Abstract This paper focuses on human behavior recognition where the main problem is to bridge the semantic gap between the analogue observations of the real world and the symbolic ...
Abstract. In speech recognition, phonemes have demonstrated their efficacy to model the words of a language. While they are well defined for languages, their extension to human act...
Kaustubh Kulkarni, Edmond Boyer, Radu Horaud, Amit...
The paper introduces an action recognition framework that uses concepts from the theory of chaotic systems to model and analyze nonlinear dynamics of human actions. Trajectories o...
Most of the methods for human motion tracking are based on the modeling of human dynamics in action execution. In even small example space of human activities, the variation in ac...
We propose a new method for human action recognition from video sequences using latent topic models. Video sequences are represented by a novel “bag-of-words” representation, w...