ABSTRACT
This paper describes an unsupervised learning technique for modeling human locomotion styles, such as distinct related activities (e.g. running and striding) or variations of the same motion performed by different subjects. Modeling motion styles requires identifying the common structure in the motions and detecting style-specific characteristics. We propose an algorithm that learns a hierarchical model of styles from unlabeled motion capture data by exploiting the cyclic property of human locomotion. We assume that sequences with the same style contain locomotion cycles generated by noisy, temporally warped versions of a single latent cycle. We model these style-specific latent cycles as random variables drawn from a common "parent" cycle distribution, representing the structure shared by all motions. Given these hierarchical priors, the algorithm learns, in a completely unsupervised fashion, temporally aligned latent cycle distributions, each modeling a specific locomotion style, and computes for each example the style label posterior distribution, the segmentation into cycles, and the temporal warping with respect to the latent cycles. We demonstrate the flexibility of the model on several application problems such as style clustering, animation, style blending, and filling in of missing data.
- Brand, M., & Hertzmann, A. (2000). Style machines. Proc. of SIGGRAPH (pp. 183--192). Google ScholarDigital Library
- Chiappa, S., Kober, J., & Peters, J. (2009). Using bayesian dynamical systems for motion template libraries. In Adv. in Neural Inform. Proc. Systems 21, 297--304.Google Scholar
- Elgammal, A. M., & Lee, C.-S. (2004). Separating style and content on a nonlinear manifold. Proc. of Comp. Vision Pattern Recogn. (pp. 478--485).Google ScholarCross Ref
- Grochow, K., Martin, S. L., Hertzmann, A., & Popovićć, Z. (2004). Style-based inverse kinematics. ACM Trans. on Graphics, 23, 522--531. Google ScholarDigital Library
- Jordan, M. I., Ghahramani, Z., Jaakkola, T., & Saul, L. K. (1999). An introduction to variational methods for graphical models. Machine Learning, 37, 183--233. Google ScholarDigital Library
- Kovar, L., Gleicher, M., & Pighin, F. (2002). Motion graphs. ACM Trans. on Graphics, 21, 473--482. Google ScholarDigital Library
- Li, Y., Wang, T., & Shum, H.-Y. (2002). Motion texture: A two-level statistical model for character motion synthesis. ACM Trans. on Graphics, 21, 465--472. Google ScholarDigital Library
- Listgarten, J., Neal, R. M., Roweis, S. T., & Emili, A. (2005). Multiple alignment of continuous time series. In Adv. in Neural Inform. Proc. Systems 17, 817--824.Google Scholar
- Listgarten, J., Neal, R. M., Roweis, S. T., Puckrin, R., & Cutler, S. (2007). Bayesian detection of infrequent differences in sets of time series with shared structure. In Adv. in Neural Inform. Proc. Systems 19, 905--912.Google Scholar
- Liu, K., Hertzmann, A., & Popovic, Z. (2005). Learning physics-based motion style with nonlinear inverse optimization. ACM Trans. on Graphics, 24, 1071--1081. Google ScholarDigital Library
- Ormoneit, D., Black, M., Hastie, T., & Kjellströöm, H. (2005). Representing cyclic human motion using functional analysis. Image and Vision Comp., 1264--1276. Google ScholarDigital Library
- Rabiner, L. R. (1989). A tutorial on HMMs and selected applications in speech recognition. Proc. IEEE, 77. Google ScholarDigital Library
- Rose, C., Cohen, M., & Bodenheimer, B. (1998). Verbs and adverbs: multidimensional motion interpolation. IEEE Computer Graphics and Application, 18, 32--40. Google ScholarDigital Library
- Taylor, G. W., Hinton, G. E., & Roweis, S. T. (2007). Modeling human motion using binary latent variables. In Adv. in Neural Inform. Proc. Systems 19, 1345--1352.Google Scholar
- Torresani, L., Hackney, P., & Bregler, C. (2007). Learning motion style synthesis from perceptual observations. In Adv. in Neural Inform. Proc. Systems 19, 1393--1400.Google Scholar
- Urtasun, R., Fleet, D. J., Geiger, A., Popovic, J., Darrell, T., & Lawrence, N. D. (2008). Topologically-constrained latent variable models. Proc. Int. Conf. Machine Learning (pp. 1080--1087). Google ScholarDigital Library
- Wang, J. M., Fleet, D. J., & Hertzmann, A. (2007). Multi-factor gaussian process models for style-content separation. Proc. Int. Conf. Machine Learning (pp. 975--982). Google ScholarDigital Library
Index Terms
- Unsupervised hierarchical modeling of locomotion styles
Recommendations
Motion modeling for on-line locomotion synthesis
SCA '05: Proceedings of the 2005 ACM SIGGRAPH/Eurographics symposium on Computer animationIn this paper, we propose an example-based approach to on-line locomotion synthesis. Our approach consists of two parts: motion analysis and motion synthesis. In the motion analysis part, an unlabeled motion sequence is first decomposed into motion ...
Unsupervised Portrait Drawing Generation for Free Styles
Artistic portrait drawing (APDrawing) generation has seen progress in recent years. However, due to the naturally high scarcity and artistry, it is difficult to collect large-scale labeled and paired data and generally divide drawing styles into several ...
Unsupervised multi-modal modeling of fashion styles with visual attributes
AbstractFashion compatibility learning is of great practical significance to satisfy the needs of consumers and promote the development of the apparel industry. As a core task of it, fashion style modeling has received extensive attention. In ...
Highlights- The model builds a word bag for an item with a text feature and a visual feature..
Comments