Video Coding, 3D Reconstruction/Rendering, and Visual-based Odometry/SLAM

Human Sensing

World’s top-level performance in extracting human body/face images taken by a camera; emotion estimation and recognition from the physical state of a person’s image taken by a camera

Human Detection

  • World’s top-level performance in detecting pedestrians and even those in wheelchairs

  • <Application> ADAS, autonomous driving

Skeleton Extraction / Physical State Recognition

  • Understanding the bodily state of a driver and a passenger from the change of their skeletal features

  • <Application> understanding driving operations, etc.

Human Tracking by Multiple Cameras
(Person Re-Identification)

  • Continuing the tracking of a specific person across multiple cameras

  • <Application> tracking a person-of-interest using surveillance cameras

Face Recognition

  • World’s top-level face recognition performance in both verification and identification

  • <Application> security solutions using surveillance cameras

Behavior Recognition

  • Understanding human behaviors like running, jumping, and kicking by employing the 3D imaging technology

Object Detection

Estimating the shape, volume, speed and other aspects of an object detected and tracked from images acquired by a camera

Object Detection (e.g. obstacle detection)

  • Extracting important objects from images using a stereo camera

  • <Application> vehicles, labels/signs/beacons

3D Shape Estimation/Recognition

  • Estimating the volume of a complex object from images using a camera

3D Object Projection Mapping

  • Exact projection mapping of a moving 3-dimensional object

Moving Object Detection

  • Extracting the direction and speed of a moving object from images using a camera ADAS

Object Tracking

  • Continuing tracking a moving object using a monocular camera

  • <Application> Autofocus on a moving object