Action Completion: A Temporal Model for Moment Detection

Position (finance)
DOI: 10.48550/arxiv.1805.06749 Publication Date: 2018-01-01
ABSTRACT
We introduce completion moment detection for actions - the problem of locating completion, when action's goal is confidently considered achieved. The paper proposes a joint classification-regression recurrent model that predicts from given frame, and then integrates frame-level contributions to detect sequence-level moment. voting node frame's relative position by either classification or regression. method also capable detecting incompletion. For example, missed ball-catch, as well at which ball safely caught. test on 16 three public datasets, covering sports daily actions. Results show combining frames prior post detected within one second in 89% all tested sequences.
SUPPLEMENTAL MATERIAL
Coming soon ....
REFERENCES ()
CITATIONS ()
EXTERNAL LINKS
PlumX Metrics
RECOMMENDATIONS
FAIR ASSESSMENT
Coming soon ....
JUPYTER LAB
Coming soon ....