High-resolution Network Research Articles

Fall detection method based on spatio-temporal coordinate attention for high-resolution networks

Fall behavior is closely related to the high mortality rate of the elderly, so fall detection has become an important and urgent research area in human behavior recognition. However, the existing fall detection methods, suffer from the loss of detailed action information during feature extraction due to the downsampling operation, resulting in subpar performance when detecting falls with similar behaviors such as lying and sitting. To solve the challenges, this study proposes a high-resolution spatio-temporal feature extraction method based on a spatio-temporal coordinate attention mechanism. The method employs 3D convolutions to extract spatio-temporal features and utilizes gradual down-sampling to generate a multi-resolution sub-network, thus realizing multi-scale fusion and perception enhancement of details. In particular, this study designs a pseudo-3D basic block, which simulates the ability of 3D convolution, to ensure the running speed of the network while controlling the number of parameters. Further, a spatio-temporal coordinate attention mechanism is designed to accurately extract the spatio-temporal positional changes of key skeletal points and the interrelationships among them. Long-term dependencies in horizontal, vertical, temporal directions are captured through three one-dimensional global pooling operations. Then the long-range relationships and channel correlations among features are captured by cascading and slicing operations. Finally, the key information is effectively highlighted by performing dot-multiplication operations between the feature maps from the horizontal, vertical and temporal directions and the input feature maps. Experimental results on three typical public datasets show that the proposed method can better extract motion features and improve the accuracy of fall detection.

SwinInver: 3D data-driven seismic impedance inversion based on Swin Transformer and adversarial training

As deep learning becomes increasingly prevalent in seismic impedance inversion, 3D data-driven approaches have garnered substantial interest. However, two critical challenges persist. First, existing methodologies predominantly rely on Convolutional Neural Networks (CNNs), which, due to the inherent locality of convolutional operations, are inadequate in capturing the global context of seismic data. This limitation notably hinders their performance in inverting complex subsurface structures, such as salt bodies. Second, the current inversion frameworks are prone to overfitting, particularly when trained on limited seismic datasets. To address these challenges, we propose SwinInver, a novel backbone network that integrates the Swin Transformer as its fundamental unit, coupled with a high-resolution network design to facilitate comprehensive global modeling of intricate subsurface structures. Furthermore, we incorporate adversarial training to enhance the inversion process and effectively mitigate overfitting. Experimental evaluations demonstrate that SwinInver significantly surpasses conventional CNN-based approaches in both synthetic and field data scenarios, providing a more accurate and reliable framework for seismic impedance inversion.

Vehicular Mini-LED backlight display inspection based on residual global context mechanism

Mini-LED backlight has emerged as a promising technology for high performance LCDs, yet the massive detection of dead pixels and precise LEDs placement are constrained by the miniature scale of the Mini-LEDs. The high-resolution network (Hrnet) with mixed dilated convolution and dense upsampling convolution (MDC-DUC) module and a residual global context attention (RGCA) module has been proposed to detect the quality of vehicular Mini-LED backlights. The proposed model outperforms the baseline networks of Unet, Pspnet, Deeplabv3+, and Hrnet, with a mean intersection over union (Miou) of 86.91%. Furthermore, compared to the four baseline detection networks, our proposed model has a lower root-mean-square error (RMSE) when analyzing the position and defective count of Mini-LEDs in the prediction map by canny algorithm. This work incorporates deep learning to support production lines improve quality of Mini-LED backlights.Graphical abstract

Improving Human Pose Estimation with Integrated Dual Self-Attention Mechanism in High-Resolution Network

Human Pose Estimation (HPE) in computer vision (CV) has garnered significant attention due to its diverse applications. Deep convolutional neural networks (CNNs) may be solutions for addressing this challenge, but still face several critical issues. Many existing models employ serial convolution with pooling, leading to low-resolution outputs that are suboptimal for the precise localisation required in HPE. They often prioritise local feature learning, overlooking crucial contextual relationships between key-points. This work addresses these challenges by proposing a novel approach for enhancing HPE. Firstly, the paper evaluates the high-resolution network (HRNet) and its comparative advantages over other CNN architectures. Secondly, it introduces a dual self-attention (DSA) mechanism designed to enhance the model’s global awareness, thereby enriching feature maps with contextual information. Thirdly, it integrates the DSA mechanism into HRNet, crafting DSA-HRNet. The model performance was tested on the COCO Val 2017 validation dataset, showing improvements of 2.3% in mean average precision (mAP), 3% in AP at 50 (AP50), and 2.7% in AP at 75 (AP75). Finally, the work includes an investigation into the effectiveness of the DSA mechanism within the HRNet framework, through a series of experiments, showing this work offers a streamlined and effective solution for improving HPE.

