Jianrong Wang

In recent years, barefoot footprint recognition has become increasingly important in criminal investigations. However, the high complexity and diverse features of footprint images pose challenges to traditional methods and CNNs due to their limited local receptive fields. To address this problem, we propose OPTransformer, an improved Vision Transformer (ViT) that integrates global and local features for robust barefoot footprint recognition. Specifically, an overlapped patch strategy is introduced to enhance inter-patch information exchange, a depthwise separable convolution module is embedded to strengthen fine-grained local feature extraction with low computational cost, and an efficient channel attention (ECA) mechanism is used to recalibrate feature channels. Experiments on the SXU-FOOT dataset under a closed-set protocol show that OPTransformer achieves a Top-1 accuracy of 95.33%, outperforming ViT, ResNet, Swin Transformer, and EfficientNetV2. The reported ablations support the contribution of individual modules but do not isolate their pairwise interactions. This work provides an approach to closed-set barefoot footprint classification and can be extended in the future with multimodal fusion and lightweight model design.

Barefoot footprints provide valuable biometric evidence for forensic investigation and suspect identification. However, the rich structural details and diverse scale distributions of footprint images pose significant challenges for accurate fine-grained region recognition. Existing methods often struggle to effectively model contextual information and structural characteristics, limiting their ability to characterize subtle morphological variations. To address these issues, this paper proposes a robust framework for contextual–structural representation learning in barefoot footprint region recognition. A context-guided spatial feature reconstruction pyramid network (CGSFRPN) is introduced to enhance semantic consistency and feature interactions across different scales. To improve directional perception and edge detail modeling, a Direction-Aware Pinwheel Convolution (DAPConv) is incorporated to capture structural patterns with diverse orientations. Furthermore, a ConvFormer module based on the MetaFormer architecture is employed to facilitate efficient semantic representation and cross-region information integration. By jointly exploiting contextual information and structural characteristics, the proposed framework achieves robust recognition of fine-grained footprint regions, including toes, arches, and heels. Experiments conducted on a proprietary barefoot footprint dataset demonstrate that the proposed method achieves a mean Average Precision (mAP) of 83.2%, outperforming several mainstream object detection approaches and exhibiting superior visualization performance. These results highlight the effectiveness and robustness of the proposed framework for fine-grained barefoot footprint analysis and its potential applications in intelligent forensic identification.

With the frequent emergence and spread of new infectious diseases, poses severe threats to public health, and the government often relies on non-pharmaceutical interventions to cope. Meanwhile, the impact of media information on public behavior and health awareness is increasingly significant, becoming an indispensable factor in epidemic prevention and control. This paper constructs an SVEIR-M infectious disease model integrating age structure and media coverage mechanisms, depicting the differences in individuals’ acceptance of media information and the effectiveness of vaccination at different age stages. The model introduces complex factors such as immune waning, latent development age, and media information dissemination, and systematically analyzes the existence and stability of disease-free and endemic equilibrium points using partial differential equations and Volterra integral tools. It is proved that the basic reproduction number R₀ plays a threshold role in characterizing the dynamical properties of the system, and the global stability of equilibrium points under different conditions is demonstrated by constructing Lyapunov functions. In addition, the uniform persistence of the system is analyzed, and the correctness of the theoretical analysis is verified through numerical simulations, discussing the impact of different intervention measures on epidemic development. The research results show that media publicity and vaccination can significantly reduce the infection and mortality rates, and their combination can more effectively control the spread of the epidemic.

Insulator defect detection in transmission lines has become increasingly important for the safe operation of power systems within Internet of Things (IoT)-enabled smart grid infrastructures. However, complex backgrounds, diverse defect characteristics, and limited computing resources at IoT edge nodes pose significant challenges to existing detection methods. In existing defect detection methods, the fixed receptive field of convolutional neural networks (CNNs) limits their ability to adapt to multiscale defects. Defect detection methods based on the Transformer architecture offer advantages in global modeling, but they are limited by their inability to accurately locate defect locations and high computational complexity, hindering their deployment on resource-constrained unmanned aerial vehicle (UAV) platforms as intelligent IoT sensing devices. Therefore, this article proposes an insulator defect monitoring framework based on dynamic adaptive perception and a local–global attention (LGA) enhancement fusion network, i.e., dynamic perception and precise localization detection Transformer (DPL-DETR), effectively addressing the core challenges of real-time intelligent insulator defect detection at the IoT edge. First, a dynamic perception scale-adaptive network (DSNet) is designed, which leverages a learnable dynamic convolution kernel mechanism to significantly enhance the model’s adaptive perception capability for defects of varying scales while maintaining lightweight characteristics essential for edge deployment. Second, a precise attention spatial enhancement network (PASEN) is constructed to achieve accurate localization and feature enhancement of defect regions in complex background environments while substantially reducing computational complexity to meet real-time IoT application requirements. In addition, an LGA fusion (LGAF) module is proposed, which employs a dual-branch collaborative mechanism to effectively fuse local detail features with global semantic information, thereby enhancing the model’s capability to recognize defects of varying morphologies under diverse environmental conditions. Experiments on the Transmission Line-ID and OPDL datasets show that our proposed defect detection method DPL-DETR achieves an effective balance between performance and efficiency, with mAP50 reaching 98.25% and mAP50–95 reaching 68.74%. Furthermore, the number of parameters is reduced by nearly 50%, and the computational complexity decreases to 79.1 GFLOPs, facilitating efficient deployment on UAV-mounted IoT edge devices. The results of all evaluation metrics surpass the best existing detection methods, demonstrating the potential of DPL-DETR as a reliable intelligent perception solution for IoT-based power infrastructure monitoring.

