Issue No. 12 - Dec. (2013 vol. 35)
Jamie Shotton , Microsoft Res., Cambridge, UK
Ross Girshick , EERES-COENG Eng. Res., Univ. of California, Berkeley, Berkeley, CA, USA
Andrew Fitzgibbon , Microsoft Res., Cambridge, UK
Toby Sharp , Microsoft Res., Cambridge, UK
Mat Cook , Microsoft Res., Cambridge, UK
Mark Finocchio , Microsoft Corp., Redmond, WA, USA
Pushmeet Kohli , Microsoft Res., Cambridge, UK
Antonio Criminisi , Microsoft Res., Cambridge, UK
Alex Kipman , Microsoft Corp., Redmond, WA, USA
Andrew Blake , Microsoft Res., Cambridge, UK
We describe two new approaches to human pose estimation. Both can quickly and accurately predict the 3D positions of body joints from a single depth image without using any temporal information. The key to both approaches is the use of a large, realistic, and highly varied synthetic set of training images. This allows us to learn models that are largely invariant to factors such as pose, body shape, field-of-view cropping, and clothing. Our first approach employs an intermediate body parts representation, designed so that an accurate per-pixel classification of the parts will localize the joints of the body. The second approach instead directly regresses the positions of body joints. By using simple depth pixel comparison features and parallelizable decision forests, both approaches can run super-real time on consumer hardware. Our evaluation investigates many aspects of our methods, and compares the approaches to each other and to the state of the art. Results on silhouettes suggest broader applicability to other imaging modalities.
Pose estimation, Cameras, Human factors, Shape analysis, Feature extraction, Rendering (computer graphics)
J. Shotton et al., "Efficient Human Pose Estimation from Single Depth Images," in IEEE Transactions on Pattern Analysis & Machine Intelligence, vol. 35, no. 12, pp. 2821-2840, 2013.