Deep Learning-Based Transmitter Localization in Sparse Wireless Sensor Networks.

In the field of wireless communication, transmitter localization technology is crucial for achieving accurate source tracking. However, the extant methodologies for localization face numerous challenges in wireless sensor networks (WSNs), particularly due to the constraints posed by the sparse distribution of sensors across large areas. We present DSLoc, a deep learning-based approach for transmitter localization in sparse WSNs. Our method is based on an improved high-resolution network model in neural networks. To address localization in sparse wireless sensor networks, we design efficient feature enhancement modules, and propose to locate transmitter locations in the heatmap using an image centroid-based method. Experiments conducted on WSNs with a 0.01% deployment density demonstrate that, compared to existing deep learning models, our method significantly reduces the transmitter miss rate and improves the localization accuracy by more than double. The results indicate that the proposed method offers more accurate and robust performance in sparse WSN environments.

Soft computing-driven infrared and visible image fusion network for security application service

The high-level visual tasks (HLVT)-specific infrared and visible image fusion networks aim to guide the fusion network to focus on specific feature regions. The resulting fused images enhance the performance of HLVT in adverse conditions. However, cross-layer visual task fusion may diminish the fusion network’s adaptability to diverse data. This deficiency results in a lack of robustness and security in the fusion network. Consequently, the fusion network is rendered insufficient to fulfill the requirements of HLVT security applications, including object detection and security systems. To address these challenges, we propose a soft computing-driven infrared and visible image fusion network based on layer separation and dual-stream adversarial learning, referred to as LsDSGFuse. Firstly, we employ a layer segmentation network based on a high-resolution network to create salient object segmentation masks for the original image. This helps mitigate the impact of task-specific guidance on the fusion network learning process, preventing overfitting to HLVT. Secondly, we use the original image after information separation as the input of the generator. The separated network inputs clearly define foreground objects and background environments, aiding the generator in better fusing foreground and background. Lastly, we adopt a no-loading training strategy to balance adversarial learning, prevent discriminator overfitting, and reduce resource waste. Experimental results demonstrate the outstanding performance of our proposed image fusion method in terms of fusion quality, surpassing several representative approaches. Furthermore, experiments in HLVT indicate that our method facilitates HLVT in better responding to potential threats and unknown variables. This improvement significantly enhances HLVT’s security performance.

Accurate UAV Small Object Detection Based on HRFPN and EfficentVMamba.

(1) Background: Small objects in Unmanned Aerial Vehicle (UAV) images are often scattered throughout various regions of the image, such as the corners, and may be blocked by larger objects, as well as susceptible to image noise. Moreover, due to their small size, these objects occupy a limited area in the image, resulting in a scarcity of effective features for detection. (2) Methods: To address the detection of small objects in UAV imagery, we introduce a novel algorithm called High-Resolution Feature Pyramid Network Mamba-Based YOLO (HRMamba-YOLO). This algorithm leverages the strengths of a High-Resolution Network (HRNet), EfficientVMamba, and YOLOv8, integrating a Double Spatial Pyramid Pooling (Double SPP) module, an Efficient Mamba Module (EMM), and a Fusion Mamba Module (FMM) to enhance feature extraction and capture contextual information. Additionally, a new Multi-Scale Feature Fusion Network, High-Resolution Feature Pyramid Network (HRFPN), and FMM improved feature interactions and enhanced the performance of small object detection. (3) Results: For the VisDroneDET dataset, the proposed algorithm achieved a 4.4% higher Mean Average Precision (mAP) compared to YOLOv8-m. The experimental results showed that HRMamba achieved a mAP of 37.1%, surpassing YOLOv8-m by 3.8% (Dota1.5 dataset). For the UCAS_AOD dataset and the DIOR dataset, our model had a mAP 1.5% and 0.3% higher than the YOLOv8-m model, respectively. To be fair, all the models were trained without a pre-trained model. (4) Conclusions: This study not only highlights the exceptional performance and efficiency of HRMamba-YOLO in small object detection tasks but also provides innovative solutions and valuable insights for future research.

