Research activities focus on the development of technologies for comprehensive analysis and annotation of audio and video contents using signal analysis and machine learning. The extraction of metadata from media data provides the foundation for numerous applications such as automatic tagging, content-based search, and recommendation systems.
The use and exploitation of audiovisual content depend on the availability of meaningful metadata (data describing data). They provide the basis for locating, organizing, and classifying specific content, as well as implementing recommendation systems. Technologies for the automatic extraction of metadata are therefore crucial to make media content truly accessible and usable.
The development of technologies for the automatic analysis and annotation of audiovisual data requires a solid understanding of signal processing and machine learning, along with a good comprehension of the underlying requirements.
Another challenge lies in multimodal analysis and orchestration: extracting metadata from audio, video, and image files involves a variety of processes ranging from preprocessing to feature extraction and classification. Different methods and technologies are employed, requiring flexible integration and orchestration. The integration of heterogeneous data from different sources and formats also requires the selection or development of suitable data models and metadata standards. Media archives often deal with large volumes of data, imposing specific requirements on system architecture, efficiency, and the optimization of the algorithms used.
Furthermore, we are involved in metadata standards, as well as the integration and orchestration of analysis components. We also address privacy concerns and other aspects of trustworthy AI, aiming to provide comprehensive solutions for specific application requirements.
These technologies are applied particularly for tagging and indexing A/V archives, recommendation systems, program analysis, content tracking, and rights management. They are also used for audio-visual biodiversity measurement and to support disinformation detection.
| Jahr Year | Titel/Autor:in Title/Author | Publikationstyp Publication Type |
|---|---|---|
| 2024 | Visual and audio scene classification for detecting discrepancies in video: a baseline method and experimental protocol Apostolidis, Konstantinos; Abeßer, Jakob; Cuccovillo, Luca; Vasileios, Mezaris |
Konferenzbeitrag Conference Paper |
| 2022 | Construction-sAIt: Multi-modal AI-driven technologies for construction site monitoring Abeßer, Jakob; Loos, Alexander; Sharma, Prachi |
Konferenzbeitrag Conference Paper |
| 2016 | A workflow for cross media recommendations based on linked data analysis Aichroth, P.; Berndl, E.; Weißgerber, T.; Kosch, H.; Köllmer, T. |
Konferenzbeitrag Conference Paper |
| 2015 | MICO - Media in Context Aichroth, P.; Kurz, T.; Stadler, H.; Drewes, F.; Björklund, J.; Schlegel, K.; Berndl, E.; Perez, A.; Bowyer, A.; Volpini, A.; Weigel, C. |
Konferenzbeitrag Conference Paper |
| 2011 | Automated detection of errors and quality issues in audio-visual content Kühhirt, U.; Paduschek, R.; Nowak, S. |
Konferenzbeitrag Conference Paper |
| 2008 | Personal television: A crossmodal analysis approach Dunker, Peter; Gruhne, Matthias; Sturtz, S. |
Konferenzbeitrag Conference Paper |
Fraunhofer Institute for Digital Media Technology IDMT