Relative Positional Relationship Research Articles

Clustering-based mask recovery for image captioning

Region features play a huge role in image captioning. However, obtaining region features requires pre-training an object detector by using a large number of object detection datasets. It may be impossible for the end-to-end training. And if there is a large distribution difference between the object detection datasets and the image captioning datasets, the object detector may not be able to extract accurate region features. This makes it limited in application. In this paper, we propose a clustering-based mask recovery for image captioning. In the encoder, the pseudo-region features are obtained by clustering the grid features, which are extracted using Swin Transformer. Then we input the grid features together with the pseudo-region features into the decoder, and make the model to dynamically learns the weights of the two features in the decoding process to minimize the effect of errors caused by clustering. By using a clustering method to generate pseudo-region features for images, not only does the training process become end-to-end, but there is no need to introduce additional object detection datasets to train the object detector. In addition, the Transformer decoder has a misplaced problem in the decoding process. This means that the positional information used by the model when generating a word is not the same as the positional information used when it continues to use the word to reason. This may have some negative impact on the position encoding of the model. Therefore, we changed the original decoding method to mask recovery. Furthermore, a masked multi-head attention module with relative position is proposed in the decoder to integrate the information in the fusion features, and reconstruct the relative position relationship between words. We conduct experiments on MSCOCO 2014 dataset. The experiment results show that our model obtains 144.3% (single model) and 147.0% (ensemble of 4 models) CIDEr scores on ‘Karpathy’ offline test split, and 143.2% (c40) CIDEr scores on the official online test server.

Landslide hazard assessment of the fault zone considering the fault effect: a case study of the Lixian–Luojiabu fault zone in Gansu Province (China)

The earthquake landslide hazard assessment method is mainly based on the traditional Newmark model. However, when the landslide hazard assessment is carried out along the fault zone, the calculated results are often different from the actual situation because the influence of fault effect is not fully considered. Therefore, how to construct a landslide hazard assessment model suitable for the fault zone is a technical problem to be solved by researchers. Taking the Lixian–Luojiabu fault zone in Gansu Province in China as the study area, this paper put forward the concept of fault effect correction coefficient exploringly, systematically studied the relative distance relationship between the landslide and fault zone, and the relative position relationship between landslide and upper and lower sides of the fault zone. The value table of the fault effect correction coefficient along the Lixian–Luojiabu fault zone was established, and the corresponding distribution map of the fault effect correction coefficient was drawn. Based on this, an improved Newmark model for the landslide hazard assessment along the fault zone was constructed. On the basis of systematic analysis of the slope and engineering geological rock group in the study area, the traditional Newmark model and improved Newmark model considering fault effects were used, respectively, to carry out the earthquake landslide hazard assessment under the condition of 10% exceeding probability in 50 years, and the ROC curve and Kappa coefficient methods were used to compare and analyze the evaluation results. The results showed that the AUC value and Kappa coefficient of the danger area obtained by the improved model with the Newmark model were 0.841 and 0.822, respectively, which were significantly higher than the calculated values of the traditional Newmark model, indicating that the model had a good improvement effect. The Newmark improved model, considering the fault effect, fully considered the influence of distance from the fault zone and fault upper and lower side effects, and the research results can provide a new reference for the landslide hazard assessment along the fault zone.

AUV Obstacle Avoidance Framework Based on Event-Triggered Reinforcement Learning

Autonomous Underwater Vehicles (AUVs), as a member of the unmanned intelligent ocean vehicle group, can replace human beings to complete dangerous tasks in the ocean. It is of great significance to apply reinforcement learning (RL) to AUVs to realize intelligent control. This paper proposes an AUV obstacle avoidance framework based on event-triggered reinforcement learning. Firstly, an environment perception model is designed to judge the relative position relationship between the AUV and all unknown obstacles and known targets. Secondly, considering that the detection range of AUVs is limited, and the proposed method needs to deal with unknown static obstacles and unknown dynamic obstacles at the same time, two different event-triggered mechanisms are designed. Soft actor–critic (SAC) with a non-policy sampling method is used. Then, improved reinforcement learning and the event-triggered mechanism are combined in this paper. Finally, a simulation experiment of the obstacle avoidance task is carried out on the Gazebo simulation platform. Results show that the proposed method can obtain higher rewards and complete tasks successfully. At the same time, the trajectory and the distance between each obstacle confirm that the AUV can reach the target well while maintaining a safe distance from static and dynamic obstacles.

