dc.creator | Rendón Segador, Fernando José | es |
dc.creator | Álvarez García, Juan Antonio | es |
dc.creator | Enríquez de Salamanca Ros, Fernando | es |
dc.creator | Deniz, Oscar | es |
dc.date.accessioned | 2021-09-08T09:41:22Z | |
dc.date.available | 2021-09-08T09:41:22Z | |
dc.date.issued | 2021 | |
dc.identifier.citation | Rendón Segador, F.J., Álvarez García, J.A., Enríquez de Salamanca Ros, F. y Deniz, O. (2021). ViolenceNet: Dense Multi-Head Self-Attention with Bidirectional Convolutional LSTM for Detecting Violence. Electronics, 10 (13), 1-16. | |
dc.identifier.issn | 2079-9292 | es |
dc.identifier.uri | https://hdl.handle.net/11441/125566 | |
dc.description.abstract | Introducing efficient automatic violence detection in video surveillance or audiovisual
content monitoring systems would greatly facilitate the work of closed-circuit television (CCTV)
operators, rating agencies or those in charge of monitoring social network content. In this paper we
present a new deep learning architecture, using an adapted version of DenseNet for three dimensions,
a multi-head self-attention layer and a bidirectional convolutional long short-term memory (LSTM)
module, that allows encoding relevant spatio-temporal features, to determine whether a video is
violent or not. Furthermore, an ablation study of the input frames, comparing dense optical flow and
adjacent frames subtraction and the influence of the attention layer is carried out, showing that the
combination of optical flow and the attention mechanism improves results up to 4.4%. The conducted
experiments using four of the most widely used datasets for this problem, matching or exceeding in
some cases the results of the state of the art, reducing the number of network parameters needed
(4.5 millions), and increasing its efficiency in test accuracy (from 95.6% on the most complex dataset
to 100% on the simplest one) and inference time (less than 0.3 s for the longest clips). Finally, to check
if the generated model is able to generalize violence, a cross-dataset analysis is performed, which
shows the complexity of this approach: using three datasets to train and testing on the remaining one
the accuracy drops in the worst case to 70.08% and in the best case to 81.51%, which points to future
work oriented towards anomaly detection in new datasets. | es |
dc.description.sponsorship | Ministerio de Economía y Competitividad TIN2017-82113-C2-1-R | es |
dc.description.sponsorship | MInisterio de Economía y Competitividad TIN2017-82113-C2-2-R | es |
dc.format | application/pdf | es |
dc.format.extent | 16 | es |
dc.language.iso | eng | es |
dc.publisher | MDPI | es |
dc.relation.ispartof | Electronics, 10 (13), 1-16. | |
dc.rights | Attribution-NonCommercial-NoDerivatives 4.0 Internacional | * |
dc.rights.uri | http://creativecommons.org/licenses/by-nc-nd/4.0/ | * |
dc.subject | violence detection | es |
dc.subject | fight detection | es |
dc.subject | deep learning | es |
dc.subject | dense net | es |
dc.subject | bidirectional ConvLSTM | es |
dc.title | ViolenceNet: Dense Multi-Head Self-Attention with Bidirectional Convolutional LSTM for Detecting Violence | es |
dc.type | info:eu-repo/semantics/article | es |
dcterms.identifier | https://ror.org/03yxnpp24 | |
dc.type.version | info:eu-repo/semantics/publishedVersion | es |
dc.rights.accessRights | info:eu-repo/semantics/openAccess | es |
dc.contributor.affiliation | Universidad de Sevilla. Departamento de Lenguajes y Sistemas Informáticos | es |
dc.relation.projectID | TIN2017-82113-C2-1-R | es |
dc.relation.projectID | TIN2017-82113-C2-2-R | es |
dc.relation.publisherversion | https://www.mdpi.com/2079-9292/10/13/1601/htm | es |
dc.identifier.doi | 10.3390/electronics10131601 | es |
dc.contributor.group | Universidad de Sevilla. TIC-134: Sistemas Informáticos | es |
dc.journaltitle | Electronics | es |
dc.publication.volumen | 10 | es |
dc.publication.issue | 13 | es |
dc.publication.initialPage | 1 | es |
dc.publication.endPage | 16 | es |
dc.contributor.funder | Ministerio de Economía y Competitividad (MINECO). España | es |