Gait is a unique biometric trait with several useful properties. It can be recognized remotely and without the cooperation of the individual, with low-resolution cameras, and it is difficult to obscure. Therefore, it is suitable for crime investigation, surveillance, and access control. Existing approaches for gait recognition generally belong to the supervised learning domain, where all samples in the dataset are annotated. In the real world, annotation is often expensive and time-consuming. Moreover, convolutional neural networks (CNNs) have dominated the field of gait recognition for many years and have been extensively researched, while other recent methods such as vision transformer (ViT) remain unexplored. In this manuscript, we propose a self-supervised learning (SSL) approach for pretraining the feature extractor using the DINO model to automatically learn useful gait features with the vision transformer architecture. The feature extractor is then used for extracting gait features on which the fully connected neural network classifier is trained using the supervised approach. Experiments on CASIA-B and OU-MVLP gait datasets show the effectiveness of the proposed approach.

The problem of finding the optimal deployment of sensors is becoming increasingly important with the growing expansion of the Internet of Things paradigm and increased usage of sensor networks in different applications. During the installation of sensor networks, sensor placement directly affects the performance of the system. The general problem of determining the position and orientation of the sensors with the goal of optimal coverage of a given environment is NP-hard. In this manuscript, an effective stochastic method for the placement of sensors in arbitrarily given two-dimensional and three-dimensional environments is proposed. The method uses models of generic isotropic and directional sensors with the defined probabilistic coverage. The optimization function combining the environment and sensor models based on the area coverage metric is proposed. Three optimization algorithms are compared with regard to obtained coverage score, execution time, and reliability, and the results are presented and discussed.

Each individual describes unique patterns during their gait cycles. This information can be extracted from the live video stream and used for subject identification. In appearance based recognition methods, this is done by tracking silhouettes of persons across gait cycles. In recent years, there has been a profusion of sensors that in addition to RGB video images also provide depth data in real-time. When such sensors are used for gait recognition, existing RGB appearance based methods can be extended to get a substantial gain in recognition accuracy. In this paper, this is accomplished using information fusion techniques that combine features from extracted silhouettes, used in traditional appearance based methods, and the height feature that can now be estimated using depth data. The latter is estimated during the silhouette extraction step with minimal additional computational cost. Two approaches are proposed that can be implemented easily as an extension to existing appearance based methods. An extensive experimental evaluation was performed to provide insights into how much the recognition accuracy can be improved. The results are presented and discussed considering different types of subjects and populations of different height distributions.

Motor imagery (MI) is a technique of imagining the performance of a motor task without actually using the muscles. When employed in a brain–computer interface (BCI) supported by electroencephalographic (EEG) sensors, it can be used as a successful method of human–computer interaction. In this paper, the performance of six different classifiers, namely linear discriminant analysis (LDA), support vector machine (SVM), random forest (RF), and three classifiers from the family of convolutional neural networks (CNN), is evaluated using EEG MI datasets. The study investigates the effectiveness of these classifiers on MI, guided by a static visual cue, dynamic visual guidance, and a combination of dynamic visual and vibrotactile (somatosensory) guidance. The effect of filtering passband during data preprocessing was also investigated. The results show that the ResNet-based CNN significantly outperforms the competing classifiers on both vibrotactile and visually guided data when detecting different directions of MI. Preprocessing the data using low-frequency signal features proves to be a better solution to achieve higher classification accuracy. It has also been shown that vibrotactile guidance has a significant impact on classification accuracy, with the associated improvement particularly evident for architecturally simpler classifiers. These findings have important implications for the development of EEG-based BCIs, as they provide valuable insight into the suitability of different classifiers for different contexts of use.

This study presents a comprehensive approach for single sensor placement optimization in two-dimensional and three-dimensional spaces. A traditional exhaustive search technique and a novel method called recursive exhaustive search are used to place a sensor in a way that maximizes the area coverage metric. Exhaustive search provides a baseline by methodically evaluating all potential placements, while recursive exhaustive search innovates by segmenting the search process into more manageable, recursive steps. Our findings highlight the significant impact of two key parameters, the number of evaluations and the rasterization value, on the achieved coverage and computation time. The results show that the right choice of parameters can significantly reduce the computational effort without compromising the quality of the solution. This underlines the critical need for a balanced approach that considers both computational complexity and placement efficacy. We show that exhaustive search is not feasible for three-dimensional environment models and propose to establish a modified exhaustive search method as a ground truth for the single sensor placement problem. We then explore nature-inspired genetic algorithms and the impact of the number of evaluations of the optimization function for these algorithms on both accuracy and computational cost.

In this study, we present a novel method for creating an environment model suitable for addressing the sensor placement problem. We extract a detailed environment model from a 3D point cloud by identifying spatial boundaries and furniture in indoor spaces and representing them as a series of polygons. To validate our method, we compare its performance against ground truth data, demonstrating high accuracy in both simple and complex environments. Subsequently, we employ the obtained models in a comprehensive experiment that evaluates the effectiveness of six metaheuristic optimization algorithms in solving the sensor placement problem. We examine how the choice of optimization algorithm and the number of sensors impacts the achieved coverage through statistical analysis. With this study, we gain insights into the comparative effectiveness of various evolutionary algorithms in enhancing sensor network design within indoor spaces. In particular, the Artificial Bee Colony algorithm consistently delivered superior results.