Mashable, Engadget, and Wired, among others. This technology was running live on stage at TED 2016. I also worked on hand tracking from depth cameras.
I received a PhD in computer science from the University of Maryland. I worked with Larry Davis at the Institute for Advanced Computer Studies (UMIACS). My thesis was on multi-person activity recognition from video sequences. I also interned at Siemens, IST Austria, MSR Cambridge, and SRI International.
I received a MSc in computer science from the University of Western Ontario, where I worked with Yuri Boykov. My thesis was on semantic image segmentation. I also wrote a parallel out-of-core push-relabel algorithm.
@inproceedings{valentin-siggraph-asia2018, author = {Julien Valentin and Adarsh Kowdle and Jonathan T Barron and Neal Wadhwa and Max Dzitsiuk and Michael Schoenberg and Vivek Verma and Ambrus Csaszar and Eric Turner and Ivan Dryanovski and Joao Afonso and Jose Pascoal and Konstantine Tsotsos and Mira Leung and Mirko Schmidt and Onur Guleryuz and Sameh Khamis and Vladimir Tankovitch and Sean Fanello and Shahram Izadi and Christoph Rhemann}, title = {Depth from Motion for Smartphone AR}, booktitle = {ACM SIGGRAPH Conference on Computer Graphics and Interactive Techniques in Asia}, year = {2018} }
@inproceedings{kowdle-siggraph-asia2018, author = {Adarsh Kowdle and Christoph Rhemann and Sean Fanello and Andrea Tagliasacchi and Jonathan Taylor and Philip Davidson and Mingsong Dou and Kaiwen Guo and Cem Keskin and Sameh Khamis and David Kim and Danhang Tang and Vladimir Tankovich and Julien Valentin and Shahram Izadi}, title = {The Need 4 Speed in Real-Time Dense Visual Tracking}, booktitle = {ACM SIGGRAPH Conference on Computer Graphics and Interactive Techniques in Asia}, year = {2018} }
@inproceedings{dou-siggraph2017, author = {Dou, Mingsong and Davidson, Philip and Fanello, Sean Ryan and Khamis, Sameh and Kowdle, Adarsh and Rhemann, Christoph and Tankovich, Vladimir and Izadi, Shahram}, title = {Motion2fusion: real-time volumetric performance capture}, booktitle = {ACM SIGGRAPH Conference on Computer Graphics and Interactive Techniques}, year = {2017} }
@inproceedings{orts-escolano-uist2016, author = {Sergio Orts-Escolano and Christoph Rhemann and Sean Fanello and Wayne Chang and Adarsh Kowdle and Yury Degtyarev and David Kim and Philip Davidson and Sameh Khamis and Mingsong Dou and Vladimir Tankovich and Charles Loop and Qin Cai and Philip Chou and Sarah Mennicken and Julien Valentin and Vivek Pradeep and Shenlong Wang and Sing Bing Kang and Pushmeet Kohli and Yuliya Lutchyn and Cem Keskin and Shahram Izadi}, title = {Holoportation: Virtual 3D Teleportation in Real-time}, booktitle = {ACM User Interface Software and Technology Symposium}, year = {2016} }
@inproceedings{dou-siggraph2016, author = {Mingsong Dou and Sameh Khamis and Yury Degtyarev and Philip Davidson and Sean Fanello and Adarsh Kowdle and Sergio Orts Escolano and Christoph Rhemann and David Kim and Jonathan Taylor and Pushmeet Kohli and Vladimir Tankovich and Shahram Izadi}, title = {Fusion4D: Real-time Performance Capture of Challenging Scenes}, booktitle = {ACM SIGGRAPH Conference on Computer Graphics and Interactive Techniques}, year = {2016} }
@inproceedings{taylor-siggraph2016, author = {Jonathan Taylor and Lucas Bordeaux and Thomas Cashman and Bob Corish and Cem Keskin and Eduardo Soto and David Sweeney and Julien Valentin and Benjamin Luff and Arran Topalian and Erroll Wood and Sameh Khamis and Pushmeet Kohli and Toby Sharp and Shahram Izadi and Richard Banks and Andrew Fitzgibbon and Jamie Shotton}, title = {Efficient and Precise Interactive Hand Tracking through Joint, Continuous Optimization of Pose and Correspondences}, booktitle = {ACM SIGGRAPH Conference on Computer Graphics and Interactive Techniques}, year = {2016} }
@inproceedings{tan-cvpr2016, author = {David Joseph Tan and Thomas Cashman and Jonathan Taylor and Andrew Fitzgibbon and Daniel Tarlow and Sameh Khamis and Shahram Izadi and Jamie Shotton}, title = {Fits Like a Glove: Rapid and Reliable Hand Shape Personalization}, booktitle = {IEEE Conference on Computer Vision and Pattern Recognition}, year = {2016} }
@inproceedings{khamis-cvpr2015, author = {Sameh Khamis and Jonathan Taylor and Jamie Shotton and Cem Keskin and Shahram Izadi and Andrew Fitzgibbon}, title = {Learning an Efficient Model of Hand Shape Variation from Depth Images}, booktitle = {IEEE Conference on Computer Vision and Pattern Recognition}, year = {2015} }
@inproceedings{khamis-cvprw2015, author = {Sameh Khamis and Larry S. Davis}, title = {Walking and Talking: A Bilinear Approach to Multi-Label Action Recognition}, booktitle = {CVPR Workshop on Group And Crowd Behavior Analysis And Understanding}, year = {2015} }
@inproceedings{khamis-bmvc2014, author = {Sameh Khamis and Christoph H. Lampert}, title = {CoConut: Co-Classification with Output Space Regularization}, booktitle = {British Machine Vision Conference}, year = {2014} }
@inproceedings{khamis-eccvw2014, author = {Sameh Khamis and Cheng-Hao Kuo and Vivek K. Singh and Vinay Shet and Larry S. Davis}, title = {Joint Learning for Attribute-Consistent Person Re-Identification}, booktitle = {ECCV Workshop on Visual Surveillance and Re-Identification}, year = {2014} }
@inproceedings{kuo-wacv2013, author = {Cheng-Hao Kuo and Sameh Khamis and Vinay Shet}, title = {Person Re-identification using Semantic Color Names and RankBoost}, booktitle = {IEEE Workshop on the Applications of Computer Vision}, year = {2013} }
@inproceedings{london-cvprw2013, author = {Ben London and Sameh Khamis and Stephen H. Bach and Bert Huang and Lise Getoor and Larry S. Davis}, title = {Collective Activity Detection using Hinge-loss Markov Random Fields}, booktitle = {CVPR Workshop on Structured Prediction: Tractability, Learning and Inference}, year = {2013} }
@inproceedings{khamis-eccv2012, author = {Sameh Khamis and Vlad I. Morariu and Larry S. Davis}, title = {Combining Per-Frame and Per-Track Cues for Multi-Person Action Recognition}, booktitle = {European Conference on Computer Vision}, year = {2012} }
In the published version, equation (3) should use the distance directly, and not its pseudo-probability: v_{ab}(i, j) = \lambda_d\ d(i, j) - \lambda_c \log(p_{ab})
@inproceedings{khamis-cvpr2012, author = {Sameh Khamis and Vlad I. Morariu and Larry S. Davis}, title = {A Flow Model for Joint Action Recognition and Identity Maintenance}, booktitle = {IEEE Conference on Computer Vision and Pattern Recognition}, year = {2012} }
In the published version, equation (3) should use the distance directly, and not its pseudo-probability: v_{ab}(i, j) = \lambda_d\ d(i, j) - \lambda_c \log(p_{ab})