Collaboration Initiatives

Name: Kliegr Tomas Research Activity: Exploiting free-text annotations accompanying images as a complementary resource in image classification. Online demo at http://nb.vse.cz/~klit01/hypernym_discovery Collaborations: Currently at PhD exchange at MMV group - QMU. Name: Vallet Félicien Research Activity: Segmentation and structuration of a Talk-Show using audio, vidéo and text modalities. My background is signal processing oriented but right now I'd like to work on the automatic extraction of semantic knowledge from texts to incorporate high-level interpretations. Collaborations: Maybe a collaboration with DFKI (action point for next meeting I think). Nothing really done so far. Name: HAN Yina Research Activity: My research interest includes pattern recognition and machine learning, image and vision processing, face recognition, and biometrics. Collaborations: Please provide a brief note on the nature of joint research activities. Name: Achanta Radhakrishna Research Activity: My area of research is Image Content Analysis aided by Image Metadata like Exif, for use in Automatic Image Annotation. I could use help in Machine Learning techniques, Object Recognition and techniques for using metadata for image content analysis. Collaborations: Please provide a brief note on the nature of joint research activities. Name: Hughes Mark Research Activity: Computer Vision, Social Networking, Gis Collaborations: Name: Bailer Werner Research Activity: analysis and summarisation of unedited scripted/unscripted video content such as rushes in digital cinema post-production (including multi-view content for 3D cinema) and surveillance; metadata models for audiovisual content links to other activities: evaluation of browsing/skimming approaches, user evaluation of browsing/summarisation/skimming tools Collaborations: - TRECVID rushes activities - metadata for TRECVID search activities - metadata validation (VAMP service) together with CWI - metadata for audio analysis (with GET) Name: Saathoff Carsten Research Activity: I am mainly concerned with efficient annotation. The paper is specifically about region calssification using only few training examples and combining low-level processing with high-level features, in this case spatial features. This is clearly realted to image classification, knowledge-assisted analysis and semantics. Besides that I am also interested in other kinds of semi-automatic annotation, like clustering for event detection, interactive approaches, using backgroundknowledge etc. Collaborations: I am leading the KAT development. Besides that I only collaborate with other kspace participants within KU. Name: Emiya Valentin Research Activity: Automatic transcription of music consists in analyzing a recording in order to extract and recognize musical events (e.g. notes, rhythm, tempo, etc). My PhD research project focuses on automatic transcription of piano music, with particular interests in multipitch estimation and in specific piano aspects (inharmonicity, etc.). Links: speech and audio signal processing, Music Information Retrieval. Collaborations: Some common interests in multipitch estimation and piano sound analysis with Eurecom. Name: Seneviratne Lasantha Research Activity: My research project involves the study of minimize the semantic gap in computer vision using image annotation (tagging) through gaming. I consider the human psychological aspects to create motivation through developing an attractive gaming environment. The main goal of the research is to verify the tagging process accuracy, using low-level features. Research interests: Image classification, Object detection, Image annotation Collaborations: Please provide a brief note on the nature of joint research activities. Name: Nemrava Jan Research Activity: We use text mining approaches (ontology-based information extraction) for names and entities extraction from structured and unstructured data to use these information for video annotation, retrieval and event detection. Collaborations: In the field of video analysis we cooperate with DCU and TUB and in the field of information extraction from texts we have nice collaboration with DFKI. Name: SCHUTZ Antony Research Activity: My PhD research focuses on multipitch estimation for automatic transcription and on interpretation effect detection. Collaborations: This work is the result of an internal collaboration between the multimedia and mobile Department of Eurecom. Name: Piatrik Tomas Research Activity: My research interests are in the fields of semantic multimedia retrieval, biologically inspired methods and multimodal feature mapping. My current work includes knowledge based image classification and clustering based on Ant Colony Optimization. Additional proposal for video summarization is based on Ant-Tree Clustering method. Collaborations: I contributed to K-Space retrieval system and my classification module was used for the feature extraction task of TRECVid 2006/2007/2008. Video summarization module was used in collaborative activity on TRECVid 2008 summarization task. Name: Chen Yi Research Activity: My focus is on exploration of efficient retrieval of personal experiences and knowledge from human digital memory archives containing heterogeneous content in various media. The emphasis in my research is on exploring how to make use of human memory and context information to faciltate various modes of searching and exploration, and particular in information searching. This requires content analysis of stored items and modeling of contextual relationships between them. Collaborations: Please provide a brief note on the nature of joint research activities. Name: Hildebrand Michiel Research Activity: We investigate to what extent explicit semantics can be used to support end users with the exploration of a large heterogeneous collection. In particular we consider cultural heritage, a knowledge-rich domain in which collections are typically described by multiple thesauri. We focus on three types of end user functionality: searching for terms within multiple thesauri, semantic keyword search and faceted browsing. Collaborations: Please provide a brief note on the nature of joint research activities. Name: Amin Alia Research Activity: I am interested in investigating