Robust, low-cost, non-intrusive sensing and recognition of seated postures
Proceedings of the 20th annual ACM symposium on User interface software and technology
Social signal processing: state-of-the-art and future perspectives of an emerging domain
MM '08 Proceedings of the 16th ACM international conference on Multimedia
| Hi-index | 0.00 |
In this paper, we propose an algorithm for human postureand activity recognition for uncompressed and compressed(MPEG) video inputs. A real-time compression domain techniqueis developed to recognize different postures such asstanding, pointing left/right, opening arms, etc. by using aneigenspace representation of human silhouettes obtained fromAC-DCT coefficients. The system stores frames with specificpostures and finds global activity of the human body in thecompressed domain. In the uncompressed domain, this informationis used as an input for the activity/gesture recognitionalgorithm. First part of our approach is invariant to changesin intensity, color and textures and has the advantage of usingthe available data in the standard compression algorithms.Second part of the system can recognize activities in a set offrames starting with a recognized posture that is classified asa reference movement by the system. A prototype system isdeveloped with two camera nodes each consists of a standardcamera and a video processing board.