Gait recognition is a method of person identification at a distance through vision technology. The method performs effective information extraction by analyzing the contour changes of pedestrians during movement. Low-quality silhouettes are inevitably present in the sequences due to the tendency of gait silhouette data to shape features and the outdoor environment’s discomfort with the original segmentation method. In this paper, we propose a gait recognition method GaitSE with enhanced spatial extraction capability, which combines the Global View Capture Module(GCM) and Contour Multi-Frame Attention (CMFA) mechanism. First, the global view capture module is constructed based on the concept of large convolutional kernels to obtain an Effective Receptive Field(ERF) close to the global one. Specifically, a reparameterization approach is used in GCM to fuse multi-scales dilated convolutional kernels to capture the fine-grained information ignored by large convolutional kernels. In addition, we propose a novel contour multi-frame attention mechanism, which can adaptively adjust the model’s attention to different contour frames in the same sequence according to the input data, thus with obtaining more discriminative features. Experiments show that our proposed method produces effective boosts on all popular outdoor gait datasets. GaitSE achieves 73.1% accuracy on Gait3D and 71.6% on GREW.

Significant advancements have been made in developing parametric models for digital humans, with various approaches concentrating on parts such as the human body, hand, or face. Nevertheless, connectors such as the neck have been overlooked in these models, with rich anatomical priors often unutilized. In this paper, we introduce HACK (Head-And-neCK), a novel parametric model for constructing the head and cervical region of digital humans. Our model seeks to disentangle the full spectrum of neck and larynx motions, facial expressions, and appearance variations, providing personalized and anatomically consistent controls, particularly for the neck regions. To build our HACK model, we acquire a comprehensive multi-modal dataset of the head and neck under various facial expressions. We employ a 3D ultrasound imaging scheme to extract the inner biomechanical structures, namely the precise 3D rotation information of the seven vertebrae of the cervical spine. We then adopt a multi-view photometric approach to capture the geometry and physically-based textures of diverse subjects, who exhibit a diverse range of static expressions as well as sequential head-and-neck movements. Using the multi-modal dataset, we train the parametric HACK model by separating the 3D head and neck depiction into various shape, pose, expression, and larynx blendshapes from the neutral expression and the rest skeletal pose. We adopt an anatomically-consistent skeletal design for the cervical region, and the expression is linked to facial action units for artist-friendly controls. We also propose to optimize the mapping from the identical shape space to the PCA spaces of personalized blendshapes to augment the pose and expression blendshapes, providing personalized properties within the framework of the generic model. Furthermore, we use larynx blendshapes to accurately control the larynx deformation and force the larynx slicing motions along the vertical direction in the UV-space for precise modeling of the larynx beneath the neck skin. HACK addresses the head and neck as a unified entity, offering more accurate and expressive controls, with a new level of realism, particularly for the neck regions. This approach has significant benefits for numerous applications, including geometric fitting and animation, and enables inter-correlation analysis between head and neck for fine-grained motion synthesis and transfer.

In this paper, we study the spectrum efficiency (SE), energy efficiency (EE), and economic efficiency (ECE) for a heterogeneous cellular architecture that separates the indoor and outdoor scenarios for beyond 5G (B5G) wireless communication systems. For outdoor scenarios, massive multiple-input-multiple-output (MIMO) technologies and distributed antenna systems (DASs) at sub-6 GHz frequency bands are used for long-distance communications. For indoor scenarios, millimeter-wave (mmWave) and beamforming communication technologies are deployed at wireless indoor access points (IAPs) to provide high-speed short-range services to indoor users. Mathematical expressions for the system capacity, SE, EE, and ECE are derived using a proposed realistic power consumption model. The results shed light on the fact that the proposed network architecture is able to improve SE and EE by more than three times compared to those conventional network architectures. The analysis of system performance in terms of SE, EE, ECE, and their trade-off results in the observation that the proposed network architecture offers a promising solution for future B5G communication systems.

Cooperative localization has emerged as a promising technique which can complement or even replace global positioning systems (GPS) in many practical scenarios, such as GPS-denied environments. In this paper, we concentrate on the distributed cooperative localization design. The conventional distributed cooperative localization approaches usually yield high computational complexity and communication overhead, due to the lack of an efficient link selection mechanism. For this purpose, we propose a novel concept named opportunistic cooperative localization, based on which each agent is able to select the most informative links rather than utilize all the possible links in a distributed, self-organized, and self-optimized manner. To achieve effective opportunistic selection, overlapping coalition formation (OCF) game is employed. In addition, we also provide an optimized terminating criterion, based on which the agents will be able to know whether and when they are well localized, and thus can terminate their localization procedure efficiently. Through simulations, we observe that by virtue of the proposed OCF game based opportunistic cooperative localization scheme along with the provided terminating criterion, the limitations of conventional distributed cooperative localization can be alleviated at the cost of negligible performance degradation.

Nowadays, more and more communication systems are deployed under mobile environments with very high velocity. This paper focuses on the channel estimation problem in vehicular scenarios, which is very challenging in view of the extremely time-varying characteristics of mobile channels. Specifically, we propose a novel channel estimator named constructed data pilot (CDP) estimator for the current communication standards by fully exploiting the channel correlation characteristics across two concatenated symbols. On the basis of the CDP estimator, we further resort to two efficient techniques to improve its performance over the entire signal-to-noise-ratio (SNR) region. For the first technique, the time-variant mobile channel is modeled as a first-order Markov process so that the exact autocorrelation value of the two adjacent symbols can be derived. For the second technique, the SNR is estimated and serves as a priori information. Simulation results reveal that our proposed channel estimators outperform existing alternatives with lower computational complexity.

Recently, localization has become an indispensable technique for wireless applications. In view of the limitation of global position system (GPS) in certain environments, alternative approaches are in demand. In this paper, we consider a cooperative localization approach named sum-product algorithm over a wireless network (SPAWN). Although SPAWN theoretically facilitates cooperative localization, it has several practical limitations. Specifically, SPAWN results in high computational complexity and increased network traffic. The main complexity of SPAWN lies in the selection of agents/anchors involved in the cooperative localization. To this end, we formulate the agent/anchor selection problem into a network formation game. Together with a practical limit on the number of agents/anchors used for cooperative localization, our proposed approach can markedly reduce the computational complexity and the resultant network traffic. Simulations show that these advantages come with a slight degradation in the localization mean squared error (MSE) performance.