Object Trajectory-Based Activity Classification and Recognition Using Hidden Markov Models

Abstract

Motion trajectories provide rich spatiotemporal information about an object's activity. This paper presents novel classification algorithms for recognizing object activity using object motion trajectory. In the proposed classification system, trajectories are segmented at points of change in curvature, and the subtrajectories are represented by their principal component analysis (PCA) coefficients. We first present a framework to robustly estimate the multivariate probability density function based on PCA coefficients of the subtrajectories using Gaussian mixture models (GMMs). We show that GMM-based modeling alone cannot capture the temporal relations and ordering between underlying entities. To address this issue, we use hidden Markov models (HMMs) with a data-driven design in terms of number of states and topology (e.g., left-right versus ergodic). Experiments using a database of over 5700 complex trajectories (obtained from UCI-KDD data archives and Columbia University Multimedia Group) subdivided into 85 different classes demonstrate the superiority of our proposed HMM-based scheme using PCA coefficients of subtrajectories in comparison with other techniques in the literature.

Action recognition using global spatio-temporal features derived from sparse representations

“…Over the past decade, the problem of action recognition in videos has been looked at from multiple perspectives, with a diverse set of solutions suggested. Bashir et al [8] demonstrate the use of HMMs with spatio-temporal curvature representations and PCA decompositions for trajectory based human action classification. Another powerful representation method involves describing actions as a sequence of shapes [9], and has gained popularity due to its invariance properties.…”

A Survey on Video Transmission Techniques

“…In the particle filter, many features can be used. Color and edge contour [26] are two common features that express targets.…”

Warped K-Means: An algorithm to cluster sequentially-distributed data

“…Approaches in [48,60,61] Supervised [9,47,59] of full sequences [10,27,38] Figure 1: Taxonomy of related work. For the sake of brevity, we cite in this figure at most three of the references that relate to our work.…”

“…In the "supervised" case, different kinds of models such as Hidden Markov Models [9,52,57] or Neural Networks [29,35,59] are trained for different segments using a suitable amount of manually segmented and labeled trajectories. This clearly differs from our proposal, since we are only interested in unsupervised methods, which do not require the availability of (otherwise costly) training data.…”