Searching for Complex Human Activities with No Visual Examples

Abstract

We describe a method of representing human activities that allows a collection of motions to be queried without examples, using a simple and effective query language. Our approach is based on units of activity at segments of the body, that can be composed across space and across the body to produce complex queries. The presence of search units is inferred automatically by tracking the body, lifting the tracks to 3D and comparing to models trained using motion capture data. Our models of short time scale limb behaviour are built using labelled motion capture set. We show results for a large range of queries applied to a collection of complex motion and activity. We compare with discriminative methods applied to tracker data; our method offers significantly improved performance. We show experimental evidence that our method is robust to view direction and is unaffected by some important changes of clothing.

Object, Scene and Actions: Combining Multiple Features for Human Action Recognition

“…This makes the comprehensive listing of the related literature impossible, while Forsyth et al [10] presents an extensive review of the subject. Some of the recent works include [8,29,18,20,16,13]. In most of the earlier works, the focus is on simpler scenarios, where the background was stable and the foreground human figure is easy to extract [4,18].…”

“…However, this scenario is hardly realistic; videos from the real world are fairly complicated, especially when taken in uncontrolled environments. Some recent approaches try to deal with such complex scenarios [19,25,20,16].…”

Histograms of oriented optical flow and Binet-Cauchy kernels on nonlinear dynamical systems for the recognition of human actions

“…[17,10,30] to represent human activity in a video. Niebles [20] presented an unsupervised method similar to the bagof-words approach for learning the probability distributions of space-time interest points in human action videos.İkizler et al [16] presented a method whereby limb motion model units are learnt from labeled motion capture data and used to detect more complex unseen motions in a test video using search queries specific to the limb motion in the desired activity. However, an important limitation of the aforementioned approaches is that they do not incorporate global characteristics of the activity as a whole.…”

On Supervised Human Activity Analysis for Structured Environments

“…Parameswaran et al [1] detects a number of body joints and analyses their trajectories in 2D invariance space. Detecting and tracking body parts have also been used to infer the higher level activities [2,3]. In this, state space methods have been employed to analyse a sequence of lower level events.…”