PSFHS: Intrapartum ultrasound image dataset for AI-based segmentation of pubic symphysis and fetal head

During the process of labor, the intrapartum transperineal ultrasound examination serves as a valuable tool, allowing direct observation of the relative positional relationship between the pubic symphysis and fetal head (PSFH). Accurate assessment of fetal head descent and the prediction of the most suitable mode of delivery heavily rely on this relationship. However, achieving an objective and quantitative interpretation of the ultrasound images necessitates precise PSFH segmentation (PSFHS), a task that is both time-consuming and demanding. Integrating the potential of artificial intelligence (AI) in the field of medical ultrasound image segmentation, the development and evaluation of AI-based models rely significantly on access to comprehensive and meticulously annotated datasets. Unfortunately, publicly accessible datasets tailored for PSFHS are notably scarce. Bridging this critical gap, we introduce a PSFHS dataset comprising 1358 images, meticulously annotated at the pixel level. The annotation process adhered to standardized protocols and involved collaboration among medical experts. Remarkably, this dataset stands as the most expansive and comprehensive resource for PSFHS to date.

Improved textile-vamp punching points location using CNN and priori regional template matching in the footwear industry

The accurate and efficient positioning of punching points on textile-vamps is critical for successful footwear manufacturing. Manual methods often lead to errors, increased labor costs, and reduced efficiency. To address these challenges, this paper takes a full account of punching point layout regularity in the vamp and proposes an approach for locating punching points using priori regional template matching and convolutional neural network. Least-squares ellipse fitting is used for coordinate system establishment based on a fixed pattern of 3D additive on the vamp. This facilitates the description of the placement and orientation of the vamp. Then, the prior region and template images are extracted and recorded in accordance with the relative positional relationship between punching points and the vamp. This can limit the template search to designated regions and improve the speed and accuracy of punching point location. Leveraging the determined coordinates system, the extracted prior regions universally applicable to vamps in any placement and orientation. To automatize the adjustment of punching point matching strategies, a textile-vamp classification CNN model is trained to distinguish between different vamp types. Experiments demonstrate the effectiveness of the proposed method on various vamp types and placement, achieving a high accuracy rate of 98%. Therefore, this research offers a promising solution for practical applications in the footwear industry, providing accurate and rapid textile-vamp punching point location.

Co-Operatively Increasing Smoothing and Mapping Based on Switching Function

Collaborative localization is a technique that utilizes the exchange of information between multiple sensors or devices to improve localization accuracy and robustness. It has a wide range of applications in autonomous driving and unmanned aerial vehicles (UAVs). In the field of UAVs, collaborative localization can help UAVs perform autonomous navigation and mission execution in complex environments. However, when GNSS is not available, it becomes challenging to position the UAV swarm relative to each other. This is because the swarm loses its perception and constraint of the position relationship between each member. Consequently, the swarm faces the problem of the serious drift of relative accuracy for an extended period. Furthermore, when the environment is faced with complex obstruction challenges or a camera with low texture scenes, noise can make it more difficult to solve the relative position relationship between drones, and a single UAV may lose positioning capability. To solve these specific problems, this paper studies a swarm co-operative localization method in a GNSS-denied environment with loud noise interference. In this paper, we proposed a method that utilizes a distributed scheme based on an incremental smoothing and mapping (iSAM) algorithm for state estimation. It incorporates new anchor-free topological constraints to prevent positioning failures and significantly improve the system’s robustness. Additionally, a new switching function is applied in front of each factor of the loss function, which adjusts the switches in real time in response to the input information to improve observably the accuracy of the system. A novel co-operative incremental smoothing and mapping (CI-SAM) method is proposed and the method does not require a complete relative position measurement, which reduces the need for vehicle measurement equipment configuration. The effectiveness of the method is verified by simulation.

