The Best Model of Convolutional Neural Networks Combined with LSTM for the Detection of Interpersonal Physical Violence in Videos

Hugo Calderon-Vilca; Kent Cuadros Ramos; Elmer Diaz Quiroz; Jorge Angeles Rojas; René Calderon Vilca; Alejandro Apaza Tarqui

doi:10.23919/FRUCT52173.2021.9435563

The Best Model of Convolutional Neural Networks Combined with LSTM for the Detection of Interpersonal Physical Violence in Videos

Hugo Calderon-Vilca, Kent Cuadros Ramos, Elmer Diaz Quiroz, Jorge Angeles Rojas, René Calderon Vilca, Alejandro Apaza Tarqui

Producción científica: Capítulo del libro/informe/acta de congreso › Contribución a la conferencia › revisión exhaustiva

2 Citas (Scopus)

Resumen

Citizen insecurity is directly related to interpersonal physical violence, there are algorithms that allow detecting violence in videos; therefore, it is necessary to know which is the best model for detecting violence. For this research, three convolutional neural network models were compared: Xception, InceptionV3 and VGG16 each together with a recurrent LSTM network, to find out which of the models is the best for the detection of interpersonal violence in videos. The three models were trained using the Real Life Violence Situations dataset, then violence and non-violence were classified, as a result, the InceptionV3 model is the best model, managing to classify with an accuracy of 94% compared to the VGG16 and Xception models, which obtained 88% and 93% respectively. Therefore, we recommend the InceptionV3 model for the detection of interpersonal physical violence in citizen security videos.

Idioma original	Inglés
Título de la publicación alojada	Proceedings of the 29th Conference of Open Innovations Association FRUCT, FRUCT 2021
Editores	Sergey Balandin, Yevgeni Koucheryavy, Tatiana Tyutina
Editorial	IEEE Computer Society
Páginas	81-86
Número de páginas	6
ISBN (versión digital)	9789526924458
DOI	https://doi.org/10.23919/FRUCT52173.2021.9435563
Estado	Publicada - 12 may. 2021
Evento	29th Conference of Open Innovations Association FRUCT, FRUCT 2021 - Virtual, Tampere, Finlandia Duración: 12 may. 2021 → 14 may. 2021

Serie de la publicación

Nombre	Conference of Open Innovation Association, FRUCT
Volumen	2021-May
ISSN (versión impresa)	2305-7254

Conferencia

Conferencia	29th Conference of Open Innovations Association FRUCT, FRUCT 2021
País/Territorio	Finlandia
Ciudad	Virtual, Tampere
Período	12/05/21 → 14/05/21

Nota bibliográfica

Publisher Copyright:
© 2021 FRUCT.

Acceder al documento

10.23919/FRUCT52173.2021.9435563

Otros archivos y enlaces

Enlace a la publicación en Scopus

Citar esto

Calderon-Vilca, H., Ramos, K. C., Quiroz, E. D., Rojas, J. A., Calderon Vilca, R., & Tarqui, A. A. (2021). The Best Model of Convolutional Neural Networks Combined with LSTM for the Detection of Interpersonal Physical Violence in Videos. En S. Balandin, Y. Koucheryavy, & T. Tyutina (Eds.), Proceedings of the 29th Conference of Open Innovations Association FRUCT, FRUCT 2021 (pp. 81-86). Artículo 9435563 (Conference of Open Innovation Association, FRUCT; Vol. 2021-May). IEEE Computer Society. https://doi.org/10.23919/FRUCT52173.2021.9435563

Calderon-Vilca, Hugo ; Ramos, Kent Cuadros ; Quiroz, Elmer Diaz et al. / The Best Model of Convolutional Neural Networks Combined with LSTM for the Detection of Interpersonal Physical Violence in Videos. Proceedings of the 29th Conference of Open Innovations Association FRUCT, FRUCT 2021. editor / Sergey Balandin ; Yevgeni Koucheryavy ; Tatiana Tyutina. IEEE Computer Society, 2021. pp. 81-86 (Conference of Open Innovation Association, FRUCT).

@inproceedings{d37730c429614e6c82e57966ff66dc5f,

title = "The Best Model of Convolutional Neural Networks Combined with LSTM for the Detection of Interpersonal Physical Violence in Videos",

abstract = "Citizen insecurity is directly related to interpersonal physical violence, there are algorithms that allow detecting violence in videos; therefore, it is necessary to know which is the best model for detecting violence. For this research, three convolutional neural network models were compared: Xception, InceptionV3 and VGG16 each together with a recurrent LSTM network, to find out which of the models is the best for the detection of interpersonal violence in videos. The three models were trained using the Real Life Violence Situations dataset, then violence and non-violence were classified, as a result, the InceptionV3 model is the best model, managing to classify with an accuracy of 94% compared to the VGG16 and Xception models, which obtained 88% and 93% respectively. Therefore, we recommend the InceptionV3 model for the detection of interpersonal physical violence in citizen security videos.",

author = "Hugo Calderon-Vilca and Ramos, {Kent Cuadros} and Quiroz, {Elmer Diaz} and Rojas, {Jorge Angeles} and {Calderon Vilca}, Ren{\'e} and Tarqui, {Alejandro Apaza}",

note = "Publisher Copyright: {\textcopyright} 2021 FRUCT.; 29th Conference of Open Innovations Association FRUCT, FRUCT 2021 ; Conference date: 12-05-2021 Through 14-05-2021",

