Abstract | ||
---|---|---|
•A novel depth maps and skeleton-based action descriptor is proposed.•Local motion patterns between consecutive depth maps are captured at joint positions.•Feature selection is based on inter-class correlation and redundancy reduction.•Late fusion takes advantage of multiple feature sets based on different modalities.•Extensive experiments confirmed the superiority of the proposed framework. |
Year | DOI | Venue |
---|---|---|
2019 | 10.1016/j.jvcir.2019.03.026 | Journal of Visual Communication and Image Representation |
Keywords | Field | DocType |
Action recognition,Action descriptors,Depth maps,Feature selection,Multimodal representation,Decision-level fusion | Inertial frame of reference,Computer vision,Units of measurement,Pattern recognition,Feature selection,Support vector machine,Redundancy (engineering),Multilayer perceptron,Artificial intelligence,RGB color model,Discriminative model,Mathematics | Journal |
Volume | ISSN | Citations |
61 | 1047-3203 | 0 |
PageRank | References | Authors |
0.34 | 0 | 2 |
Name | Order | Citations | PageRank |
---|---|---|---|
Filip Malawski | 1 | 6 | 3.19 |
Bogdan Kwolek | 2 | 328 | 40.16 |