A benchmarking campaign for the multimodal detection of violent scenes in movies

Claire-Hélène Demarty (1) , Cédric Penet (1, 2) , Guillaume Gravier (2) , Mohammad Soleymani (3)

1 Technicolor R & I [Cesson Sévigné]
2 TEXMEX - Multimedia content-based indexing
3 Department of Computing [London]

We present an international benchmark on the detection of violent scenes in movies, implemented as a part of the multimedia benchmarking initiative MediaEval 2011. The task consists in detecting portions of movies where physical violence is present from the automatic analysis of the video, sound and subtitle tracks. A dataset of 15 Hollywood movies was carefully annotated and divided into a development set and a test set containing 3 movies. Annotation strategies and resolution of borderline cases are discussed at length in the paper. Results from 29 runs submitted by the 6 participating sites are analyzed. The first year's results are promising, but considering the use case, there is still a large room for improvement. The detailed analysis of the 2011 benchmark brings valuable insight for the implementation of future evaluation on violent scenes detection in movies.

Domaines: Multimédia [cs.MM]

Claire-Hélène Demarty, Cédric Penet, Guillaume Gravier, Mohammad Soleymani. A benchmarking campaign for the multimodal detection of violent scenes in movies. European Conference on Computer Vision, Workshop on Information Fusion in Computer Vision for Concept Recognition, 2012, Italy. ⟨hal-00767036⟩