Topological optimization design for support structure under multiple loading conditions

In view of the lightweight and high stiffness requirements of a certain type of spacecraft support, combined with the special performance and environment of the support, topology optimization technology is introduced into the optimal design of the support structure in the space field. The topology optimization and design of the support are carried out with the stiffness as the objective function and the quality as the constraint condition, and the design of "space hugging configuration" is proposed, and the simulation verification and experimental verification are carried out. Firstly, different load conditions of the support were analyzed and studied. Secondly, the initial model was established based on the relative position relationship between the support and the mounting surface of the loaded equipment. Finally, a design method of the support structure was established based on the combination of topology optimization under multi-load conditions and size optimization considering the influence of strength. The stiffness and strength of the optimized structure were checked under multiple load conditions, and the fundamental frequency was 2.05 times of the index requirement. The maximum response stress is lower than the yield limit of the material used, which meets the strength requirement. Finally, the vibration test is carried out to verify the support. After the test is completed, the position accuracy of the support interface meets the requirements of the index. The obtained bracket has high structural stability, meets the high performance requirements of the load equipment, verifies the feasibility of the topology optimization method, and provides an effective method for the design of lightweight high-stiffness structures.

Multipath-Closure Calibration of Stereo Camera and 3D LiDAR Combined with Multiple Constraints

Stereo cameras can capture the rich image textures of a scene, while LiDAR can obtain accurate 3D coordinates of point clouds of a scene. They complement each other and can achieve comprehensive and accurate environment perception through data fusion. The primary step in data fusion is to establish the relative positional relationship between the stereo cameras and the 3D LiDAR, known as extrinsic calibration. Existing methods establish the camera–LiDAR relationship by constraints of the correspondence between different planes in the images and point clouds. However, these methods depend on the planes and ignore the multipath-closure constraint among the camera–LiDAR–camera sensors, resulting in poor robustness and accuracy of the extrinsic calibration. This paper proposes a trihedron as the calibration object to effectively establish various coplanar and collinear constraints between stereo cameras and 3D LiDAR. With the various constraints, the multipath-closure constraint between the three sensors is further formulated for the extrinsic calibration. Firstly, the coplanar and collinear constraints between the camera–LiDAR–camera are built using the trihedron calibration object. Then, robust and accurate coplanar constraint information is extracted through iterative maximum a posteriori (MAP) estimation. Finally, a multipath-closure extrinsic calibration method for multi-sensor systems is developed with structurally mutual validation between the cameras and the LiDAR. Extensive experiments are conducted on simulation data with different noise levels and a large amount of real data to validate the accuracy and robustness of the proposed calibration algorithm.

Latent Attention Network With Position Perception for Visual Question Answering.

For exploring the complex relative position relationships among multiobject with multiple position prepositions in the question, we propose a novel latent attention (LA) network for visual question answering (VQA), in which LA with position perception is extracted by a novel LA generation module (LAGM) and encoded along with absolute and relative position relations by our proposed position-aware module (PAM). The LAGM reconstructs original attention into LA by capturing the tendency of visual attention shifting according to the position prepositions in the question. The LA accurately captures the complex relative position features of multiple objects and helps the model locate the attention to the correct object or region. The PAM adopts latent state and relative position relations to enhance the capability of comprehending the multiobject correlations. In addition, we also propose a novel gated counting module (GCM) to strengthen the sensitivity of quantitative knowledge for effectively improving the performance of counting questions. Extensive experiments demonstrate that our proposed method achieves excellent performance on VQA and outperforms state-of-the-art methods on the widely used datasets VQA v2 and VQA v1.

Early Postoperative Change in Hip Rotation Angle and Factors Contributing to It for Patients Undergoing Total Hip Arthroplasty.

