"Recognition" : Interactive motion tracking installation with neural network agent performer

Public installation running throughout the night at Cube 37 gallery. Morphing forms are projected onto the glass front of the gallery, facing the street. Human perticipants interact with the forms whose morphology changes with the movements of the participants. When no human participants are present, a neural network based agent that has learnt how to dance from a trained dancer, takes over and the forms follow the agent's movement.

Snapshots of complexity: using motion capture and principal component analysis to reconceptualise dance

This article brings together the disparate worlds of dance practice, motion capture and statistical analysis. Digital technologies such as motion capture offer dance artists new processes for recording and studying dance movement. Statistical analysis of these data can reveal hidden patterns in movement in ways that are semantically ‘blind’, and are hence able to challenge accepted culturo-physical ‘grammars’ of dance creation. The potential benefit to dance artists is to open up new ways of understanding choreographic movement. However, quantitative analysis does not allow for the uncertainty inherent in emergent, artistic practices such as dance. This article uses motion capture and principal component analysis (PCA), a common statistical technique in human movement recognition studies, to examine contemporary dance movement, and explores how this analysis might be interpreted in an artistic context to generate a new way of looking at the nature and role of movement patterning in dance creation.

Face recognition from face motion manifolds using robust kernel resistor-average distance

In this work we consider face recognition from face motion manifolds. An information-theoretic approach with Resistor-Average Distance (RAD) as a dissimilarity measure between distributions of face images is proposed. We introduce a kernel-based algorithm that retains the simplicity of the closed-form expression for the RAD between two normal distributions, while allowing for modelling of complex, nonlinear manifolds. Additionally, it is shown how errors in the face registration process can be modelled to significantly improve recognition. Recognition performance of our method is experimentally demonstrated and shown to outperform state-of-the-art algorithms. Recognition rates of 97–100% are consistently achieved on databases of 35– 90 people.

Contextually Learnt Detection of Unusual Motion-Based Behaviour in Crowded Public Spaces

In this paper we are interested in analyzing behaviour in crowded publicplaces at the level of holistic motion. Our aim is to learn, without user input, strong scene priors or labelled data, the scope of ‘‘normal behaviour’’ for a particular scene and thus alert to novelty in unseen footage. The first contribution is a low-level motion model based on what we term tracklet primitives, which are scenespecific elementary motions. We propose a clustering-based algorithm for tracklet estimation from local approximations to tracks of appearance features. This is followed by two methods for motion novelty inference from tracklet primitives: (a) an approach based on a non-hierarchial ensemble of Markov chains as a means of capturing behavioural characteristics at different scales, and (b) a more flexible alternative which exhibits a higher generalizing power by accounting for constraints introduced by intentionality and goal-oriented planning of human motion in a particular scene. Evaluated on a 2 h long video of a busy city marketplace, both algorithms are shown to be successful at inferring unusual behaviour, the latter model achieving better performance for novelties at a larger spatial scale.

An information-theoretic approach to face recognition from face motion manifolds

In this work, we consider face recognition from face motion manifolds (FMMs). The use of the resistor-average distance (RAD) as a dissimilarity measure between densities confined to FMMs is motivated in the proposed information-theoretic approach to modelling face appearance. We introduce a kernel-based algorithm that makes use of the simplicity of the closed-form expression for RAD between two Gaussian densities, while allowing for modelling of complex and nonlinear, but intrinsically low-dimensional manifolds. Additionally, it is shown how geodesically local FMM structure can be modelled, naturally leading to a stochastic algorithm for generalizing to unseen modes of data variation. Recognition performance of our method is demonstrated experimentally and is shown to exceed that of state-of-the-art algorithms. Recognition rate of 98% was achieved on a database of 100 people under varying illumination

Dancing in suits: a performer's perspective on the collaborative exchange between self, body, motion capture, animation and audience

The motion capture process places unique demands on performers. The impact of this process on the simultaneously artistic/somatic nature of dance practice is profound. This paper explores, from a performer’s perspective, how the process of performing in an optical motion capture system can impact and limit, but also expand and reconfigure a dancer’s somatic practice. This paper argues that working within motion capture processes affects not only the immediate contexts of capture and interactive performance, but also sets up a dialogue between dance practices within and beyond the motion capture studio.

Resting toucher: a time and motion analysis of elite lawn bowls

Whilst numerous investigations have explored the physical demands placed upon competitive sportspeople from a wide array of sports little is known about the physical demands placed on lawn bowlers. The purpose of this study was to ascertain the movement activities of Australian representative singles and pairs players and to determine the frequency and duration of these activities. One match each of two male and two female players (one singles and one pairs player per gender) were videotaped during an international tournament. During playback of the videotaped matches (n = 4), a single observer coded the players’ activities into five distinct categories (waiting, walking forward, walking backward, jogging and bowling) using a computerised video editing system (Gamebreaker™ Digital Video Analysis System). Field calibration of players over 30m for forward motions and 15m for the backward motion was performed to allow for the estimation of total distance covered during the match. Heart rate was monitored during each match. The duration of a match was found to be (mean ± SD) 1hr 28 ± 15mins. The total distance covered during each match was 2093 ± 276m. The mean percentage of match time spent in each motion was: waiting, 61.8 ± 9.3%; walking forward, 22.3 ± 5.6%; walking backward, 2.0 ± 0.4%; jogging, 1.1 ± 0.5%; and bowling, 8.5 ± 4.2%. Average heart rate was found to be 57 ± 7% of age-predicted HRmax with a maximum of 78 ± 9% of age-predicted HRmax. The results of this study suggest that playing lawn bowls at an international level requires light-moderate intensity activity similar to that reported for golf.