Anastasios Dimou

An Empirical Study of Multi-label Learning Methods for Video Annotation

This paper presents an experimental comparison of different approaches to learning from multi-labeled video data. We compare state-of-the-art multi-label learning methods on the Media mill Challenge dataset. We employ MPEG-7 and SIFT-based global image descriptors independently and in conjunction using variations of the stacking approach for their fusion. We evaluate the results comparing the different classifiers using both MPEG-7 and SIFT-based descriptors and their fusion. A variety of multi-label evaluation measures is used to explore advantages and disadvantages of the examined classifiers. Results give rise to interesting conclusions.

Local Invariant Feature Tracks for high-level video feature extraction

This paper builds upon previous work on local interest point detection and description to propose the extraction and representation of novel Local Invariant Feature Tracks (LIFT). These features compactly capture not only the spatial attributes of 2D local regions, as in SIFT and related techniques, but also their long-term trajectories in time. This and other desirable properties of LIFT allow the generation of Bags-of-Spatiotemporal-Words models that facilitate capturing the dynamics of video content, which is necessary for detecting high-level video features that by definition have a strong temporal dimension. Preliminary experimental evaluation and comparison of the proposed approach reveals promising results.

VERGE: A video interactive retrieval engine

This paper presents the video retrieval engine VERGE, which combines indexing, analysis and retrieval techniques in various modalities (i.e. textual, visual and concept search). The functionalities of the search engine are demonstrated through the supported user interaction modes.

High-level event detection system based on discriminant visual concepts

This paper demonstrates a new approach to detecting high-level events that may be depicted in images or video frames. Given a non-annotated content item, a large number of previously trained visual concept detectors are applied to it and their responses are used for representing the content item with a model vector in a high-dimensional concept space. Subsequently, an improved subclass discriminant analysis method is used for identifying a concept subspace within the aforementioned concept space, that is most appropriate for detecting and recognizing the target high-level events. In this subspace, the nearest neighbor rule is used for comparing the non-annotated content item with a few known example instances of the target events. The high-level events used as target events in the present version of the system are those defined for the TRECVID 2010 Multimedia Event Detection (MED) task.

On the use of feature tracks for dynamic concept detection in video

This paper proposes the use of feature tracks for the detection of concepts in video, particularly dynamic concepts. Feature tracks are defined as sets of local interest points found in different frames of a video shot that exhibit spatio-temporal and visual continuity, defining a trajectory in the 2D+Time space. The extraction of feature tracks and the selection and representation of an appropriate subset of them allow the generation of a Bag-of-Spatiotemporal-Words model for the shot, which facilitates capturing the dynamics of video content. The experimental evaluation of the proposed approach highlights how the selection of such feature tracks for the definition of the Bag-of-Spatiotemporal-Words model enhances the results of traditional keyframe-based concept detection techniques.