aceMedia
From Wikipedia, the free encyclopedia
This article is orphaned as few or no other articles link to it. Please help introduce links in articles on related topics. (February 2008) |
This article does not cite any references or sources. (February 2008) Please help improve this article by adding citations to reliable sources. Unverifiable material may be challenged and removed. |
aceMedia is an EU-funded project that aims at creating a system for multimedia content management.[1]
Features of the aceMedia software include:
- it runs on several platforms (PC, set-top box, mobile phone, web interface)
- it manages different users that may that share content, and their associated access rights
- it includes image, language and video analysis functions that can automatically annotate content or handle textual queries in natural language
It was developed by 15 European partners[2] that can be classified in:
- research laboratories that provide innovative technologies from their expertise fields
- high-tech companies that may re-use or sell (parts of) the software
- companies that manage image or video databases, that provide the test content
The project extended from 2004 to 2007, with the final review presented in February 2008.
Contents |
[edit] Description
The software comes in two flavours, the Personal Content System (PCS) application for home users and the Commercial Content Management (CCM) framework.
[edit] The PCS app
The PCS is a GUI where the user can import images and videos, organise in folders, annotate them and call content analysis tools on them.
The content analysis tools include:
- a person detector that finds standing people in images
- a face analysis method that finds and recognizes faces in images
- a classifier that determines properties on a whole image (like whether it was taken from inside or outside, whether it was by night, etc.)
- a segmentation engine that detects textured areas on images (like trees, sky, rock, etc.)
[edit] The CCM app
It includes a web interface from which users can rate images and videos, and search them. Searching can be done with different criteria:
- textual search. Text queries in various languages are "understood" by the system, and searches are performed using all available information, including that from the image analysis algorithms
- global image similarity, using global image descriptors
- precise image similarity: the user can define image regions (s)he is interested in on an example image, and the system finds the most relevant images
A reasoning system combines annotations to make sure they are coherent. For example, on an image, a sky region is supposed to be above a rock region.
[edit] Architecture
The PCS application and most of the CCM application are based on a Java framework that manages the content (stored as files) and the metadata (stored in a MySQL database).
The content analysis tools are implemented in C or C++ and called by the framework.
The server side of the web interface is in C#.
[edit] Future
Some technologies developed during the lifespan of the project will be used by the participating companies. The main framework is to be published in open-source form.