A possible impairment in hip proprioception after total hip arthroplasty (THA) has been an issue of concern. The aims of this study were to investigate the extent of early postoperative change in standing hip rotation angle (HRAng) in patients with osteoarthritis (OA) undergoing THA and to consider a possible mechanism behind this. A total of 82 hips (82 patients; 63 women and 19 men) undergoing unilateral primary THA with total capsulectomy were included. We characterized the standing HRAng and internal/external range of motion (ROM) in the prone position before THA and 2 weeks after THA. Acetabular/cup and femoral/stem anteversion, combined anteversion (CA), and leg length discrepancy were also characterized. Correlations were examined postoperatively between the HRAng and the other analyzed variables. The median standing HRAng showed a significant internal shift from external to more medial position (6.3° to 1.7°) 2 weeks after THA (P<.0001). The postoperative change in standing HRAng was significantly negatively correlated with the difference between the postoperative femoral anteversion and the stem anteversion (rs=-0.429, P<.0001) and with the pre- to postoperative change in CA (rs=-0.3012, P=.0063). This study demonstrated that the extent of the rotational shift of the distal femur toward medial direction was significantly associated with increasing stem anteversion and CA. This phenomenon can be interpreted as a compensatory mechanism for maintaining the relative positional relationship between the pelvis and the proximal femur using proprioception. Therefore, we conclude that the extracapsular/extra-articular components may be prominent determinants of joint position sense. [Orthopedics. 2024;47(3):e114-e118.].

Endpoint Detection Based on Optical Method in Chemical Mechanical Polishing.

Endpoint detection is an important technology in chemical mechanical polishing (CMP), which is used to capture the material interface and compensate the variations of consumables and incoming wafer thickness. This paper aimed to apply optical detection in metal CMP. An in situ optical measurement system was developed for a 12-inch CMP tool. Kinematic analysis of the scanning trajectory of the laser device indicated the relative position relationship between the device and the wafer. Average optical data within the wafer described the material removal of metal CMP. Furthermore, optical data and location described the non-uniformity of the entire wafer surface. In this research, the polishing condition and the residual of the wafer edge are characterized by optical trace. Pauta Criterion is used to discriminate the inflexion point of the material interface. The results reveal that the interface capture is accurate and effective.

Research on cone-beam CT of mandibular foramen location in children aged 7-10 years

Objective: Cone-beam CT (CBCT) images were used to investigate the relative position changes of mandibular foramen in the mandible of children and the relative position relationship with the occlusal plane, so as to provide clinical guidance for inferior alveolar nerve block (IANB) anesthesia of children. Methods: The CBCT data of 202 children aged 7-10 years in the image database of the First Affiliated Hospital of Zhengzhou University from March 2021 to February 2023 were included. Patients were divided into 4 groups according to age diffrences as 7-year-old, 8-year-old, 9-year-old and 10-year-old. There were 20 males and 22 females in the 7-year-old group, 31 males and 28 females in the 8-year-old group, 30 males and 26 females in the 9-year-old group, and 22 males and 23 females in the 10-year-old group, respectively. Forty-six adults aged 25-30 years were selected as control group, 24 males and 22 females included. The distance between the center point of mandibular foramen with the anterior edge of ascending ramus of mandible (MF-A), the posterior edge of the ascending ramus of mandible (MF-P) and the shortest distance between the center point of mandibular foramen with occlusal plane (MF-OP) were measured. The angle between the center point of the mandibular foramen with the sagittal plane of the mandibular first deciduous molar (or mandibular first premolar) and mandibular second deciduous molar (or mandibular second premolar) (∠A) was measured. The data of mandibular foramen were compared between the left and right sides and among different genders and different age groups. Results: The position of mandibular foramen in children aged 7-10 years maintained bilateral symmetry, and mandibular growth and development were relatively consistent between different genders (P>0.05). MF-A increased with age, from (15.83±1.28) mm in 7-year-old group to (17.10±1.60) mm in 10-year-old group gradually. There were significant differences in MF-A between the 10-year-old group with the 7-year-old group, the 8-year-old group [(15.98±1.53) mm] and the 9-year-old group [(16.43±1.49) mm] respectively (P<0.05). MF-P increased with age, from (9.12±1.17) mm in 7-year-old group to (11.25±1.60) mm in 10-year-old group. There were statistically significant differences in MF-P among all age groups (P<0.05). MF-OP increased with age, from below the plane (-0.24±2.31) mm in the 7-year-old group to above the plane (1.08±1.95) mm in the 10-year-old group. There were significant differences between the 10-year-old group with the 7-year-old group, the 8-year-old group [(-0.01±1.93) mm], and the 9-year-old group [(0.31±1.95) mm] (P<0.05). The ratio of MF-A to MF-P decreased as the age increased, from 1.77±0.30 in the 7-year-old group to 1.55±0.29 in the 10-year-old group. There were statistically significant differences in MF-A/MF-P among all age groups (P<0.05), except for between the 8-year-old group (1.66±0.19) and the 9-year-old group (1.65±0.28) (P>0.05). The ∠A of children in all age groups was significantly greater than the reference value (45°) (P<0.05), and there was no statistical significance among all groups (P>0.05). The differences of MF-A, MF-P, MF-OP, MF-A/MF-P and ∠A between children of all age groups and the control group were statistically significant (P<0.05). Conclusions: In children aged 7-10 years, the mandibular foramen is located behind the midpoint of the anteroposterior diameter of the mandibular ramus. With the increase of age, the mandibular foramen gradually moves from below the occlusal plane to above, and is flush with the occlusal plane at the age of 8 years. Compared with adults, the mandibular foramen in children is more backward and lower on the medial side of the mandibular ramus. When IANB is operated to children, the syringe can be moved distally from the contact area of the contralateral deciduous molars or premolars, so that the injection angle can be greater than the reference value 45° to improve the accuracy of IANB.

