2606 results about "Pedestrian" patented technology

The invention discloses a long-time pedestrian re-identification method based on dual-path cooperation and key frame guided reconstruction. The method comprises the steps of firstly collecting a pedestrian video to be recognized, and extracting a video feature sequence; space and time position coding is introduced into the video feature sequence; capturing local fine-grained dynamic features through a local dynamic feature capturing path, and modeling long-range time sequence association through a cross-frame global feature modeling path; then, dual-path feature complementation is realized through bidirectional gating interaction; further screening out key frames, and realizing feature reconstruction through a full-frame attention propagation mechanism; and finally fusing the dual-path fusion features, the key frame guide reconstruction features and the refined features to generate pedestrian identity features. And processing pedestrian identity features to obtain standardized feature vectors, performing similarity comparison on the standardized feature vectors and pedestrian features in an image library, and returning a matching list. According to the method, video time sequence information is fully utilized, and the problem of insufficient robustness caused by appearance change in long-time pedestrian re-identification is effectively solved.

The invention discloses a target detection tracking method and device based on radar visual fusion perception, and a medium, and the method employs visual detection as a leading part to establish a radar visual fusion tracking module, and solves a problem that the detection precision, tracking stability and environment robustness are difficult to give consideration to the existing single-mode perception in a complex traffic environment at the same time. Through unified multi-modal fusion of distance and speed information of visual detection, visual tracking and millimeter wave radar, stable, continuous and reliable target identification and track output of targets such as pedestrians and vehicles under a low-computing-power platform are realized.

An operating method of a first device (100) in a wireless communication system is presented. The method may comprise the steps of: determining that a first device (100) interrupts walking of a pedestrian; and determining, on the basis of the determination that the first device (100) interrupts walking of the pedestrian, whether to perform a first operation for preventing the interruption.

The invention discloses an occluded pedestrian re-identification method based on cross-layer frequency domain enhancement and multi-view fusion, and the method comprises the steps: employing ResNet-50 as a backbone network, inputting a plurality of pedestrian images of the same identity according to groups, obtaining multiple layers, carrying out the frequency separation and frequency enhancement, forming a significant mask which is more sensitive to the occlusion and background, and carrying out the recognition of the pedestrian images. And meanwhile, the lay1 obtains global features with the same scale as the lay4 through GCSA, and weighted fusion is carried out on the global features and the lay4 features according to the occlusion score, so that the occluded area is completed. Group-level representation with multi-view attention fusion, output information complementation and noise suppression is adopted. In the training stage, the identity classification loss of each branch and the cross-branch consistency loss are jointly optimized, and gradient cutting is matched to improve the stability. According to the method, on the premise that a ResNet-50 backbone structure is not changed, cross-layer frequency domain prior is used for accurate shielding positioning, global-local adaptive fusion based on shielding scores and multi-view weighting are combined, and the pedestrian re-recognition precision and robustness in a shielding scene are remarkably improved.

The invention belongs to the field of visual tracking, and relates to a multi-target visual tracking method based on cooperation of a dynamic neural field and a neural network, which takes a cross-modal cooperation architecture as a core and comprises a dynamic neural field module based on multi-target trajectory maintenance and shielding matching and an improved MoESDQ neural network module. Meanwhile, a collaborative decision-making mechanism is designed, when the activation peak value of the dynamic neural field is attenuated to a preset threshold value, neural network feature matching is triggered, and disappearance target reproduction correlation is achieved based on cosine similarity. The objective of the invention is to solve the visual tracking capability under the condition of scene and target motion change in a monitoring range, for example, under an intelligent traffic intersection scene. The problems of high ID switching rate, multi-target misassociation and low tracking precision under a real-time tracking background caused by scene change or frequent shielding of vehicles and pedestrians, similar target appearances, transient disappearance and reproduction of the targets and sudden illumination change are solved.

The invention discloses an unmanned aerial vehicle intelligent obstacle avoidance decision-making method and system for a complex environment, and belongs to the technical field of unmanned aerial vehicle obstacle avoidance. The method comprises the steps of S1, environment perception and data acquisition, S2, multi-source data hierarchical fusion processing, S3, complex environment obstacle recognition and risk assessment, S4, dynamic obstacle avoidance path decision and optimization, and S5, obstacle avoidance execution and real-time adjustment. Omnibearing acquisition of obstacle distance, environment images, unmanned aerial vehicle attitude and position data is realized; the method combines an improved YOLOv8 model (adding a CBAM attention mechanism), strengthens the obstacle feature extraction capability under complex weathers such as rain and fog, strong light and dust, can accurately distinguish static obstacles such as buildings and trees from dynamic obstacles such as pedestrians, vehicles and other aircrafts, and solves the problems that a single sensor perceives blind areas and is insufficient in precision in a complex environment.

The invention discloses a sidewalk damage detection method and device based on deep learning, electronic equipment and a program product. The method is realized through a trained detection model, a DS-HAF module is introduced into a model neck network, and multi-scale feature fusion, channel-space joint attention enhancement and bidirectional residual error guided deep and shallow feature dynamic enhancement fusion are performed, so that the collaborative effect of detail and semantic expression is optimized, and complex background interference is effectively inhibited. An MAGRDet detection head is introduced into the detection network, multi-branch feature extraction and alignment fusion are realized through an MAGR module, and the detection capability of small targets, low-contrast targets and shielded crack targets is improved through a channel-space cascade attention modulation fusion result. And meanwhile, an MDPAR module is introduced into the backbone network, so that efficient and robust cross-scale feature extraction is realized, and the perception capability and the anti-interference performance of the model on multi-scale sidewalk cracks and damages are improved while low calculation overhead is maintained.