Comparative analysis of hyperspectral Image reconstruction using deep learning for agricultural and biological applications

Hyperspectral imaging (HSI) has become a key technology for non-invasive quality evaluation in various fields, offering detailed insights through spatial and spectral data. Despite its efficacy, the complexity and high cost of HSI systems have hindered their widespread adoption. This study addressed these challenges by exploring deep learning-based hyperspectral image reconstruction from RGB (Red, Green, Blue) images, particularly for agricultural products. Specifically, different hyperspectral reconstruction algorithms, such as Hyperspectral Convolutional Neural Network - Dense (HSCNN-D), High-Resolution Network (HRNET), and Multi-Scale Transformer Plus Plus (MST++), were compared to assess the dry matter content of sweet potatoes. Among the tested reconstruction methods, HRNET demonstrated superior performance, achieving the lowest mean relative absolute error (MRAE) of 0.07, root mean square error (RMSE) of 0.03, and the highest peak signal-to-noise ratio (PSNR) of 32.28 decibels (dB). Some key features were selected using the genetic algorithm (GA), and their importance was interpreted using explainable artificial intelligence (XAI). Partial least squares regression (PLSR) models were developed using the RGB, reconstructed, and ground truth (GT) data. The visual and spectra quality of these reconstructed methods was compared with GT data, and predicted maps were generated. The results revealed the prospect of deep learning-based hyperspectral image reconstruction as a cost-effective and efficient quality assessment tool for agricultural and biological applications.

Study on a Landslide Segmentation Algorithm Based on Improved High-Resolution Networks

Landslides are a kind of geological hazard with great destructive potential. When a landslide event occurs, a reliable landslide segmentation method is important for assessing the extent of the disaster and preventing secondary disasters. Although deep learning methods have been applied to improve the efficiency of landslide segmentation, there are still some problems that need to be solved, such as the poor segmentation due to the similarity between old landslide areas and the background features and missed detections of small-scale landslides. To tackle these challenges, a proposed high-resolution semantic segmentation algorithm for landslide scenes enhances the accuracy of landslide segmentation and addresses the challenge of missed detections in small-scale landslides. The network is based on the high-resolution network (HR-Net), which effectively integrates the efficient channel attention mechanism (efficient channel attention, ECA) into the network to enhance the representation quality of the feature maps. Moreover, the primary backbone of the high-resolution network is further enhanced to extract more profound semantic information. To improve the network’s ability to perceive small-scale landslides, atrous spatial pyramid pooling (ASPP) with ECA modules is introduced. Furthermore, to address the issues arising from inadequate training and reduced accuracy due to the unequal distribution of positive and negative samples, the network employs a combined loss function. This combined loss function effectively supervises the training of the network. Finally, the paper enhances the Loess Plateau landslide dataset using a fractional-order-based image enhancement approach and conducts experimental comparisons on this enriched dataset to evaluate the enhanced network’s performance. The experimental findings show that the proposed methodology achieves higher accuracy in segmentation performance compared to other networks.

Using high-density SNP data to unravel the origin of the Franches-Montagnes horse breed

BackgroundThe Franches-Montagnes (FM) is the last native horse breed of Switzerland, established at the end of the 19th century by cross-breeding local mares with Anglo-Norman stallions. We collected high-density SNP genotype data (Axiom™ 670 K Equine genotyping array) from 522 FM horses, including 44 old-type horses (OF), 514 European Warmblood horses (WB) from Sweden and Switzerland (including a stallion used for cross-breeding in 1990), 136 purebred Arabians (AR), 32 Shagya Arabians (SA), and 64 Thoroughbred (TB) horses, as introgressed WB stallions showed TB origin in their pedigrees. The aim of the study was to ascertain fine-scale population structures of the FM breed, including estimation of individual admixture levels and genomic inbreeding (FROH) by means of Runs of Homozygosity.ResultsTo assess fine-scale population structures within the FM breed, we applied a three-step approach, which combined admixture, genetic contribution, and FROH of individuals into a high-resolution network visualization. Based on this approach, we were able to demonstrate that population substructures, as detected by model-based clustering, can be either associated with a different genetic origin or with the progeny of most influential sires. Within the FM breed, admixed horses explained most of the genetic variance of the current breeding population, while OF horses only accounted for a small proportion of the variance. Furthermore, we illustrated that FM horses showed high TB admixture levels and we identified inconsistencies in the origin of FM horses descending from the Arabian stallion Doktryner. With the exception of WB, FM horses were less inbred compared to the other breeds. However, the relatively few but long ROH segments suggested diversity loss in both FM subpopulations. Genes located in FM- and OF-specific ROH islands had known functions involved in conformation and behaviour, two traits that are highly valued by breeders.ConclusionsThe FM remains the last native Swiss breed, clearly distinguishable from other historically introgressed breeds, but it suffered bottlenecks due to intensive selection of stallions, restrictive mating choices based on arbitrary definitions of pure breeding, and selection of rare coat colours. To preserve the genetic diversity of FM horses, future conservation managements strategies should involve a well-balanced selection of stallions (e.g., by integrating OF stallions in the FM breeding population) and avoid selection for rare coat colours.