Camera calibration using the dual double-contact property of circles

To mitigate the influence of the relative positional relationship between circles on camera calibration, this study explores the double contact of dual circles with dual circular points. We demonstrate that one of the generalized eigenvectors of two dual circles corresponds to the line passing through their respective centers. Moreover, we establish a methodology to identify this line from the three generalized eigenvectors and show that, regardless of the positional arrangement of the three co-planar circles, the image of the circular points can be constructed by calculating the image of this line. Consequently, the applicability of circles as calibration templates is expanded, enabling the development of a novel optimization technique for fitting circular images with enhanced calibration precision.

A Lidar-Assisted Self-Localization Technology for Indoor Wireless Sensor Networks

The self-localization of wireless sensor networks (WSNs) is facing the problem of insufficient positioning accuracy in indoor environment due to multipath and interference issues. At the same time, without external references, through mutual measurement nodes can calculate only a set of relative coordinates. Therefore, it is difficult to achieve the mapping of coordinate values to the physical world. The development of simultaneous localization and mapping (SLAM) technologies has provided new opportunities to solve the above problems by making it easier to obtain real-time indoor maps. This paper proposes a Lidar-assisted self-localization (LASL) technique to further improve the localization accuracy of WSNs in indoor scenes by combining spatial constraint information obtained from real-time maps, and to place the relative node coordinate network in the visualized maps. Based on the general assumption that the nodes are deployed on the surface of the object, the proposed technique combines the spatial constraints obtained by plane fitting of a local point cloud map (PCM) and finite area approximation of object surfaces with the distance constraints provided by radio ranging. Subsequently, the self-localization results under the joint constraints are solved by the alternating coordinate descent (ACD) method. Simulations and experiments demonstrate that the proposed technique can effectively combine the spatial constraints provided by the Lidar PCM to further improve the self-localization accuracy of the sensor nodes, and further optimize the relative position relationship between each node and the map environment to achieve better matching and integration of the WSN and the real-time map.

Feasibility analysis of implant movement along arc trajectory under non-contact control in magnetic stereotaxic system

In this paper, the non-contact control of magnetic implants by changing the external magnetic field in the magnetic stereotaxic system is introduced, and the feasibility of making them move along the arc trajectory is analyzed. Through COMSOL software, the process of moving the miniature magnetic implant along the arc trajectory was simulated, the change of the micro-magnetic implant trajectory after the external magnetic field was changed, the relative position relationship between the large permanent magnets was determined, and the mechanical analysis of the miniature magnetic implant moving along the arc trajectory was carried out. In this experiment, we fix a large permanent magnet, only move the second permanent magnet, first, observe the process of small permanent magnets moving along a straight trajectory, determine the position of the large permanent magnet magnetic field when it contacts the small permanent magnet, and then, analyze the force of the small permanent magnet through the force calculation module, and determine the relative position relationship between the two large permanent magnets by comparing and , and when the small permanent magnet will start to move along the arc trajectory. Then, according to the previous data, we move two adjacent large permanent magnets at the same time at a certain interval, record the movement trajectory of the small magnet, Finally, with the force calculation module of the COMSOL software, force analysis of small permanent magnets moving along arc trajectories. The data from this experiment will be used to determine the relative position relationship between two large permanent magnets adjacent to each other during the actual experiment, and under what conditions the small permanent magnets will move along the arc trajectory. The purpose of this experiment is to provide theoretical and data support for the subsequent practical experiments of the magnetic stereotactic system, and all parameters in the COMSOL software are derived from the actual measurement data, so as to improve the reliability of the simulation results.