Infectious diseases not only pose a threat to human health, but also have an impact on social order and economic stability. Given the complexity and dynamism of the transmission process of infectious diseases, single dimensional information is difficult to provide comprehensive and accurate understanding of infectious diseases when characterizing transmission patterns and evolutionary characteristics. Therefore, it is crucial to delve into the interaction mechanism of multiple types of information, including positive, negative, and policy regulatory information, in the process of infectious disease transmission. Based on this, this paper innovatively constructs an infectious disease transmission model based on multidimensional media information interaction, which fully considers the heterogeneity characteristics of networks blocks and the co-evolution between information transmission and disease transmission. First, a two-layer infectious disease networks model based on multidimensional media information influence and regional heterogeneity was constructed. The basic reproduction number [ ] of the model was derived by using methods such as differential equation dynamics and matrix analysis. And with the help of Lyapunov function stability theory and Lasalle invariance principle, the dynamic properties of disease-free equilibrium points and endemic equilibrium points were rigorously proved. Second, in order to achieve effective intervention and optimized control of the transmission process of infectious diseases, the optimal control equation for infectious disease transmission was designed with the objective function of minimizing the number of infected people and control costs. The numerical solution of the optimal control strategy was obtained by introducing the Pontryagin maximum principle and using numerical algorithms. Finally, a series of numerical simulation experiments were conducted to verify the effectiveness and accuracy of the theoretical analysis of the model. Therefore, it is of great significance to reasonably guide and regulate the dissemination characteristics and content of multidimensional media information in order to achieve effective prevention and control of infectious diseases.

To address the problems of relying on electronic repositories and being vulnerable to network influence in obtaining key information of literature in mainstream literature management software, an optimized Cascade R-CNN-based algorithm is proposed for detecting key information of literature. Based on the cascade regional convolutional neural network (Cascade R-CNN), a Res2Net feature extraction network embedded with SimAM and CBAM attention modules was used to acquire features at different scales. The bi-directional feature pyra-mid network (BiFPN) is introduced. The BiFPN enhances the feature represen-tation capability by adding residual links and weight fusion to better balance the feature information at different scales; improves the algorithm’s ability to distinguish hard samples by using the focal loss function to balance the unbalanced number of positive and negative samples; and optimizes the anchor mechanism to accelerate the network convergence. The experimental results show that the optimized Cascade R-CNN object detection algorithm has the best comprehensive performance, and the average accuracy (AP_0.5) reaches 95.5%, which is 3.5% better than the original model, and takes 7.95 FPS to achieve better detection results.

The equipment in railway station is complicated and diverse, and the health status assessment and prediction of equipment is crucial to the safe and stable operation of stations. Graph Neural Networks (GNNs) effectively combine graph data with deep learning technology, which has stronger data and knowledge representation capability and can efficiently handle some non-Euclidean spatial data problems with irregular station equipment associated network structure. Based on this, this paper takes the automatic gate machine and X-ray security checker as an example and proposes a health status assessment and prediction scheme for railway passenger station equipment based on Graph Long Short-Term Memory (G-LSTM) neural network. This paper first analyzes the main factors affecting the health status of passenger station equipment, as well as the correlation between the equipment. Then, the initial graph network structure of the passenger station equipment is constructed, and the G-LSTM model is used to evaluate and predict the health status of the passenger station equipment. Finally, this paper takes the automatic gate machine and X-ray security checker of a high-speed railway station in Beijing as an example to verify the proposed method. The experimental results show that all evaluation metrics perform well, indicating that the G-LSTM model has high accuracy in assessing and predicting the health status of automatic gate machine and X-ray security checker. This paper realizes the health status assessment and prediction of railway passenger station equipment, which can provide some reference for the Prognostics and Health Management (PHM) of equipment in railway stations.

In the future, the Internet of things will reduce the cell radius and increase the number of low-power nodes to support thousands of times of traffic growth under 5G. As a virtual multiple input multiple output technology, cooperative communication technology can solve these problems effectively. According to the evolution characteristics of cooperative communication networks, a multi-domain cooperative communication network evolution model with preferential attachment and random attachment is constructed in this article. And then, the network properties and robustness are analyzed using the mean-field method and different attacks. Aiming at the resource constraints and resource allocation problems of communication nodes, a relay selection strategy based on the combination of maximum degree and minimum clustering coefficient is proposed. The simulation results show that the relay node selection strategy based on the combination of maximum degree and minimum clustering coefficient has significant advantages in selection steps and selection time, which greatly enhanced the performance of relay selection in multi-domain cooperative communication networks. Through real-time monitoring and updating of the performance and security indicators of the multi-domain cooperative communication networks, it provides a strong guarantee for the node deployment and security management of the Internet of things cooperative communication system.