Automatic soft-tissue analysis on orthodontic frontal and lateral facial photographs based on deep learning.

To establish the automatic soft-tissue analysis model based on deep learning that performs landmark detection and measurement calculations on orthodontic facial photographs to achieve a more comprehensive quantitative evaluation of soft tissues. A total of 578 frontal photographs and 450 lateral photographs of orthodontic patients were collected to construct datasets. All images were manually annotated by two orthodontists with 43 frontal-image landmarks and 17 lateral-image landmarks. Automatic landmark detection models were established, which consisted of a high-resolution network, a feature fusion module based on depthwise separable convolution, and a prediction model based on pixel shuffle. Ten measurements for frontal images and eight measurements for lateral images were defined. Test sets were used to evaluate the model performance, respectively. The mean radial error of landmarks and measurement error were calculated and statistically analysed to evaluate their reliability. The mean radial error was 14.44 ± 17.20 pixels for the landmarks in the frontal images and 13.48 ± 17.12 pixels for the landmarks in the lateral images. There was no statistically significant difference between the model prediction and manual annotation measurements except for the mid facial-lower facial height index. A total of 14 measurements had a high consistency. Based on deep learning, we established automatic soft-tissue analysis models for orthodontic facial photographs that can automatically detect 43 frontal-image landmarks and 17 lateral-image landmarks while performing comprehensive soft-tissue measurements. The models can assist orthodontists in efficient and accurate quantitative soft-tissue evaluation for clinical application.

HR-YOLO: A Multi-Branch Network Model for Helmet Detection Combined with High-Resolution Network and YOLOv5

Automatic detection of safety helmet wearing is significant in ensuring safe production. However, the accuracy of safety helmet detection can be challenged by various factors, such as complex environments, poor lighting conditions and small-sized targets. This paper presents a novel and efficient deep learning framework named High-Resolution You Only Look Once (HR-YOLO) for safety helmet wearing detection. The proposed framework synthesizes safety helmet wearing information from the features of helmet objects and human pose. HR-YOLO can use features from two branches to make the bounding box of suppression predictions more accurate for small targets. Then, to further improve the iterative efficiency and accuracy of the model, we design an optimized residual network structure by using Optimized Powered Stochastic Gradient Descent (OP-SGD). Moreover, a Laplace-Aware Attention Model (LAAM) is designed to make the YOLOv5 decoder pay more attention to the feature information from human pose and suppress interference from irrelevant features, which enhances network representation. Finally, non-maximum suppression voting (PA-NMS voting) is proposed to improve detection accuracy for occluded targets, using pose information to constrain the confidence of bounding boxes and select optimal bounding boxes through a modified voting process. Experimental results demonstrate that the presented safety helmet detection network outperforms other approaches and has practical value in application scenarios. Compared with the other algorithms, the proposed algorithm improves the precision, recall and mAP by 7.27%, 5.46% and 7.3%, on average, respectively.

Deep learning-based automatic measurement system for patellar height: a multicenter retrospective study