user information seeking needs in different domains. Specifically, I am interested in information seeking needs elicitation, including interfaces/technologies that might help users search and browse information. Collaborations: Please provide a brief note on the nature of joint research activities. Name: Kaiser Florian Research Activity: Research field : audio summarization including speaker analysis, music analysis, source separation, machine learning. Collaborations: Please provide a brief note on the nature of joint research activities. Name: Philippeau Jérémy Research Activity: With the aim of audiovisual database consulting, without being limited to a predefined applicative context, the prospect of a user-dependent interactive visual organization should be enviable: with S a same small subset of documents, a user should have the possibility to explore several graphical arrangements of it; the rest of the corpus (or a part of it) has to reorganize itself regarding the notion of similarity expressed by the human on S. Collaborations: Name: Athanasiadis Thanos Research Activity: Briefly state the nature of your research and with possible links to other research activities Collaborations: Both EURECOM and ITI-IVML research groups have strong interest in multimedia content representation and its semantic interpretation. The objective of this exchange was to examine how semantic interpretation of multimedia content can be improved using existing experience and tools of the two groups. The outcome was to extend semantic labeling of images to video shot sequences and achieve efficient and semantic-aware spatiotemporal video segmentation. Name: Simou Nikolaos Research Activity: My research interests include knowledge representation both under certain and uncertain or imprecise knowledge based on crisp and fuzzy descriptions logics, ontologies, implementation and optimization of reasoning systems and applications for the Semantic Web. Collaborations: I have participated in a collaboration with UEP and DFKI for building an architecture for mining resources complementary to audio-visual streams. Name: De Simone Francesca Research Activity: My research focuses on the objective assessment of the quality of multimedia content, i.e. still pictures (= visual quality assessment) and video sequences (= visual quality, audio quality, and joint audio-visual quality assessment). Thus, connections with the research in the field of multimedia content analysis are very easy. In particular I am currently working on the analysis of color information thus it would be worth to know more about approaches for the classification and analysis of color data, which I suppose being relevant in the field of multimedia content analysis. Collaborations: Please provide a brief note on the nature of joint research activities. Name: Mailis Theofilos Research Activity: My research interests are Knowledge Representation and Reasoning both under perfect and imperfect knowledge. In particular I work in the area of Semantic Web, ontologies and Description Logics. Collaborations: Name: Uscumlic Marija Research Activity: My field of research is multimedia content annotation. Accent is on the fusion of the results obtained by content analysis with textual tags. When users in social networks annotate and rate multimedia content, they provide objective cognitive information and subjective opinions at a level that cannot typically be handled by multimedia content analysis. Research is inspired by this phenomenon considering obvious limitation of automatic technique for content analysis. P.S. as I started PhD in 2008, paper will be more "state of the art" oriented Collaborations: Please provide a brief note on the nature of joint research activities. Name: Leelanupab Teerapong Research Activity: Multimedia Information Retrieval dealing with designing and developing interfaces for Personal Digital Information Management. This deals with content extraction, and multimedia annotation as well as classification, searching and browsing techniques. I'm currently finalising and planning about evaluation on a multi-faceted ostensive browser system with story creation function. This is a part of my warmed-up project and initial research. Collaborations: Present I'm a visiting student at DCU as a part of joint activities under K-space cooperation. Name: durrieu jean-louis Research Activity: Signal processing applied to music transcription: use of NMF (non negative matrix factorisation) in order to extract the main melody from polyphonic music audio. Also involves (audio) source separation. The NMF technique can be used for other domains such as image and video processing. Collaborations: Jan Weil from TUB spent two months (May and June 2008) in our lab, at Telecom ParisTech. There, we collaborated on the topic I will be presenting during the jamboree. With his chord detection and our melody extraction, we desire to produce so-called "lead-sheets" from polyphonic music. The obtained results can also be used in order to compare two songs in applications such as cover song detection or song similarity (automatic generation of playlists)... Name: Papadopoulos Georgios Research Activity: The main aim of my research activity is the detection and recognition of high-level semantic concepts in still images (using multiple sources of information such as color, texture, shape) and semantic events in video sequences (using multiple modalities such as motion, color, audio). Most of the developed algorithms are based on Machine Learning methods for implicit knowledge acquisition. Additionally, I have studied the issue of incorporating contextual information (spatial, temporal) in the semantic multimedia analysis procedure, and examined individual approaches for realizing information fusion. Possible links to my research constitute reasoning approaches (for further improving the analysis results) and automatic speech recognition techniques (for detecting more complex semantic concepts in video sequences). Collaborations: Two joint publications have been made so far: a) a paper that presents and comparatively evaluates 4 individual approaches to semantic image analysis (ITI, QMUL), b) a publication that outlines an integrated framework developed within WP4 (Eurecom, ITI, QMUL)