Combining discriminative appearance and segmentation cues for articulated human pose estimation
Sam Johnson and Mark Everingham
In: 2nd IEEE International Workshop on Machine Learning for Vision-based Motion Analysis (MLVMA09), 28 Sep 2009, Kyoto, Japan.
We address the problem of articulated 2-D human pose estimation in unconstrained natural images. In previous work the Pictorial Structure Model approach has proven particularly successful, and is appealing because of its moderate computational cost. However, the accuracy of resulting pose estimates has been limited by the use of simple representations of limb appearance. We propose strong discriminatively trained limb detectors combining gradient and color segmentation cues. Our main contribution is a novel method for capturing coherent appearance properties of a limb using efficient color segmentation applied to every limb hypothesis during inference. The approach gives state-of-the-art results improving significantly on the “iterative image parsing” method, and shows significant promise for combination with other models of pose and appearance.
|EPrint Type:||Conference or Workshop Item (Oral)|
|Project Keyword:||Project Keyword UNSPECIFIED|
|Deposited By:||Mark Everingham|
|Deposited On:||08 March 2010|