Multi-modal based violent movies detection in video sharing sites

Authors:
Xingyu Zou;Ou Wu;Qishen Wang;Weiming Hu;Jinfeng Yang
Affiliations:
College of Aviation Automation, Civil Aviation University of China, Tianjin, China;National Laboratory of Pattern Recognition, Chinese Academy of Sciences, Beijing, China;National Laboratory of Pattern Recognition, Chinese Academy of Sciences, Beijing, China;National Laboratory of Pattern Recognition, Chinese Academy of Sciences, Beijing, China;College of Aviation Automation, Civil Aviation University of China, Tianjin, China
Venue:
IScIDE'12 Proceedings of the third Sino-foreign-interchange conference on Intelligent Science and Intelligent Data Engineering
Year:
2012

Citing 7
Cited 0

Automatic partitioning of full-motion video

Multimedia Systems
Person-on-Person Violence Detection in Video Data

ICPR '02 Proceedings of the 16 th International Conference on Pattern Recognition (ICPR'02) Volume 1 - Volume 1
Semantic context detection based on hierarchical audio models

MIR '03 Proceedings of the 5th ACM SIGMM international workshop on Multimedia information retrieval
Automatically selecting shots for action movie trailers

MIR '06 Proceedings of the 8th ACM international workshop on Multimedia information retrieval
Violence content classification using audio features

SETN'06 Proceedings of the 4th Helenic conference on Advances in Artificial Intelligence
Audio-Visual fusion for detecting violent scenes in videos

SETN'10 Proceedings of the 6th Hellenic conference on Artificial Intelligence: theories, models and applications
Low bit rate transparent audio compression using adapted wavelets

IEEE Transactions on Signal Processing

Quantified Score

Hi-index	0.00

Visualization

Abstract

In this paper we present a method for the detection of violent movies in video sharing sites. The proposed method operates on three modalities: text, video and audio, the former being collected from the accompanying synopsis and user comments. Towards our goal, a multi-step approach is followed: initially, the text information is utilized to build a pre-classifier which selects the potential violent movie segments. At a second stage, a classifier is adopted, which combines the visual and audio information, in order to classify the potential violent movie segments as "violent" or "non-violent". The experimental results on 220 movie segments from YouKu and TuDou show the effectiveness of our method.