Call for papers
The explosive growth of new media distribution channels in the Internet and the resulting new production workflows based on computerized tools are forcing media industry to adapt new business models and supporting technologies.
The Fourth Workshop on Automated Media Analysis and Production for Novel TV Services (AIEMPro 2011) aims at fostering the exchange of ideas and best practices between leading experts in industrial and academic research, academic stakeholders of prospective methods, and leading actors in the media industry.
The workshop aims to catalyze the migration towards new ways of producing, broadcasting, and presenting media content, via the introduction of tools for automated multimedia analysis and understanding. At the same time, the workshop will help academic researchers better understand the real-life key requirements that enable higher impact and wider adoption of these methods.
Media production workflows are substantially influenced by new ways of acquiring, elaborating, and publishing audiovisual material, as well as by bandwidth adaptive streaming through Internet portals. In this context, automatic information extraction techniques based on audiovisual content analysis are seen as an interesting and promising option to streamline these processes and lower the total cost of new productions, and as a way to help to disseminate existing archives. This line of development has been fully recognized by the European Broadcasting Union, Time Warner, and other major players in industry. However, so far the potential of these methods is yet to be leveraged in everyday use.
On the delivery side, TV content is generally available as continuous streams of audio-visual frames. For example, in linear television, useful TV programs and inter-programs (like commercials, trailers) are concatenated and broadcasted without any precise and reliable flags that identify the boundaries and times of interest. Thus, media streams have to be first macro-segmented and segments must be classified and mined with the objective to recover the original structure of the stream.
In general, programs need to be described, structured, summarized, classified following their genre and theme, indexed, stored and retrieved. Events of interest also need to be detected in programs. Events may concern for instance goals and important actions in football footage, but also each time a specific person appears on screen. Automatically understanding program storylines is also an important issue when creating summaries. It also provides high-level metadata that improves the effectiveness of the search.
At previous editions of this workshop (AIEMPro 2010 and CBTV 2009 ), work showing important progress in several steps of the media production process has been presented. But there are still many challenges ahead for leveraging the application these promising research results in production processes, e.g., generalization across content types, formats and genres, adaptation to requirements in different sub-processes, reliable evaluation methods assessing the fitness of a method for a certain purpose and novel intuitive user interfaces.
The workshop aims at attracting the attention of researchers and practitioners in the field of automatic information extraction based on audiovisual content analysis on the problems related with media production processes. In particular, the objective is to analyze the impact and performances of these tools in real-life applications, and on real-life material.
Authors are encouraged to submit papers on which they enlighten the features of existing or novel systems and applications in the key aspects of future media production based on automated information extraction. This includes acquisition, editing, search and retrieval, publishing, archiving and repurposing of audiovisual material.
Topics of interest include, but are not limited to:
- Multimodal topic and concept detection, categorization, and genre / editorial format detection
- Automatic speech recognition and keyword spotting
- Automated copyright infringement detection and watermarking
- Duplicate and near duplicate detection
- Content segmentation (video and/or audio)
- Content summarization
- High- and low-level acoustic, visual, and multimodal indexing
- Multimodal personality identification (e.g., combined face and speaker identification)
- Visual and acoustic event detection in multimedia streams
- Thematic and semantic clustering of television programmes
- (Semi-) Automated repurposing of archived material
- Computer-assisted news production
- Context based retrieval and indexing of news content
- Assisted material selection
- Efficient navigation and retrieval of live multimedia streams
- Automated trust estimation and opinion mining for news
- Ontologies and metadata schemes for media production and their applications
- HCI for efficient annotation and retrieval
- Automated cross-media linking
- Evaluation methods for multimedia content analysis tools data sets and standard resources
- Architectures and best-practices for integrating annotation tools in production workflows
- Tools for automatic handling information extraction tools results by using MPEG-7 AVDP
- User studies and usage trends
We are particularly interested in tools and solutions that are scalable and operate in real-time.
A considerable amount of audiovisual material taken from some of the major European broadcasters' archives is available for researchers. Perspective authors are encouraged to test their technologies on this real-life media production material.
Test material for experimentation is available at the Online Media Asset Management System "Mammie" provided by the organizers. Download is conditioned to some terms and conditions for the use of the material. You can find the full text on the register page of the Mammie system.
By clicking on the link below, you will be redirected to the "Mammie" system. Use the Register link and fill out the form as accurate as possible to request your access to the system. After manual verification of your request (which can take up to 5 days), you will be notified about your approval to use the system.