BackgroundThe patellar height index is important; however, the measurement procedures are time-consuming and prone to significant variability among and within observers. We developed a deep learning-based automatic measurement system for the patellar height and evaluated its performance and generalization ability to accurately measure the patellar height index.MethodsWe developed a dataset containing 3,923 lateral knee X-ray images. Notably, all X-ray images were from three tertiary level A hospitals, and 2,341 cases were included in the analysis after screening. By manually labeling key points, the model was trained using the residual network (ResNet) and high-resolution network (HRNet) for human pose estimation architectures to measure the patellar height index. Various data enhancement techniques were used to enhance the robustness of the model. The root mean square error (RMSE), object keypoint similarity (OKS), and percentage of correct keypoint (PCK) metrics were used to evaluate the training results. In addition, we used the intraclass correlation coefficient (ICC) to assess the consistency between manual and automatic measurements.ResultsThe HRNet model performed excellently in keypoint detection tasks by comparing different deep learning models. Furthermore, the pose_hrnet_w48 model was particularly outstanding in the RMSE, OKS, and PCK metrics, and the Insall–Salvati index (ISI) automatically calculated by this model was also highly consistent with the manual measurements (intraclass correlation coefficient [ICC], 0.809–0.885). This evidence demonstrates the accuracy and generalizability of this deep learning system in practical applications.ConclusionWe successfully developed a deep learning-based automatic measurement system for the patellar height. The system demonstrated accuracy comparable to that of experienced radiologists and a strong generalizability across different datasets. It provides an essential tool for assessing and treating knee diseases early and monitoring and rehabilitation after knee surgery. Due to the potential bias in the selection of datasets in this study, different datasets should be examined in the future to optimize the model so that it can be reliably applied in clinical practice.Trial registrationThe study was registered at the Medical Research Registration and Filing Information System (medicalresearch.org.cn) MR-61-23-013065. Date of registration: May 04, 2023 (retrospectively registered).

An intelligent MRI assisted diagnosis and treatment system for osteosarcoma based on super-resolution

Magnetic resonance imaging (MRI) examinations are a routine part of the cancer treatment process. In developing countries, disease diagnosis is often time-consuming and associated with serious prognostic problems. Moreover, MRI is characterized by high noise and low resolution. This creates difficulties in automatic segmentation of the lesion region, leading to a decrease in the segmentation performance of the model. This paper proposes a deep convolutional neural network osteosarcoma image segmentation system based on noise reduction and super-resolution reconstruction, which is the first time to introduce super-resolution methods in the task of osteosarcoma MRI image segmentation, effectively improving the Model generalization performance. We first refined the initial osteosarcoma dataset using a Differential Activation Filter, separating those image data that had little effect on model training. At the same time, we carry out rough initial denoising of the image. Then, an improved information multi-distillation network based on adaptive cropping is proposed to reconstruct the original image and improve the resolution of the image. Finally, a high-resolution network is used to segment the image, and the segmentation boundary is optimized to provide a reference for doctors. Experimental results show that this algorithm has a stronger segmentation effect and anti-noise ability than existing methods. Code: https://github.com/GFF1228/NSRDN.

Real-time segmentation of short videos under VR technology in dynamic scenes

Abstract This work addresses the challenges of scene segmentation and low segmentation accuracy in short videos by employing virtual reality (VR) technology alongside a 3D DenseNet model for real-time segmentation in dynamic scenes. First, this work extracted short videos by frame and removed redundant background information. Then, the volume rendering algorithm in VR technology was used to reconstruct short videos in dynamic scenes in 3D. It enriched the detailed information of short videos, and finally used the 3D DenseNet model for real-time segmentation of short videos in dynamic scenes, improving the accuracy of segmentation. The experiment compared the performance of High resolution network, Mask region based convolutional neural network, 3D U-Net, Efficient neural network models on the Densely annotation video segmentation dataset. The experimental results showed that the segmentation accuracy of the 3D DenseNet model has reached 99.03%, which was 15.11% higher than that of the ENet model. The precision rate reached 98.33%, and the average segmentation time reached 0.64 s, improving the segmentation accuracy and precision rate. It can adapt to various scene situations and has strong robustness. The significance of this research lies in its innovative approach in tackling these issues. By integrating VR technology with advanced deep learning models, we can achieve more precise segmentation of dynamic scenes in short videos, enabling real-time processing. This has significant practical implications for fields such as video editing, VR applications, and intelligent surveillance. Furthermore, the outcomes of this research contribute to advancing computer vision in video processing, providing valuable insights for the development of future intelligent video processing systems.

Building pose detection for the characterization of reinforced concrete buildings

SummaryThe automated identification of building characteristics for seismic vulnerability remains a challenge for governments due to the high number of buildings in cities. The diverse architectural styles of these buildings complicate the automated identification of building information (e.g., number of stories, structural system, and material type). Deep learning techniques lose accuracy as they generalize information, while the visual contents of a building exhibit a considerable range and diversity. This study leverages the pose detection technique to tackle such issues by focusing on a common construction style: reinforced concrete buildings representing columns, beams, or floors on the façade. With an aim to enable the assessment of seismic vulnerability, the technique developed herein is conceived for buildings with up to six stories that are more likely to be moment‐frame buildings. The AI‐enabled proposed framework starts with collecting building images and categorizing those containing this specific building type. A bounding box detector is then used to isolate building facades, for the subsequent identification of the structural frame with the High‐Resolution Network (HR‐Net). For demonstration, we illustrate this technique by identifying the structural frame on concrete buildings with a sample dataset developed based on buildings found in Mexico City in a pre‐earthquake event state.

