Daniel Rowe

Understanding dynamic scenes based on human sequence evaluation

In this paper, a Cognitive Vision System (CVS) is presented, which explains the human behaviour of monitored scenes using natural-language texts. This cognitive analysis of human movements recorded in image sequences is here referred to as Human Sequence Evaluation (HSE) which defines a set of transformation modules involved in the automatic generation of semantic descriptions from pixel values. In essence, the trajectories of human agents are obtained to generate textual interpretations of their motion, and also to infer the conceptual relationships of each agent w.r.t. its environment. For this purpose, a human behaviour model based on Situation Graph Trees (SGTs) is considered, which permits both bottom-up (hypothesis generation) and top-down (hypothesis refinement) analysis of dynamic scenes. The resulting system prototype interprets different kinds of behaviour and reports textual descriptions in multiple languages.

Unconstrained multiple-people tracking

This work presents two main contributions to achieve robust multiple-target tracking in uncontrolled scenarios. A novel system which consists on a hierarchical architecture is proposed. Each level is devoted to one of the main tracking functionalities: target detection, low-level tracking, and high-level tasks such as target-appearance representation, or event management. Secondly, tracking performances are enhanced by on-line building and updating multiple appearance models. Successful experimental results are accomplished on sequences with significant illumination changes, grouping, splitting and occlusion events.

Improving Background Subtraction Based on a Casuistry of Colour-Motion Segmentation Problems

The basis for the high-level interpretation of observed patterns of human motion still relies on motion segmentation. Popular approaches based on background subtraction use colour information to model each pixel during a training period. Nevertheless, a deep analysis on colour segmentation problems demonstrates that colour segmentation is not enough to detect all foreground objects in the image, for instance when there is a lack of colour necessary to build the background model. In this paper, our segmentation procedure is based not only on colour, but also on intensity information. Consequently, the intensity model enhances segmentation when the use of colour is not feasible. Experimental results demonstrate the feasibility of our approach.

Improving tracking by handling occlusions

Keeping track of a target by successive detections may not be feasible, whereas it can be accomplished by using tracking techniques. Tracking can be addressed by means of particle filtering. We have developed a new algorithm which aims to deal with some particle-filter related problems while coping with expected difficulties. In this paper, we present a novel approach to handling complete occlusions. We focus also on the target-model update conditions, ensuring proper tracking. The proposal has been successfully tested in sequences involving multiple targets, whose dynamics are highly non-linear, moving over clutter.

Robust Multiple-People Tracking Using Colour-Based Particle Filters

Robust and accurate people tracking is a key task in many promising computer-vision applications. One must deal with non-rigid targets in open-world scenarios, whose shape and appearance evolve over time. Targets may interact, causing partial or complete occlusions. This paper improves tracking by means of particle filtering, where occlusions are handled considering the targets predicted trajectories. Model drift is tackled by careful updating, based on the history of likelihood measures. A colour-based likelihood, computed from histogram similarity, is used. Experiments are carried out using sequences from the CAVIAR database.

On tracking inside groups

This work develops a new architecture for multiple-target tracking in unconstrained dynamic scenes, which consists of a detection level which feeds a two-stage tracking system. A remarkable characteristic of the system is its ability to track several targets while they group and split, without using 3D information. Thus, special attention is given to the feature-selection and appearance-computation modules, and to those modules involved in tracking through groups. The system aims to work as a stand-alone application in complex and dynamic scenarios. No a-priori knowledge about either the scene or the targets, based on a previous training period, is used. Hence, the scenario is completely unknown beforehand. Successful tracking has been demonstrated in well-known databases of both indoor and outdoor scenarios. Accurate and robust localisations have been yielded during long-term target merging and occlusions.

On reasoning over tracking events

High-level understanding of motion events is a critical task in any system which aims to analyse dynamic human-populated scenes. However, current tracking techniques still do not address complex interaction events among multiple targets. In this paper, a principled event-management framework is proposed, and it is included in a hierarchical and modular tracking architecture. Multiple-target interaction events, and a proper scheme for tracker instantiation and removal according to scene events, are considered. Multiple-target group management allows the system to switch among different operation modes. Robust and accurate tracking results have been obtained in both indoor and outdoor scenarios, without considering a-priori knowledge about either the scene or the targets based on a previous training period.

Robust particle filtering for object tracking

This paper addresses the filtering problem when no assumption about linearity or gaussianity is made on the involved density functions. This approach, widely known as particle filtering, has been explored by several previous algorithms, including Condensation. Although it represented a new paradigm and promising results have been achieved, it has several unpleasant behaviours. We highlight these misbehaviours and propose an algorithm which deals with them. A test-bed, which allows proof-testing of new approaches, has been developed. The proposal has been successfully tested using both synthetic and real sequences.

A 3d dynamic model of human actions for probabilistic image tracking

In this paper we present a method suitable to be used for human tracking as a temporal prior in a particle filtering framework such as CONDENSATION [5]. This method is for predicting feasible human postures given a reduced set of previous postures and will drastically reduce the number of particles needed to track a generic high-articulated object. Given a sequence of preceding postures, this example-driven transition model probabilistically matches the most likely postures from a database of human actions. Each action of the database is defined within a PCA-like space called UaSpace suitable to perform the probabilistic match when searching for similar sequences. So different, but feasible postures of the database become the new predicted poses.

3D action modeling and reconstruction for 2d human body tracking

In this paper we present a technique for predicting the 2D human body joints and limbs position in monocular image sequences, and reconstructing its corresponding 3D postures using information provided by a 3D action model. This method is used in a framework based on particle filtering, for the automatic tracking and reconstruction of the 3D human body postures. A set of the reconstructed postures up to time t are projected on the action space defined in this work, which is learnt from Motion Capture data, and provides us a principled way to establish similarity between body postures, natural occlusion handling, invariance to viewpoint, robustness, and is able to handle different people and different speeds while performing an action. Results on manually selected joint positions on real image sequences are shown in order to prove the correctness of this approach.