Special Issue "Recent Advances in Deep Learning, Information Fusion, and Features Selection for Video Surveillance Application"

Submission Deadline: 15 April 2021 (closed)
Submit to Special Issue
Guest Editors
Dr. Seifedine Kadry, Beirut Arab University, Lebanon.
Dr. Shuihua Wang, University of Leicester, UK.
Dr. V. Rajinikanth, St Joseph’s College, India.
Dr. Muhammad Attique Khan, HITEC University Taxila, Pakistan.

Summary

In the area of computer vision, human action recognition, gait recognition, and gesture recognition (HARGRGR) are important research areas from the last decade. The most important application of HARGRGR is video surveillance. As the imaging technique improvements and the camera expedient promotions, novel approaches for HAR continuously arise. Nowadays, through camera networks, a lot of videos are captured for human activities. Through these activities, it can be possible to predict the future activities of a human. For this purpose, many automated systems are proposed by computer vision researchers using machine learning algorithms. However, the question is how these systems can handle a large number of videos? Also, how they remove redundant or irrelevant information to monitor the required activities? The more recent, deep learning gain a huge success in the area of machine learning to handle a large amount of data with more accuracy as compared to classical techniques. For HARGRGR, deep learning can be more useful because it requires a large amount of data for training.

Sometimes, the deep learning models are trained on complex imaging datasets and due to these complex datasets, the required accuracy cannot be achieved. Therefore, it is possible to fuse two or more than two deep neural networks (layers information, features, etc.). But the question is that how the fusion process impact the system computational time? This problem can be resolve by employing feature reduction techniques.

This special issue aims to gather the achievemen of deep learning, information fusion, and feature selection in fields of action recognition, gait recognition, and gesture recognition.


Keywords
• Human action recognition using deep learning for large video datasets
• Human gait recognition using deep learning
• Human gesture recognition using deep learning
• Deep learning models information fusion for action recognition, gait recognition, and gesture recognition
• Features fusion for action recognition, gait recognition, and gesture recognition
• Features selection and action recognition
• Gesture recognition and features selection
• Gait recognition and features selection
• Gait recognition in the real-time camera network using deep learning
• Learning a deep learning model using body parts for action recognition

Published Papers

  • Video Analytics Framework for Human Action Recognition
  • Abstract Human action recognition (HAR) is an essential but challenging task for observing human movements. This problem encompasses the observations of variations in human movement and activity identification by machine learning algorithms. This article addresses the challenges in activity recognition by implementing and experimenting an intelligent segmentation, features reduction and selection framework. A novel approach has been introduced for the fusion of segmented frames and multi-level features of interests are extracted. An entropy-skewness based features reduction technique has been implemented and the reduced features are converted into a codebook by serial based fusion. A custom made genetic algorithm is implemented on… More
  •   Views:45       Downloads:19        Download PDF

  • Convolutional Bi-LSTM Based Human Gait Recognition Using Video Sequences
  • Abstract Recognition of human gait is a difficult assignment, particularly for unobtrusive surveillance in a video and human identification from a large distance. Therefore, a method is proposed for the classification and recognition of different types of human gait. The proposed approach is consisting of two phases. In phase I, the new model is proposed named convolutional bidirectional long short-term memory (Conv-BiLSTM) to classify the video frames of human gait. In this model, features are derived through convolutional neural network (CNN) named ResNet-18 and supplied as an input to the LSTM model that provided more distinguishable temporal information. In phase II,… More
  •   Views:180       Downloads:173        Download PDF