Recently, the emergence of Kinect systems has demonstrated the benefits of predicting an intermediate body part labeling for 3D human pose estimation, in conjunction with RGB-D imagery. The availability of depth information plays a critical role, so an important question is whether a similar representation can be developed with sufficient robustness in order to estimate 3D pose from RGB images. This paper provides evidence for a positive answer, by leveraging (a) 2D human body part labeling in images, (b) second-order label-sensitive pooling over dynamically computed regions resulting from a hierarchical decomposition of the body, and (c) iterative structured-output modeling to contextualize the process based on 3D pose estimates. For robustness and generalization, we take advantage of a recent large-scale 3D human motion capture dataset, Human3.6M that also has human body part labeling annotations available with images. We provide extensive experimental studies where alternative intermediate representations are compared and report a substantial 33% error reduction over competitive discriminative baselines that regress 3D human pose against global HOG features.
C. Ionescu, J. Carreira, C. Sminchisescu. Iterated Second-Order Label-Sensitive Pooling for 3D Human Pose Estimation,
IEEE International Conference on Computer Vision and Pattern Recognition (CVPR), June 2014. pdf. The CVPR 2014 talk (with slides) can be seen here.
Code and Results
Feel free to contact us with questions, suggestions and/or bug reports.