367. SPATIAL ORGANIZATION OF TUMOR-INFILTRATING LYMPHOCYTES PREDICTS PROGNOSIS OF PATIENTS WITH ESOPHAGEAL CANCER RECEIVING IMMUNOTHERAPY—A DEEP-LEARNING APPROACH

Abstract Background Neoadjuvant immunotherapy plus chemotherapy has shown satisfying efficacy and manageable safety in locally advanced esophageal squamous cell carcinoma (ESCC) patients. Tumor-infiltrating lymphocytes (TILs) have been shown to be associated with the prognosis of various cancer types. It is unknown whether these TILs-associated characteristics can be adopted as a reliable prognostic biomarker for immunotherapy-based regimen. The current study aimed to assess the prognostic power of spatial organization of TILs in patients with advanced ESCC treated with immunochemotherapy. Methods In this retrospective, multicenter cohort study, pathologically confirmed locally advanced ESCC that was potentially resectable after neoadjuvant immunochemotherapy were included. Image segmentation was performed using UNet++ algorithm and H&amp;E-stained nuclei were segmented using the StarDist deep-learning image segmentation. For the nuclei classification task, we choose to train and validate the classic convolutional neural network (CNN) Resnet50. We calculated tumor density and TiLs density in the tumor region (T_TiLs), TILs in the peritumoral region (P_TiLs) as well as TiLs density in the stromal region (S_TiLs). Based on the relative position relationship of tumor cells and lymphocytes, we subclassified the TILs spatial patterns into TILs-dominant, TILs-infiltrating, and TILs-circulating. Results Among the enrolled 86 patients, the male to female ratio approached 4:1. In this study, the mean IoU (mIoU) for nuclei segmentation is 0.9369 and the mean Dice coefficient is 0.9209. The areas under curve (AUC) for lymphocyte, tumor cell, and stromal cells classification are 0.92, 0.90 and 0.86, respectively. Through univariate analysis, TILs in tumor region (HR = 1.285, 95% CI: 1.042–1.586, p = 0.019), and TILs spatial organization (HR = 3.652, 95% CI: 1.216–10.967, p = 0.021) were identified as significant prognostic factors for OS. Multivariate analysis further revealed TILs in tumor region as the independent prognostic factor (HR = 1.248, 95% CI: 1.006–1.547, p = 0.044). Discussion The present study indicates that spatial organization and T_TILs are significant prognostic factors for ESCC patients receiving neoadjuvant immunochemotherapy. Future prospective studies with large sample size are needed to further validate the prognostic role of interaction between TILs spatial characteristic and distribution density in predicting overall survival.

A sound frequency ridge positioning model oriented to mine equipment

The positioning of equipment sound frequency ridges is the foundation in monitoring the state of mine equipment. However, existing models are difficult to identify equipment sound frequency ridges from sounds generated by mine equipment under the environment with too complex noise. Facing to the mine equipment, this paper proposed an equipment sound frequency ridges positioning model (namely, HVDG-RSAGE model) which contains the Horizontal Visibility Directed Graph conjoin the Resampling graph SAmpling and aggreGatE. Firstly, the mine equipment sound is captured by the microphone, and then a suggested preprocessing decomposes the sound into frequency components. Secondly, a horizontal visibility directed criterion is proposed to create horizontal visibility directed graph which can describe the relative position relationship of frequency components. Finally, the Resampling graph SAmpling and aggreGatE, an embedding model included a resampling module, is proposed to identify the position of equipment frequency ridges, which can avoid model overfitting. The experiments show that the HVDG-RSAGE model can improve the accuracy positioning of equipment sound frequency ridges.

