@INPROCEEDINGS{1479Senst2015,
	AUTHOR = {Tobias Senst and Volker Eiselein and Thomas Sikora},
	TITLE = {[Best Paper Award] A Local Feature based on Lagrangian Measures for Violent Video Classification},
	BOOKTITLE = {6th IET International Conference on Imaging for Crime Detection and Prevention},
	YEAR = {2015},
	MONTH = jul,
	EDITOR = {Georgios Chaitas, Sergio A Velastin},
	PUBLISHER = {IET Digital Library},
	PAGES = {1--6},
	ADDRESS = {UK, London},
	NOTE = {ISBN: 978-1-78561-131-5},
	PDF = {http://elvera.nue.tu-berlin.de/files/1479Senst2015.pdf},
	URL = {http://elvera.nue.tu-berlin.de/files/1479Senst2015.pdf},
	ABSTRACT = {Lagrangian theory provides a diverse set of tools for continuous motion analysis. Existing work shows the applicability of Lagrangian method for video analysis in several aspects. In this paper we want to utilize the concept of Lagrangian measures to detect violent scenes. Therefore we propose a local feature based on the SIFT algorithm that incooperates appearance and Lagrangian based motion models. We will show that the temporal interval of the used motion information is a crucial aspect and study its influence on the classification performance. The proposed LaSIFT feature outperforms other state-of-the-art local features, in particular in uncontrolled realistic video data. We evaluate our algorithm with a bag-of-word approach. The experimental results show a significant improvement over the state-of-the-art on current violent detection datasets, i.e. Crowd Violence, Hockey Fight.}
}