Wednesday, 21 Feb 2018
You are here:

Subscribe to our Newsletter

How many eyes has a typical person? (ex: 1)

Twitter Feed

EMail Print


VideoSense ECCV 2012 Workshop in Florence, Italy
13.10.2012 - 13.10.2012
Palazzo dei Congressi - Florence
Special Sessions


VideoSense organised the International Workshop on Information Fusion in Computer Vision for Concept Recognition (IFCVCR 2012) held in conjunction with the 12th European Conference on Computer Vision (ECCV 2012), on Oct 13, 2012, in Firenze, Italy. The workshop has been organised as a supportive MediaEval event to bring together Trecvid and MediaEval community under the same roof and discuss information fusion techniques for concept recognition in video events.

Motivation and Scope: 

Since the last decade large attention has been paid to the methods of recognition of visual concepts both in image and video documents. In visual scenes, the concept is translated as objects and actions/activities of the latter and also of the context in which these activities are performed. Therefore, machine learning approaches have received increasing importance for the building of object models and decision schemes in the whole task of recognition.

Furthermore, the recent advances of the community in this field show that the concept recognition passes through aggregation of multiple sources of information. Visual information can benefit from additional cues extracted from audio or text, involving integration of speech recognition, user-contributed information (e.g. tags, tweets), viewer affective response, social media analysis, but as well as of information recorded from additional sensors such as geo-taggers or Global Positioning System (GPS). Hence the primary problem consists in the fusion of multiple multi-modal information cues.  This problem is of importance in an actual information context as fusion often means "dimensionality reduction" which is of primarily importance for large scale information mining.

Topics of interest included:

  • Feature (early) fusion for visual content description
  • Kernel fusion methods
  • Classifier (late) fusion methods, stacking
  • Bagging and boosting methods
  • Multimodal fusion for multimedia tagging
  • Feature fusion for object detection, recognition and object tracking
  • Geo-tags and other sensor-based social information fusion
  • Normalization for fusion
  • Fusion and dimensionality reduction, PCA, LDA
  • Contextual fusion, metadata
  • Hierarchical and community-based fusion
  • Social media fusion and mining
  • Metric spaces
  • Legal, ethical and social concepts fusion

The call for paper and workshop agenda including list of presentations can be found in Annex 5.

More information can be found on following link:


Palazzo dei Congressi
Country: it