This page provides an overview of results from TOSCA-MP. Click on a link in the left column to access a more detailed description and contact information.

A list of open source resources can be found here.

Result name Short description
Metadata Production Management Framework Controlling entity that orchestrates the analysis of AV essence, based on a BPMN2.0 process. It interfaces with administrator GUIs on one hand and with various feature analysis services and storage on the other. It includes a service configuration framework and extended process error handling. (prototype)
Audio Extraction FIMS compliant service for audio essence extraction of an AV input. (prototype)
Architecture for Frame-based metadata An architecture based on REDIS key-value store for efficient storage and retrieval of frame-based metadata
Framework for cost simulation of automated information extraction tools A Java framework for simulating effects on costs of generic segmentation and classification tasks
TV genre classification system A Parallel Neural Network architecture for automatic classification of television genres
Sharpness detection module
A software module for automatically estimating the level of sharpness of video and assessing whether video has been upscaled.
Visual and multimodal linking A software module for determining links between video segments based on visual and textual similarity.
Task-modelling and task-based benchmarking tools A set of tools for working with task models of media production processes and deriving business processes, as well as tools for benchmarking results of media analysis components.
Control & Configuration UI
A web-based user interface for monitoring repository status, invoking and monitoring automatic processing and calling viewing and annotation tools.
AVDP Viewer A lightweight web user interface for visualising analysis results described in an MPEG-7 AVDP document.
Metadata model A metadata model based on MPEG-7 AVDP, along with usage guidelines.
Networked Media Search Engine The Networked Media Search engine provides the mechanisms to analyse and annotate multimedia assets at different level of details –from low-level multimedia features to semantic concepts, for example– and to perform precise and efficient searches
Exploratory-based interface for search The exploratory-based search interface has been designed and implemented with the clear goal in mind of visualize search results from very large and disparate amounts of information, whilst enabling better and faster information access, sharing and discovery
Distributed Repository Framework Framework of modules for realization of customized networked distributed content and metadata repositories for media production
Automatic Speech Transcription Effective automatic transcription systems for several languages
Machine Translation Machine translation
Spoken Language Identification Effective spoken language identification system to process the audio tracks of videos
Soccer analysis System for the analysis of broadcasted soccer games. Included are features as camera View Classification, Player Detection, and Highlight Detection.
Quality analysis Framework for the quality analysis of video content. Included are single and overall quality measures. These do not require any reference video sequence for the quality prediction.
Visual concept detection System for the classification of video content into trained concepts.
Content set The subset of the content collection provided on the EBU MAMMIE (see D4.2.3)
Visual clustering methods Visual clustering methods for novel search visualization paradigms
Named entity recognition in speech Expansion methods for missing named entities in transcribed speech
Cross-lingual clustering Methods for cross-lingual clustering of events based on bilingual probabilistic topic models
Names and faces linking in video Methods for learning name-face alignments in video
Unsupervised recognition and summarization of sport games in sports video Methods for segmentation and summarization of sport games based on textual and visual features
Event detection in wedding videos Methods for detection and summarization of key events in wedding videos


© 2017 TOSCA-MP - Task-Oriented Search and Content Annotation for Media Production
The research leading to the presented results has received funding from the European Union's
Seventh Framework Programme (FP7/2007-2013) under grant agreement n° 287532. - Imprint