TY - JOUR
T1 - A Novel Violent Video Detection Scheme Based on Modified 3D Convolutional Neural Networks
AU - Song, Wei
AU - Zhang, Dongliang
AU - Zhao, Xiaobing
AU - Yu, Jing
AU - Zheng, Rui
AU - Wang, Antai
N1 - Funding Information:
This work was supported in part by the National Science Foundation Project of China under Grant 61503424, Grant 61331013, and Grant 61701554, in part by the Promotion Plan for Young Teachers’ Scientific Research Ability of the Minzu University of China, in part by the MUC 111 Project, in part by the First Class University and First Class Discipline of the Minzu University of China (Intelligent Computing and Network Security), and in part by the Youth Team Leadership Program.
Publisher Copyright:
© 2013 IEEE.
PY - 2019
Y1 - 2019
N2 - Violent video constitutes a threat to public security, and effective detection algorithms are in urgent need. In order to improve the detection accuracy of 3D convolutional neural networks (3D ConvNet), a novel violent video detection scheme based on the modified 3D ConvNet is proposed. In this paper, the preprocessing method of data is improved, and a new sampling method by using the key frame as dividing nodes is designed. Then, a random sampling method is adapted to produce the input frame sequence. With experimental evaluations on the crowd violence dataset, the results demonstrate the effectiveness of the proposed new sampling method. For three public violent detection datasets: hockey fight, movies, and crowd violence, individualized strategies are implemented to suit the varied clip length. For the short clips, the 3D ConvNet is constructed by using the uniform sampling method. For the longer clips, the new frame sampling strategy is adopted. The proposed scheme obtains competitive results: 99.62% on hockey fight, 99.97% on movies, and 94.3% on crowd violence. The experimental results show that our method is simple and effective.
AB - Violent video constitutes a threat to public security, and effective detection algorithms are in urgent need. In order to improve the detection accuracy of 3D convolutional neural networks (3D ConvNet), a novel violent video detection scheme based on the modified 3D ConvNet is proposed. In this paper, the preprocessing method of data is improved, and a new sampling method by using the key frame as dividing nodes is designed. Then, a random sampling method is adapted to produce the input frame sequence. With experimental evaluations on the crowd violence dataset, the results demonstrate the effectiveness of the proposed new sampling method. For three public violent detection datasets: hockey fight, movies, and crowd violence, individualized strategies are implemented to suit the varied clip length. For the short clips, the 3D ConvNet is constructed by using the uniform sampling method. For the longer clips, the new frame sampling strategy is adopted. The proposed scheme obtains competitive results: 99.62% on hockey fight, 99.97% on movies, and 94.3% on crowd violence. The experimental results show that our method is simple and effective.
KW - 3D ConvNet
KW - Violent video detection
KW - key frame extraction
UR - http://www.scopus.com/inward/record.url?scp=85065229819&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85065229819&partnerID=8YFLogxK
U2 - 10.1109/ACCESS.2019.2906275
DO - 10.1109/ACCESS.2019.2906275
M3 - Article
AN - SCOPUS:85065229819
SN - 2169-3536
VL - 7
SP - 39172
EP - 39179
JO - IEEE Access
JF - IEEE Access
M1 - 8669768
ER -