Human motion analysis: a review
Computer Vision and Image Understanding
On Intelligence
Using SVM for Efficient Detection of Human Motion
ICCCN '05 Proceedings of the 14th International Conference on Computer Communications and Networks
How the brain might work: a hierarchical and temporal model for learning and recognition
How the brain might work: a hierarchical and temporal model for learning and recognition
| Hi-index | 0.00 |
This paper introduces a novel approach to the detection of human body movements during daily life. With the sole use of one wearable wireless triaxial accelerometer attached to one's chest, this approach aims at classifying raw acceleration data robustly, to detect many common human behaviors without requiring any specific a-priori knowledge about movements. The proposed approach consists of feeding sensory data into a specifically trained Hierarchical Temporal Memory (HTM) to extract invariant spatial-temporal patterns that characterize different body movements. The HTM output is then classified using a Support Vector Machine (SVM) into different categories. The performance of this new HTM+SVM combination is compared with a single SVM using real-word data corresponding to movements like "standing", "walking", "jumping" and "falling", acquired from a group of different people. Experimental results show that the HTM+SVM approach can detect behaviors with very high accuracy and is more robust, with respect to noise, than a classifier based solely on SVMs.