year = "2021",

month = may,

day = "12",

doi = "10.23919/FRUCT52173.2021.9435563",

language = "Ingl{\'e}s",

series = "Conference of Open Innovation Association, FRUCT",

publisher = "IEEE Computer Society",

pages = "81--86",

editor = "Sergey Balandin and Yevgeni Koucheryavy and Tatiana Tyutina",

booktitle = "Proceedings of the 29th Conference of Open Innovations Association FRUCT, FRUCT 2021",

address = "Estados Unidos",

}

Calderon-Vilca, H, Ramos, KC, Quiroz, ED, Rojas, JA, Calderon Vilca, R & Tarqui, AA 2021, The Best Model of Convolutional Neural Networks Combined with LSTM for the Detection of Interpersonal Physical Violence in Videos. En S Balandin, Y Koucheryavy & T Tyutina (eds.), Proceedings of the 29th Conference of Open Innovations Association FRUCT, FRUCT 2021., 9435563, Conference of Open Innovation Association, FRUCT, vol. 2021-May, IEEE Computer Society, pp. 81-86, 29th Conference of Open Innovations Association FRUCT, FRUCT 2021, Virtual, Tampere, Finlandia, 12/05/21. https://doi.org/10.23919/FRUCT52173.2021.9435563

The Best Model of Convolutional Neural Networks Combined with LSTM for the Detection of Interpersonal Physical Violence in Videos. / Calderon-Vilca, Hugo; Ramos, Kent Cuadros; Quiroz, Elmer Diaz et al.
Proceedings of the 29th Conference of Open Innovations Association FRUCT, FRUCT 2021. ed. / Sergey Balandin; Yevgeni Koucheryavy; Tatiana Tyutina. IEEE Computer Society, 2021. p. 81-86 9435563 (Conference of Open Innovation Association, FRUCT; Vol. 2021-May).

Producción científica: Capítulo del libro/informe/acta de congreso › Contribución a la conferencia › revisión exhaustiva

TY - GEN

T1 - The Best Model of Convolutional Neural Networks Combined with LSTM for the Detection of Interpersonal Physical Violence in Videos

AU - Calderon-Vilca, Hugo

AU - Ramos, Kent Cuadros

AU - Quiroz, Elmer Diaz

AU - Rojas, Jorge Angeles

AU - Calderon Vilca, René

AU - Tarqui, Alejandro Apaza

PY - 2021/5/12

Y1 - 2021/5/12

N2 - Citizen insecurity is directly related to interpersonal physical violence, there are algorithms that allow detecting violence in videos; therefore, it is necessary to know which is the best model for detecting violence. For this research, three convolutional neural network models were compared: Xception, InceptionV3 and VGG16 each together with a recurrent LSTM network, to find out which of the models is the best for the detection of interpersonal violence in videos. The three models were trained using the Real Life Violence Situations dataset, then violence and non-violence were classified, as a result, the InceptionV3 model is the best model, managing to classify with an accuracy of 94% compared to the VGG16 and Xception models, which obtained 88% and 93% respectively. Therefore, we recommend the InceptionV3 model for the detection of interpersonal physical violence in citizen security videos.

AB - Citizen insecurity is directly related to interpersonal physical violence, there are algorithms that allow detecting violence in videos; therefore, it is necessary to know which is the best model for detecting violence. For this research, three convolutional neural network models were compared: Xception, InceptionV3 and VGG16 each together with a recurrent LSTM network, to find out which of the models is the best for the detection of interpersonal violence in videos. The three models were trained using the Real Life Violence Situations dataset, then violence and non-violence were classified, as a result, the InceptionV3 model is the best model, managing to classify with an accuracy of 94% compared to the VGG16 and Xception models, which obtained 88% and 93% respectively. Therefore, we recommend the InceptionV3 model for the detection of interpersonal physical violence in citizen security videos.

UR - http://www.scopus.com/inward/record.url?scp=85107396919&partnerID=8YFLogxK

U2 - 10.23919/FRUCT52173.2021.9435563

DO - 10.23919/FRUCT52173.2021.9435563

M3 - Contribución a la conferencia

AN - SCOPUS:85107396919

T3 - Conference of Open Innovation Association, FRUCT

SP - 81

EP - 86

BT - Proceedings of the 29th Conference of Open Innovations Association FRUCT, FRUCT 2021

A2 - Balandin, Sergey

A2 - Koucheryavy, Yevgeni

A2 - Tyutina, Tatiana

PB - IEEE Computer Society

T2 - 29th Conference of Open Innovations Association FRUCT, FRUCT 2021

Y2 - 12 May 2021 through 14 May 2021

ER -

Calderon-Vilca H, Ramos KC, Quiroz ED, Rojas JA, Calderon Vilca R, Tarqui AA. The Best Model of Convolutional Neural Networks Combined with LSTM for the Detection of Interpersonal Physical Violence in Videos. En Balandin S, Koucheryavy Y, Tyutina T, editores, Proceedings of the 29th Conference of Open Innovations Association FRUCT, FRUCT 2021. IEEE Computer Society. 2021. p. 81-86. 9435563. (Conference of Open Innovation Association, FRUCT). doi: 10.23919/FRUCT52173.2021.9435563

The Best Model of Convolutional Neural Networks Combined with LSTM for the Detection of Interpersonal Physical Violence in Videos

Resumen

Serie de la publicación

Conferencia

Nota bibliográfica

Acceder al documento

Otros archivos y enlaces

Huella

Citar esto