Stella J. George

The Recognition and Analysis of Animate Objects using Neural Networks and Active Contour Models

Abstract In this paper we describe a method for tracking walking humans in the visual field. Active contour models are used to track moving objects in a sequence of images. The resulting contours are then encoded in a scale-, location-, resolution- and control point rotation-invariant vector. These vectors are used to train and test feedforward error-backpropagation neural networks, which are able to distinguish both static and dynamic human objects from other classes of object, including horses, dogs and inanimate objects. Experimental results are presented which show the neural networks ability to successfully categorise objects which have become partially occluded. Classes of object can be distinguished by the network, and experimental results are presented which show how the representational vectors used as input patterns can be used to identify, classify and analyse the temporal behaviour of pedestrians.

Analysis of Human Motion Using Snakes and Neural Networks

A novel technique is described for analysing human movement in outdoor scenes. Following initial detection of the humans using active contour models, the contours are then re-represented as normalised axis crossover vectors. These vectors are then fed into a neural network which determines the typicality of a given human shape, allowing for a given human’s motion deformation to be analysed. Experiments are described which investigate the success of the technique being presented.

The Architecture and Performance of a Stochastic CompetitiveEvolutionary Neural Tree Network

A new dynamic tree structured network—the Stochastic Competitive Evolutionary Neural Tree (SCENT) is introduced. The network is able to provide a hierarchical classification of unlabelled data sets. The main advantage that SCENT offers over other hierarchical competitive networks is its ability to self-determine the number and structure of the competitive nodes in the network without the need for externally set parameters. The network produces stable classificatory structures by halting its growth using locally calculated, stochastically controlled, heuristics. The performance of the network is analysed by comparing its results with that of a good non-hierarchical clusterer, and with three other hierarchical clusterers and its non stochastic predecessor. SCENTs classificatory capabilities are demonstrated by its ability to produce a representative hierarchical structure to classify a broad range of data sets.

Towards computational neural systems through developmental evolution

The capability of creating artificial neural networks with biologically-plausible characteristics, is becoming ever more attainable through the greater understanding of biological neural systems and the constant increases in available desktop computing power. Designing and implementing such neural systems still however remains a complex and challenging problem. This chapter introduces a design methodology, inspired by embryonic neural development, which is capable of creating 3 dimensional morphologies of both individual neurons and networks. Examples of such morphologies are given, which are created by evolving the parameters of the developmental model using an evolutionary algorithm.

Developmental Evolution of an Edge Detecting Retina

The task addressed in this paper is the evolution of an artificial retina with an on-centre/off-surround response which performs edge detection. Evolutionary optimisation is performed on the parameters of a developmental model. The model is capable of creating three dimensional, multilayer neural networks by modelling the outgrowth of neuron-to-neuron connectivity. A genetic algorithm is used to optimise the developmental parameters, measured against a target retina structure. The first stage of evolution adapts the parameters of outgrowth rules and the developmental environment. We show that this type of development can be sensitive to noisy conditions (perturbations in neuron positions). This limitation can be overcome by incorporating overgrowth and pruning. Staged evolution of these processes is shown to result in robust development.

Designing Development Rules for Artificial Evolution

Using artificial evolution to successfully create neural networks requires appropriate developmental algorithms. The aim is to determine the least complex set of rules that allow a range of networks to evolve. This paper presents a set of generic growth rules that abstractly model the biological processes associated with the development of neuron-to-neuron connections. Substantially different 3D artificial neural structures can be grown by changing parameter values associated with the rules. A genetic algorithm has been successfully employed in determining parameter values that lead to specific neural structures.

A Hybrid Detection and Classification System for Human Motion Analysis

This paper discusses a hybrid technique for detecting and tracking moving pedestrians in a video sequence. The technique comprises two sub-systems: an active contour model for detecting and tracking moving objects in the visual field, and an MLP neural network for classifying the moving objects being tracked as ‘human’ or ‘nonhuman’. The axis crossover vector method is used for translating the active contour into a scale-. location-, resolution- and rotation-invariant vector suited for input to a neural network, and we identify the most appropriate level of detail for encoding human shape information. Experiments measuring the neural network’s accuracy at classifying unseen computer generated and real moving objects are presented, along with potential applications of the technology. Previous work has accommodated lateral pedestrian movement across the visual field; this paper describes a system which accommodates arbitrary angles of pedestrian movement on the ground plane.

Detecting, Tracking, and Classifying Human Movement Using Active Contour Models and Neural Networks

Detecting and tracking moving objects in the visual field is a task which has interested the computer vision discipline for some years [4, 7, 12, 13, 14]. A hybrid technique is described in this chapter for detecting and tracking moving objects in a sequence of images, and for identifying them as ‘human’ or ‘non-human’.