Gayathri Nadarajan

Automatic fish classification for underwater species behavior understanding

The aim of this work is to propose an automatic fish classification system that operates in the natural underwater environment to assist marine biologists in understanding subehavior. Fish classification is performed by combining two types of features: 1) Texture features extracted by using statistical moments of the gray-level histogram, spatial Gabor filtering and properties of the co-occurrence matrix and 2) Shape Features extracted by using the Curvature Scale Space transform and the histogram of Fourier descriptors of boundaries. An affine transformation is also applied to the acquired images to represent fish in 3D by multiple views for the feature extraction. The system was tested on a database containing 360 images of ten different species achieving as average correct rate of about 92%. Then, fish trajectories extracted using the proposed fish classification combined with a tracking system, are analyzed in order to understand anomalous behavior. In detail, the tracking layer computer fish trajectories, the classification layer associates trajectories to fish species and then by clustering these trajectories we are able to detect unusual fish behaviors to be further investigated by marine biologists.

Semantic-Based Workflow Composition for Video Processing in the Grid

We outline the problem of automatic video processing for the EcoGrid. This poses many challenges as there is a vast amount of raw data that need to be analysed effectively and efficiently. Furthermore, ecological data are subject to environmental changes and are exception-prone, hence their qualities vary. As manual processing by humans can be time and labour intensive, video and image processing tools can go some way to addressing such problems since they are computationally fast. However, most video analyses that utilise a combination of these tools are still done manually. We propose a semantic-based hybrid workflow composition method that strives to provide automation to speed up this process. The requirements for such a system are presented, whereby we aim for a solution that best satisfies these requirements and that overcomes the limitations of existing grid workflow composition systems

Translating a typical business process modelling language to a web services ontology through lightweight mapping

Bridging the gap between enterprise modelling methods and Semantic Web services is an important yet challenging task. For organisations with business goals, the automation of business processes as Web services is increasingly important, especially with many business transactions taking place within the Web today. Taking one approach to address this problem, a lightweight mapping between Fundamental Business Process Modelling Language (FBPML) and the Web Services Ontology (OWL-S) is outlined. The framework entails a data model translation and a process model translation via the use of ontologies and mapping principles. Several working examples of the process model translations are presented together with the implementation of an automated translator. FBPML constructs and process models that could not be translated to OWL-S equivalents highlight the differences between the languages of the two domains. It also implies that evolving Semantic Web technologies, in particular OWL-S, are not adequate for all service modelling needs and could thus benefit from the more traditional and mature BPM methods. On a more interesting note, this is effectively the first step towards enabling a semantic-based business workflow system.

SWAV: semantics-based workflows for automatic video analysis

This paper outlines the SWAV system - Semantics-based Workflows for Automatic Video Analysis. SWAV utilises ontologies and planning as core technologies to gear the composition and execution of video processing workflows. It is tailored for users without image processing expertise who have specific goals (tasks) and restrictions on these goals but not the ability to choose appropriate video processing software to solve their goals. An evaluation on a set of ecological videos has indicated that SWAV: 1) is more time-efficient at solving video classification tasks than manual processing; 2) is more adaptable in response to changes in user requests (task restrictions and video descriptions) than modifying existing image processing programs; and 3) assists the user in selecting optimal solutions by providing recommended descriptions.

A modular approach for automating video analysis

Automating the steps involved in video processing has yet to be tackled with much success by vision developers and knowledge engineers. This is due to the difficulty in formulating vision problems and their solutions in a generalised manner. In this collaborated work, we introduce a modular approach that utilises ontologies to capture the goals, domain description and capabilities for performing video analysis. This modularisation is tested on real-world videos from an ecological source and proves useful in conceptualising and generalising video processing tasks. On a more significant note, this could be used in a framework for automatic video analysis in emerging infrastructures such as the Grid.

An ontology-based conceptual mapping framework for translating FBPML to the Web services ontology

This paper presents an ontology-based conceptual mapping framework that translates a formal and visually rich business process modeling (BPM) language, Fundamental Business Process Modelling Language (FBPML) to a semantic Web-based language, the Web Services Ontology (OWL-S). The translation aims to narrow the gap between enterprise modelling methods and semantic Web services, thus bringing the two communities closer. Another significant contribution of the translation is that it allows more mature technologies such as BPM methods to be utilised within emerging fields that are constantly evolving, such as the semantic Web. The framework is divided into a data model translation and a process model translation. An implementation and an evaluation of the process model translation are demonstrated and discussed.

Goal, video description and capability ontologies for fish4knowledge domain

We created a set of domain ontologies that are based on user requirements for the Fish4Knowledge (F4K) project --- goal, video description and capability. The roles of the ontologies are to 1) support the development of appropriate functions of the projects workflow system, and 2) serve as a communication media to interface with other F4K components. The ontologies were designed with collaboration with image processing experts, marine biologists and user interface experts to capture the domain knowledge succinctly. They were utilised in the first version of our workflow composition and execution system for video classification, fish detection and counting tasks. They will continue to evolve with F4Ks needs and are envisaged to interface with other components.

Mapping fundamental business process modelling language to OWL-S

This paper presents a conceptual mapping framework between a formal and visual process modelling language, Fundamental Business Process Modelling Language (FBPML), and the Web Services Ontology (OWL-S), aiming to bridge the gap between Enterprise Modelling methods and Semantic Web services. The framework is divided into a data model and a process model component. An implementation and an evaluation of the process model mapping are demonstrated.

Intelligent Workflow Management for Fish4Knowledge Using the SWELL System

F4K’s workflow component is the chief mediator between the user interface (UI) and the video and image processing (VIP) components that reside in F4K’s high performance computing (HPC) platforms. Not only does it decompose high level user queries into lower video-level command-line invocations, it also makes its own decisions on which VIP modules and the best parameters to select, which hardware platform to perform the video processing executions on and which fault tolerance strategies to take during the executions so as to optimize the overall system’s performance. In this chapter, we describe the workings of F4K’s workflow component, SWELL (Semantic Workflows with Error Handling for Large Video Analyses) which comprises a workflow engine and a workflow monitor. We also describe the F4K domain ontologies that have heavily influenced the development of SWELL and have been used for term matching between partner components.

Real-Time Data Streaming Architecture and Intelligent Workflow Management for Processing Massive Ecological Videos

We present data collection and storage utilities and a workflow management system for handling the processing of large volumes of videos collected from an ecological source over several years and still growing. They lie in the heart of an integrated system that brings together expertise from various disciplines, including marine science, image processing, high performance computing and user interface. A real-time data streaming architecture was developed for efficient collection and storage of videos. In the analysis part, a workflow management system with two main components was deployed, i) a workflow engine and ii) a workflow monitor. The workflow engine deals with on-demand user queries and batch queries, selection of suitable computing platform and invocation of optimal software modules, while the workflow monitor handles the seamless execution and intelligent error handling of workflow jobs on a heterogeneous computing platform. We discuss the challenges that lie ahead for the workflow system such as the demand for more sophisticated scheduling and monitoring.