Microsoft Research

We propose a new 3D photo-real talking head with a personalized, photo realistic appearance. Different head motions and facial expressions can be freely controlled and rendered.

Most of us possess physical objects or heirlooms which connect us to our past and which provoke us to think about how we may someday pass on our own treasured objects to those we care about.

This talk will we will challenge the conventional wisdoms of how we build data center clusters.

Introducing how Kinect performs player identity recognition in Xbox, what we call 'Kinect Identity.' A primary challenge in developing a natural user interface for the Kinect is how to seamlessly

A quick guided introduction into key features of Time Series Foundation (TSF) - an open .NET toolset for exploring and prototyping new algorithms in time series analysis and forecasting.

Craig Mundie, Microsoft chief research and strategy officer, kicked off Microsoft ResearchΓÇÖs global 20th anniversary events at Microsoft Research Asia in Beijing, China.

HoloDesk is a novel interactive system combining an optical see through display and Kinect camera to create the illusion that users are directly interacting with 3D graphics.

Microsoft Researcher danah boydΓÇÖs research looks at the myths surrounding teens and privacy and the ways different teens reconcile technology with their public and private lives.

Handheld projector systems have the potential to enable users to dynamically augment environments with digital graphics.

Microsoft Researcher Alice Marwick explores the global fashion blogging subculture and what motivates the people interacting in this community.

This interactive demo illustrates a real-time gesture recognition system that is based on machine learning and running on top of the Kinect full body skeletal tracking pipeline.

Microsoft Research Asia senior researcher Lintao Zhang explains ΓÇ£Tiger,ΓÇ¥ the next-generation index serving platform for Bing, and shows a video describing the collaboration between Microsoft

Vermeer is a novel interactive 360┬░ viewable display suitable for a tabletop form factor.

The theory of conspicuous consumption is people buy goods to display their status and wealth to other people.

Microsoft Research Asia senior researcher Xin Tong performs a live demonstration of the Kinect-based Object Digitization technology enabling the creation of 3-D objects based on just a couple of

Non-Photorealistic Rendering describes algorithms for creating images and animations inspired by traditional media for art and illustration. I will present some of my work in this area.

Since the early 1970s, computer vision researchers have relied on concepts from physics, mathematics, and statistics to develop new approaches for many computer vision problems.

We present KinectFusion, a system that takes live depth data from a moving depth camera and in real-time creates high-quality 3D models.

Determining the pose and actions of humans is one of the central problems of image and video analysis.

By providing data sets like the PASCAL Object Recognition Database Collection and other similar data sets the computer vision community has provided an accepted benchmark that clearly defined a

The Kinect-based Object Digitization project enables the creation of 3-D objects based on just a couple of snapshots.

The use of categories to represent concepts (e.g., visual objects) is so prevalent in computer vision and machine learning that most researchers don't give it a second thought.

Microsoft Research Asia: 10 Years

Without a strong model, 3-D shape recovery of non-rigid surfaces from monocular video sequences is a severely under-constrained problem. Prior models are required to resolve the inherent ambiguities.

Bill Gates Congratulates Microsoft Research Asia on Turning 10

Motion Data Acquisition

Microsoft .NET Gadgeteer is an open-source toolkit for building small electronic devices using the .NET Micro Framework and Visual Studio or Visual C# Express (the free C# development environment).

Algorithms for learning the parameters of latent variable models are prone to getting stuck in a bad local optimum.

Activity recognition in videos is a key task in video content extraction; it is needed for applications such as monitoring and alerts, content-based indexing and human-computer interaction.

Our objective is to improve the performance of keyword based image search engines by re-ranking their baseline results. We address three limitations of existing search engines in this paper.