High Performance Moves Recognition
High Performance Moves Recognition and Sequence Segmentation Based on Key Poses Filtering
keywords
Image motion analysis; Action recgnition; Key poses filtering; LDCRF; Sequence segmentation; Motion segmentation.
Paper
C. M. de Souza Vicente, E. R. Nascimento, L. E. C. Emery, C. A. G. Flor, T. Vieira and L. B. Oliveira, "High performance moves recognition and sequence segmentation based on key poses filtering," 2016 IEEE Winter Conference on Applications of Computer Vision (WACV), Lake Placid, NY, 2016, pp. 1-8. doi: 10.1109/WACV.2016.7477711.
Methodology
1) Skeleton data is extracted by a RGB-D sensor; 2) These frames are filtered and a set of frames containing key poses are selected; 3) Training and testing with the acquired features using the discriminative LDCRF model.
Data Acquisition
Example of three moves in our dataset and their corresponding extracted key poses.
Moves Recognition
Confusion matrix of our methodology.
Average accuracy of our methodology, LDCRF+Allframes and Decision Forests.
Results
Two sequence segmentation charts for our methodology and Decision Forest.
BibTeX
@INPROCEEDINGS{
7477711,
author={C. M. de Souza Vicente and E. R. Nascimento and L. E. C. Emery and C. A. G. Flor and T. Vieira and L. B. Oliveira},
booktitle={2016 IEEE Winter Conference on Applications of Computer Vision (WACV)},
title={High performance moves recognition and sequence segmentation based on key poses filtering},
year={2016},
pages={1-8},
doi={10.1109/WACV.2016.7477711},
month={March},
}