Tracking a walking person using activity-guided annealed particle filtering

Abstract

Tracking human pose using observations from less than three cameras is a challenging task due to ambiguity in the available image evidence. This work presents a method for tracking using a pre-trained model of activity to guide sampling within an Annealed Particle Filtering framework. The approach is an example of model-based analysis-bysynthesis and is capable of robust tracking from less than 3 cameras with reduced numbers of samples. We test the scheme on a common dataset containing ground truth motion capture data and compare against quantitative results for standard Annealed Particle Filtering. We find lower absolute and relative error scores for both monocular and 2-camera sequences using 80% fewer particles.

A Determination Method for Gait Event Based on Acceleration Sensors

“…Some studies recognize and evaluate the daily activities through Microsoft Kinect sensors which measured concurrently with a 3D Mo-Cap system (gold standard), with the result showing that the Microsoft Kinect sensors can effectively identify the characteristics of patients' daily activities and, hence, that it can be used as an effective clinical evaluation method [31,32]. Wearable devices are also widely used to track human walking activities [33]. However, wearable technologies have inherent limitations, and their sensor responses are often influenced by wearers' behavior, motion, clothing and environmental factors [34].…”

Behaviour based particle filtering for human articulated motion tracking

“…[3,7] may be relaxed to one of a set of known activities. PCA is used to create multiple activity spaces from training data and HMMs are trained to guide their exploration.…”

“…To avoid searching in high dimensional feature spaces, approaches to tracking often make assumptions about the class of movement and look for solutions in low dimensional pose spaces recovered from training data [3,7]. Inspired by earlier work [5], we extend a previous approach using a single activity model [3] to give simultaneous consideration to multiple models.…”

“…Inspired by earlier work [5], we extend a previous approach using a single activity model [3] to give simultaneous consideration to multiple models. Low dimensional activity feature spaces are produced by the application of PCA to training data.…”