Output details
11 - Computer Science and Informatics
Lancaster University
Spatio-temporal motion tracking with unsynchronized cameras
<23> We present the first markerless human motion capture algorithm that operates with heterogeneous and asynchronous cameras, e.g., smartphones. Our novel formulation for the likelihood of human motion in videos applies to heterogeneous frame rates, resolution, and asynchronous global temporal alignments. This is significant as it enables motion capture in uncontrolled environments where it is difficult to use conventional synchronised motion capture systems. Published at CVPR, one of three premier venues for computer vision (acceptance 465/1933). The work is at the core of “The Captury”, a spin-off that won three venture competitions including the prestigious CeBIT Start-up Competition.