.png){kind=icon}
Return
A two-step shapelets based framework for interactional activities recognition
.png){kind=icon}
PRE
AI
DOI：10.1007/s11042-022-11987-0.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Human-human interactions recognition has high potential to have a big impact on enabling robots being able to interact with people. Recently, body sensor networks (BSNs) have been widely applied in many fields. In this paper, we investigate the problem of recognizing human interactional activities based on BSNs. Considering that individual actions may contribute differently to the final recognition of interactions, a novel two-step framework that recognizes interactional activities by fusing labels of individual actions is presented. Specifically, shapelets based method is adopted for recognizing individual actions in the first step, and recognition of interactions is realized by weighted fusion of recognized individual actions in the second step. The framework is tested on our newly collected dataset. We mainly compare the performance of the proposed framework with traditional recognition algorithms based framework. Furthermore, feature level fusion is also conducted to verify the effectiveness of the proposed framework. Experimental results show that the proposed framework has achieved promising results with an overall recognition accuracy of 99.44%.
Keywords:
Two-person interaction
Wearable sensors
Shapelets
Time series
Weighted fusion
Journal
IF:
3
Papers:
2.0W
Citations:
3.2W
Organization
Cited Papers
Regional variation of care dependency after hip fracture in Germany: A retrospective cohort study using health insurance claims data.png)
PLOS ONE
OAAI
PLOS ONEIF0
OAAIShareSaveShareSaveHuman Activity Classification in Smartphones Using Accelerometer and Gyroscope SensorsIEEE SENSORS JOURNAL
IEEE SENSORS JOURNALIF4.5
PREAI
ShareSaveShareSave
View more