|
|||||
|
|
Sequence of the most informative joints (SMIJ): A new representation for human skeletal action recognition |
|
| Institution: | 1. Tele-immersion Lab, University of California – Berkeley, Berkeley, CA, USA;2. Center for Imaging Sciences, Johns Hopkins University, Baltimore, MD, USA;1. Center for Imaging Science, Department of Biomedical Engineering, Johns Hopkins University, Baltimore, MD 21218, USA;2. Institute of Informatics and Applied Mathematics, University of Bern, 3012, Switzerland;1. University of Lille 1, Villeneuve d׳Ascq, France;2. LIFL Laboratory/UMR CNRS 8022, Villeneuve d׳Ascq, France;3. Institut Mines-Telecom/Telecom Lille, Villeneuve d׳Ascq, France;4. Florida State University, Department of Statistics, Tallahassee, USA |
| Abstract: | Much of the existing work on action recognition combines simple features with complex classifiers or models to represent an action. Parameters of such models usually do not have any physical meaning nor do they provide any qualitative insight relating the action to the actual motion of the body or its parts. In this paper, we propose a new representation of human actions called sequence of the most informative joints (SMIJ), which is extremely easy to interpret. At each time instant, we automatically select a few skeletal joints that are deemed to be the most informative for performing the current action based on highly interpretable measures such as the mean or variance of joint angle trajectories. We then represent the action as a sequence of these most informative joints. Experiments on multiple databases show that the SMIJ representation is discriminative for human action recognition and performs better than several state-of-the-art algorithms. |
| Keywords: | Human action representation Human action recognition Informative joints Bag-of-words Linear dynamical systems Normalized edit distance Cross-database generalization Berkeley MHAD HDM05 MSR Action3D |
| 本文献已被 ScienceDirect 等数据库收录！ | |