SEHRNet: A lightweight, high‐resolution network for aircraft keypoint detection

AbstractCurrent research on apron conflict detection is often limited to the interaction between the aircraft as a whole and other objects, making it difficult to accomplish targeted identification of vulnerable and high‐cost aircraft components. However, the implementation of detailed aircraft identification is of great significance to enhance the safety of airport surface operations. Based on the excellent performance of High‐Resolution Network (HRNet) in keypoint detection, a lightweight end‐to‐end keypoint detection network, namely Squeeze and Excitation High‐Resolution Network (SEHRNet), is proposed in this paper to solve the problems of HRNet's slower computation and more redundancy. First, the errors arising from coordinate transformations in the coding and decoding process are solved by an improved feature map coding and decoding process. Second, replace the BasicBlock in HRNet with the Depthwise separable convolutions based on the Squeeze‐and‐Excitation Networks, which drastically cuts the computational cost of the network. Third, the improved Bottleneck module is used to further enhance the capability of feature extraction. Experimental results prove that, based on the aircraft keypoint detection dataset, the SEHRNet proposed in this paper shows stronger applicability compared to the current mainstream networks. Compared with the original HRNet, the improved network has higher accuracy, faster speed, and lighter model for aircraft keypoint detection.

Automated 3D Perioral Landmark Detection Using High-Resolution Network: Artificial Intelligence-based Anthropometric Analysis.

Three-dimensional facial stereophotogrammetry, a convenient, noninvasive and highly reliable evaluation tool, has in recent years shown great potential in plastic surgery for preoperative planning and evaluating treatment efficacy. However, it requires manual identification of facial landmarks by trained evaluators to obtain anthropometric data, which takes much time and effort. Automatic 3D facial landmark localization has the potential to facilitate fast data acquisition and eliminate evaluator error. The aim of this work was to describe a novel deep-learning method based on dimension transformation and key-point detection for automated 3D perioral landmark annotation. After transforming a 3D facial model into 2D images, High-Resolution Network is implemented for key-point detection. The 2D coordinates of key points are then mapped back to the 3D model using mathematical methods to obtain the 3D landmark coordinates. This program was trained with 120 facial models and validated in 50 facial models. Our approach achieved a satisfactory mean [standard deviation] accuracy of 1.30 [0.68] mm error in landmark detection with a mean processing time of 5.2 [0.21] seconds per model. Subsequent analysis based on these landmarks showed mean errors of 0.87 [1.02] mm for linear measurements and 5.62° [6.61°] for angular measurements. This automated 3D perioral landmarking method could serve as an effective tool that enables fast and accurate anthropometric analysis of lip morphology for plastic surgery and aesthetic procedures.

MS-HRNet: multi-scale high-resolution network for human pose estimation

MS-HRNet: multi-scale high-resolution network for human pose estimation

High-Resolution Multi-Scale Feature Fusion Network for Running Posture Estimation

Running posture estimation is a specialized task in human pose estimation that has received relatively little research attention due to the lack of appropriate datasets. To address this issue, this paper presents the construction of a new benchmark dataset called “Running Human”, which was specifically designed for running sports. This dataset contains over 1000 images along with comprehensive annotations for 1288 instances of running humans, including bounding boxes and keypoint annotations on the human body. Additionally, a Receptive Field Spatial Pooling (RFSP) module was developed to tackle the challenge of joint occlusion, which is common in running sports images. This module was incorporated into the High-Resolution Network (HRNet) model, resulting in a novel network model named the Running Human Posture Network (RHPNet). By expanding the receptive field and effectively utilizing multi-scale features extracted from the multi-branch network, the RHPNet model significantly enhances the accuracy of running posture estimation. On the Running Human dataset, the proposed method achieved state-of-the-art performance. Furthermore, experiments were conducted on two benchmark datasets. Compared to the state-of-the-art ViTPose-L method, when applied to the COCO dataset, RHPNet demonstrated comparable prediction accuracy while utilizing only one tenth of the parameters and one eighth of the floating-point operations (FLOPs). On the MPII dataset, RHPNet achieves a PCKh@0.5 score of 92.0, which is only 0.5 points lower than the state-of-the-art method, PCT. These experimental results provide strong validation for the effectiveness and excellent generalization ability of the proposed method.