Header logo is ps

{FlowCap}: {2D} Human Pose from Optical Flow


Conference Paper


We estimate 2D human pose from video using only optical flow. The key insight is that dense optical flow can provide information about 2D body pose. Like range data, flow is largely invariant to appearance but unlike depth it can be directly computed from monocular video. We demonstrate that body parts can be detected from dense flow using the same random forest approach used by the Microsoft Kinect. Unlike range data, however, when people stop moving, there is no optical flow and they effectively disappear. To address this, our FlowCap method uses a Kalman filter to propagate body part positions and ve- locities over time and a regression method to predict 2D body pose from part centers. No range sensor is required and FlowCap estimates 2D human pose from monocular video sources containing human motion. Such sources include hand-held phone cameras and archival television video. We demonstrate 2D body pose estimation in a range of scenarios and show that the method works with real-time optical flow. The results suggest that optical flow shares invariances with range data that, when complemented with tracking, make it valuable for pose estimation.

Author(s): Romero, Javier and Loper, Matthew and Black, Michael J.
Book Title: Pattern Recognition, Proc. 37th German Conference on Pattern Recognition (GCPR)
Volume: LNCS 9358
Pages: 412--423
Year: 2015
Publisher: Springer

Department(s): Perceiving Systems
Research Project(s): 2D Pose from Optical Flow
Bibtex Type: Conference Paper (inproceedings)
Paper Type: Conference

Event Name: GCPR 2015
Event Place: Aachen

Links: video
Attachments: pdf preprint


  title = {{FlowCap}: {2D} Human Pose from Optical Flow},
  author = {Romero, Javier and Loper, Matthew and Black, Michael J.},
  booktitle = {Pattern Recognition, Proc. 37th German Conference on Pattern Recognition (GCPR)},
  volume = {LNCS 9358},
  pages = {412--423},
  publisher = {Springer},
  year = {2015}