Cattle Target Segmentation Method in Multi-Scenes Using Improved DeepLabV3+ Method.

Obtaining animal regions and the relative position relationship of animals in the scene is conducive to further studying animal habits, which is of great significance for smart animal farming. However, the complex breeding environment still makes detection difficult. To address the problems of poor target segmentation effects and the weak generalization ability of existing semantic segmentation models in complex scenes, a semantic segmentation model based on an improved DeepLabV3+ network (Imp-DeepLabV3+) was proposed. Firstly, the backbone network of the DeepLabV3+ model was replaced by MobileNetV2 to enhance the feature extraction capability of the model. Then, the layer-by-layer feature fusion method was adopted in the Decoder stage to integrate high-level semantic feature information with low-level high-resolution feature information at multi-scale to achieve more precise up-sampling operation. Finally, the SENet module was further introduced into the network to enhance information interaction after feature fusion and improve the segmentation precision of the model under complex datasets. The experimental results demonstrate that the Imp-DeepLabV3+ model achieved a high pixel accuracy (PA) of 99.4%, a mean pixel accuracy (MPA) of 98.1%, and a mean intersection over union (MIoU) of 96.8%. Compared to the original DeepLabV3+ model, the segmentation performance of the improved model significantly improved. Moreover, the overall segmentation performance of the Imp-DeepLabV3+ model surpassed that of other commonly used semantic segmentation models, such as Fully Convolutional Networks (FCNs), Lite Reduced Atrous Spatial Pyramid Pooling (LR-ASPP), and U-Net. Therefore, this study can be applied to the field of scene segmentation and is conducive to further analyzing individual information and promoting the development of intelligent animal farming.

Heartbeat information prediction based on transformer model using millimetre‐wave radar

AbstractMillimetre‐wave radar offers high ranging accuracy and can capture subtle vibration information of the human heart. This study proposes a heartbeat prediction method based on the transformer model using millimetre‐wave radar. Firstly, the millimetre‐wave radar was used to collect the heartbeat data and conduct normalisation processing. Secondly, a position coding was introduced to assign sine or cosine variables to input data and extract their relative position relationship. Subsequently, the transformer encoder was adopted to allocate attention to input data through the multi‐head attention mechanism, using a mask layer before the decoding layer to prevent the leakage of future information. Finally, we employ the fully connected layer was employed in the linear decoder for regression and output the predicted results. Our experimental results demonstrate that the proposed transformer model achieves nearly 30% higher prediction accuracy than traditional long short‐term memory models while improving both the prediction accuracy and convergence rate. The proposed method has great potential in predicting the heartbeat state of elderly and sick patients.

Strain field evolution and crack coalescence mechanism of composite strength rock-like specimens with sawtooth interface

The mechanical behaviors of composite rock masses containing a complex interface widely distributed in nature are significantly different from those of monolithic rock masses. The objective of this study is to perform a series of uniaxial compression tests to study the influence of sawtooth interface parameters on the fracture behavior of rock-like composite specimens. Meantimes, the acoustic emission (AE) and digital image correlation (DIC) technique were used to classify damage stages and record strain field evolution, respectively. The results of this study demonstrate that stress–strain curves of composite rock-like specimens with different sawtooth interfaces display characteristics of plastic failure, friction failure characteristics and two different brittle failure characteristics in the post-peak phase. The peak strength of the specimen first decreases and then increases as the inclination of the interface increases, achieving the maximum value of 28.29 MPa at the inclined interface of 90°. The strain concentration zone is mainly generated and expanded around the interface. The parameters of the sawtooth have a significant influence on the type and location of cracks developed along the interface. In addition, there is a strong relationship between the mode of failure of the specimen and the strength of the rock layer and the inclination of the interface. According to the relative positional relationship between the coalescence path of the specimen and the interface, the failure modes of the specimen are classified as single weak rock layer failure (S-AB-0, S-AB-15), oblique failure through rock layer interface (S-AB-30, S-AB-45), slip failure along the interface (S-AB-60, S-AB-75) and axisymmetric failure along the interface (S-AB-90). This result is of great importance for a thorough understanding of the crack coalescence mechanism and strain field evolution of rock-like composite specimens.