Audio-visual content search and retrieval in a distributed P2P repository


- Develop a MultiPedia content distributed repository.
- Develop an innovative content/ context-based 3D search engine based on extraction of low-level features and intuitive semantics.
- Develop a P2P distributed environment supporting PCs and mobile devices.
- Provide the tools for retrieving imminent valuable results identifying the closest and more relevant neighbours.
- Improve information access and content navigation through relevance feedback so as to significantly increase the systems ability to adjust to specific user preferences and to improve system's performance.
- Integrate novel personalized multimodal user interfaces with knowledge extraction algorithms in order to provide next generation information query and access mechanisms.
- Develop novel visualization and rendering techniques into low-processing mobile devices.
- Ensure MultiPedia content's ownership by using/developing ΟΜA and 3D object watermarking methods.



- Novel algorithms for low-level feature extraction able to be applied to 3D objects regardless of their degeneracies, formats and levels of detail.
- Novel statistic rule based algorithms for automated knowledge extraction from the low-level features taking into account the real discriminative power of each feature will be developed.
- Text, 2D/3D, 3D/3D and sketches to 3D matching algorithms.
- Enhanced relevance feedback methods adapted to 3D objects and modelling very high-level concepts matching the user perception subjectivity by analysing his intuitive operations.
- Active hidden annotation process integrating user behaviour and interests . The user’s natural interactions with the user interface will be exploited for propagating knowledge as well.
- New matching methods combining low and high-level features producing a robust descriptor able to efficiently discriminate the different 3D objects.
- Annotation propagation techniques for knowledge propagation to existing peer 3D object databases.
- Semantic technology based retrieval and access to content.


- CERTH /Informatics and Telematics Institute (GR)
- University of Ljubljana (SL)
- Politecnico Di Torino (IT)
- TWT GmbH (DE)

  Website coming soon...


In light of three dimensional (3D) object retrieval evolving from text annotation to content- based and from standalone applications to web-based search engines, the creation of a search engine for 3D and multimedia distributed content into Peer-to-Peer (Ρ2Ρ) and mobile Ρ2Ρ (m-Ρ2Ρ) networks, is proposed. Driven by the very successful concept of Wikipedia (, the first goal of the proposed project is to create the first distributed Visual Object – MultiPedia - Repository in which any peer can contribute to. The main differences between VICTORY and any known knowledge repository, is that, firstly, any visual information will be described as MultiPedia object (3D object along with its accompanied information - 2D views, text, audio, video), and secondly, the distributed nature of the repository. VICTORY’s main objective is to create a MultiPedia search engine based on the 3D objects. The user will be able to use as input any Multipedia object or a combination of them. The retrieved content will contain 3D objects and their accompanied mixed-media, such as 3D objects, images, text files, video and audio and it will be accessible from both mobile devices (cell-phones, PDAs, etc.) and standard PCs. By doing so, the heavy tasks of feature extraction, indexing and storage of 3D data is distributed into peers, and the same time another aspect to searching, in addition to the filename-based method prevalent in Ρ2Ρ applications, is introduced.



- Multimedia search and retrieval algorithms
- 3D low-level feature extraction algorithms
- Annotation mechanisms
- 3D object watermarking