WP1: Framework establishment and maintenance

This work package aims at the initial establishment and maintenance of a development framework for the remaining work packages. In this sense, it is focused on two tasks:

• T.1.1: Update, upgrade and maintenance of the System infrastructure: configuration and set-up of the basic necessary infrastructure, available from the previous projects from the Research Group, as well, as the acquisition and integration on additional infrastructure
• T.1.2: Update and maintenance of the Distributed Video Analysis (DivA) software architecture, developed within a previous project –SemanticVideo- from the Research Group, and especially focused on on-line operation and division of algorithms into stages.

WP2: Video object segmentation

This work package aims at coordinating research in video object segmentation, and to design and explore the influence of feedback strategies and alternative sensors on the algorithms here developed. In this sense, it is focused on four tasks:

• T.2.1: Region-based reflectance-invariant segmentation for static cameras.
• T.2.2: Global motion estimation and segmentation for moving cameras.
• T.2.3: Inclusion of intra-stage feedback policies and adaptation to inter-stage ones.
• T.2.4: Exploration of the adaptation of segmentation techniques to alternative sensors.

WP3: People modelling and detection

This work package aims at coordinating research in the modelling and detection of people, and to design and explore the influence of feedback strategies and alternative sensors on the algorithms here developed. In this sense, it is focused on four tasks:

• T.3.1: Combined people models adapted to people detection in dense environments.
• T.3.2: Analysis of people models oriented to detect considered people actions.
• T.3.3: Inclusion of intra-stage feedback policies and adaptation to inter-stage ones.
• T.3.4: Exploration of the adaptation of people detection techniques to alternative sensors.

WP4: Video object tracking

This work package aims at coordinating research in video object tracking, and to design and explore the influence of feedback strategies and alternative sensors on the algorithms here developed. In this sense, it is focused on four tasks:

• T.4.1: Tracking approaches for dense and cluttered environments.
• T.4.2: Visual attention techniques for partial occlusions and groupings.
• T.4.3: Inclusion of intra-stage feedback policies and adaptation to inter-stage ones.
• T.4.4: Exploration of the adaptation of tracking techniques to alternative sensors.

WP5: Integration, orchestration, testing and evaluation, and dissemination

This work package aims at integrating the algorithms developed within WP2, WP3 and WP4, in order to conform the global analysis chain to provide solutions to event detection for video surveillance and monitoring applications. Besides the more traditional sequential system integration of the analysis algorithms considered as independent, special attention will be paid to orchestrate the analysis chain via the use of feedback policies and the independent control of each analysis phase or stage to maximise results quality while minimizing overall resources. This work package also coordinates the testing and evaluation of the results from the previous work packages, and the project dissemination activities. Consequently, this work package consists of four tasks:

• T.5.1: Study, design, and integration of the analysis chain for event detection.
• T.5.2: Study, design and implementation of global feedback-driven approaches to orchestrate the operation of the analysis chain.
• T.5.3: Evaluation material and testing methodology; selection and generation, when required, of test sequences and associated ground-truth sequences, as well as the design of the evaluation approach best suited for each different algorithm.
• T.5.4: Dissemination via establishment of a web site for the compilation of results and their internal and external diffusion, with special attention to Observers (a.k.a EPOs), and the planning of publications.