Violence Detection Enhancement by Involving Convolutional Block Attention Modules Into Various Deep Learning Architectures: Comprehensive Case Study for UBI-Fights Dataset

Abbass, Mahmoud Abdelkader Bashery; Kang, Hyun‐Soo

doi:10.1109/access.2023.3267409

Cited by 7 publications

(4 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Finally, a dense layer with 10 units (i.e., one output per each driver case class) and a softmax activation function, are added to extract the final probability per each class. The hyper-parameters for the proposed architecture are adjusted based on the main author's experience of handling such cases over the course of many papers [28][29][30][31].…”

Section: Overall Proposed Architecture Layoutmentioning

confidence: 99%

MobileNet-Based Architecture for Distracted Human Driver Detection of Autonomous Cars

Abbass,

Ban

2024

Electronics

Self Cite

View full text Add to dashboard Cite

Distracted human driver detection is an important feature that should be included in most levels of autonomous cars, because most of these are still under development. Hereby, this paper proposes an architecture to perform this task in a fast and accurate way, with a full declaration of its details. The proposed architecture is mainly based on the MobileNet transfer learning model as a backbone feature extractor, then the extracted features are averaged by using a global average pooling layer, and then the outputs are fed into a combination of fully connected layers to identify the driver case. Also, the stochastic gradient descent (SGD) is selected as an optimizer, and the categorical cross-entropy is the loss function through the training process. This architecture is performed on the State-Farm dataset after performing data augmentation by using shifting, rotation, and zooming. The architecture can achieve a validation accuracy of 89.63%, a validation recall of 88.8%, a validation precision of 90.7%, a validation f1-score of 89.8%, a validation loss of 0.3652, and a prediction time of about 0.01 seconds per image. The conclusion demonstrates the efficiency of the proposed architecture with respect to most of the related work.

show abstract

Section: Overall Proposed Architecture Layoutmentioning

confidence: 99%

MobileNet-Based Architecture for Distracted Human Driver Detection of Autonomous Cars

Abbass,

Ban

2024

Electronics

Self Cite

View full text Add to dashboard Cite

show abstract

“…While there are datasets like UT-Interaction, UCF101 and the Kinetics action recognition dataset that include examples of actions like boxing and punching, these datasets feature lowerresolution inputs and do not align with the requirement of video feeds capturing physical abuse actions. After extensive search [33], [34], we were able to identify two publicly available datasets UBI-Fights [35] and UCF Crime [36], that have real life video clips of common people hitting, kicking, beating and such fight sequences. This was the nearest match that we could get for detecting physical abuse.…”

Section: Stage-3mentioning

confidence: 99%

Intelligent Surveillance Framework for Physical Abuse Detection

2023

Preprint

View full text Add to dashboard Cite

Physical abuse is a societal problem irrespective of whether the society is rich, educated or so. Mostly physically weaker section like children, women and old age people are most vulnerable to it especially in cases of domestic violence or workplace aggression. If the abuser and victim share a relationship then reporting the abuse becomes all the more challenging. Hence an intelligent surveillance system is the need of the hour. In this paper we propose a hybrid deep learning framework for physical abuse detection and prevention through human action recognition. 3D convolution neural network (CNN) based deep learning model is built to analyze this complex human action. Using transfer learning ResNet-18 and GoogleNet model are trained for publicly available UBI fight and UCF crime dataset for detecting physical abuse in the video. The 2D kernels are converted to 3D kernels for both temporal and spatial feature extraction in the video. A layer of bilinear LSTM is added to gather long term temporal information, so that there is improvement in capturing the human action information. This hybrid model used for physical abuse detection is the novel approach and the results show that the performance parameters such as precision, accuracy, specificity and sensitivity score improved as we moved from 2D kernels to 3D, and there is remarkable improvement when bilinear LSTM was added to the deep learning model.

show abstract

“…Improving the accuracy of motion recognition using both visual and motion information is a major feature of I3D. Additionally, anomaly detection studies continue to be published [51][52][53][54][55][56][57][58][59][60].…”

Section: I3dmentioning

confidence: 99%

Anomaly Detection Based on a 3D Convolutional Neural Network Combining Convolutional Block Attention Module Using Merged Frames

Hwang,

Kang

2023

Sensors

Self Cite

View full text Add to dashboard Cite

With the recent rise in violent crime, the real-time situation analysis capabilities of the prevalent closed-circuit television have been employed for the deterrence and resolution of criminal activities. Anomaly detection can identify abnormal instances such as violence within the patterns of a specified dataset; however, it faces challenges in that the dataset for abnormal situations is smaller than that for normal situations. Herein, using datasets such as UBI-Fights, RWF-2000, and UCSD Ped1 and Ped2, anomaly detection was approached as a binary classification problem. Frames extracted from each video with annotation were reconstructed into a limited number of images of 3×3, 4×3, 4×4, 5×3 sizes using the method proposed in this paper, forming an input data structure similar to a light field and patch of vision transformer. The model was constructed by applying a convolutional block attention module that included channel and spatial attention modules to a residual neural network with depths of 10, 18, 34, and 50 in the form of a three-dimensional convolution. The proposed model performed better than existing models in detecting abnormal behavior such as violent acts in videos. For instance, with the undersampled UBI-Fights dataset, our network achieved an accuracy of 0.9933, a loss value of 0.0010, an area under the curve of 0.9973, and an equal error rate of 0.0027. These results may contribute significantly to solve real-world issues such as the detection of violent behavior in artificial intelligence systems using computer vision and real-time video monitoring.

show abstract

Violence Detection Enhancement by Involving Convolutional Block Attention Modules Into Various Deep Learning Architectures: Comprehensive Case Study for UBI-Fights Dataset

Cited by 7 publications

References 29 publications

MobileNet-Based Architecture for Distracted Human Driver Detection of Autonomous Cars

MobileNet-Based Architecture for Distracted Human Driver Detection of Autonomous Cars

Intelligent Surveillance Framework for Physical Abuse Detection

Anomaly Detection Based on a 3D Convolutional Neural Network Combining Convolutional Block Attention Module Using Merged Frames

Contact Info

Product

Resources

About