The invention provides a dynamic planning method for a fire evacuation path of a deep subway station based on a cellular automaton, and the method comprises the steps: building a subway space topology through employing the cellular automaton, simulating a fire environment through employing an FDS tool, and obtaining environment data; constructing a multi-field model based on subway space topology and environment data; based on the multi-field model, the pedestrian transition probability is calculated, then the pedestrian transition probability is corrected in combination with the deep subway facility characteristics and the pedestrian consensus behaviors, and finally the corrected pedestrian transition probability is output; and optimizing a cost function of an A star algorithm based on the corrected pedestrian transition probability to obtain a dynamic cost function, further obtaining an improved A star algorithm, planning an initial path based on the improved A star algorithm, formulating a path updating rule, and implementing a smooth transition strategy to realize dynamic path planning. According to the invention, the evacuation safety and the scheduling efficiency under the fire situation can be improved.

The embodiment of the invention provides a robot activity area division method and device, a robot, a storage medium and a program product. The method comprises the steps of obtaining kinematics data of the robot; and calculating a core braking area of the robot according to the kinematics data of the robot. And acquiring external environment information and system response time. And according to the external environment information and the system response time, performing activity area expansion on the core braking area, and calculating an early warning area. The robot movement area is determined according to the core braking area and the early warning area of the robot, and the safety of the robot and pedestrians is improved.

The invention provides a pedestrian and vehicle detection method and device based on multi-scale perception fusion, and the method comprises the steps: building a multi-scale perception fusion network structure which comprises an SPDConv module, a PSSF module, an FCPAM module and a DyHead module, solving a detection error caused by the problems of target shielding, size change, far-small proportion and the like in a complex traffic environment, and improving the detection precision. And the stability, the precision and the deployability of the detection model are improved.

According to the target tracking and identification method, the monitoring device and the storage medium disclosed by the invention, the image frames are continuously acquired from the monitoring video source by extracting the appearance characteristics of the specified main target in the current image frame, and the position of each pedestrian target in the image frames is detected in real time; matching the identity label and the motion trail of each pedestrian target in the current image frame; if the identity identifier of the main target is matched in the current image frame, performing similarity judgment on the historical motion trail of the main target and the motion trails of other pedestrian targets in the current image frame, and recording; and if the number of times of similarity in the plurality of continuous frames is greater than a threshold value, querying the final visible position of the main target, screening the main target according to the final visible position, and expanding the detection area step by step until the main target is screened or the whole image is covered when the main target is not screened. Therefore, the target identity abnormity is effectively recognized, the misrecognition risk is avoided, and the recovery efficiency and accuracy are improved.

The present application relates to the field of intelligent driving. Provided are a resource scheduling method and apparatus, and a system. The method comprises: running a first application program, wherein the first application program comprises a first critical data stream, the first critical data stream is used for determining the output of the first application program, the first critical data stream is associated with N threads under M processes, the M processes comprise a first process, the first process is associated with a first thread, a first function and a first stub, and the first thread is used for running the first function, M and N being positive integers; and on the basis of the first stub, allocating a first computing resource to the first thread, wherein the first computing resource is used for the first thread to run the first function. On the basis of the method, computing resources can be reasonably allocated to threads under application nodes corresponding to different critical data streams in autonomous driving applications, thereby helping to reduce the end-to-end latency and latency jitter of critical data streams in autonomous driving systems, and further ensuring the safety of vehicle owners and pedestrians outside vehicles.

A foot bridge structure comprises a plurality of fish skeletons which are mutually spliced and a bridge body which is formed by splicing the fish skeletons, each fish skeleton comprises a main body beam and cross beams, the cross beams are connected to the two sides of the main body beam respectively, the cross beams are symmetrically arranged relative to the central axis of the main body beam, and a height difference exists between the top of the main body beam and the cross beams. The waterproof liquid is accumulated at the connecting position of the main body beam and the cross beam, so that the connecting position of the main body beam and the cross beam is prevented from being rusted, a bridge floor bottom plate is laid on the fish skeleton, the top of the fish skeleton is wrapped through the bridge floor bottom plate, namely, the bridge floor bottom plate is laid on the upper end faces of the main body beam and the cross beam, and therefore the liquid is stored on the bridge floor bottom plate. The liquid is further prevented from being in contact with the main body beams and the cross beams, so that the main body beams and the cross beams are prevented from being rusted, a bridge deck slab is arranged on the bridge deck bottom plate, and guardrails are arranged on the two sides of the bridge deck slab.

The invention discloses a high-precision indoor positioning system based on a smart bracelet, and the system comprises a data processing module which is used for obtaining and preprocessing multi-source sensing data collected by the smart bracelet; the gait reckoning module is used for executing gait event detection and pedestrian dead reckoning processing; the wireless observation module is used for carrying out preprocessing and credibility evaluation on indoor wireless positioning observation data; the fusion positioning module is used for carrying out credibility weighted fusion; the trajectory constraint module is used for applying trajectory rationality constraint and correcting abnormal position points; the error correction module is used for constructing a positioning error dynamic correction model and executing online correction processing; and the track output module is used for carrying out smoothing and time resampling processing and extracting a human body position determination result set. The method is based on the multi-source perception fusion and federated Kalman dynamic correction method, achieves the indoor high-precision positioning of the smart bracelet, and has the advantages of continuous track, controllable error and high environmental adaptability.

The invention discloses an intersection right-turn vehicle and pedestrian interaction trajectory prediction method and computer equipment. The method comprises the following steps: acquiring a human-vehicle historical trajectory and initial state information in intersection right-turn vehicle and pedestrian interaction; a confrontation reinforcement learning structure is improved through KL regularization, and combined learning of a reward function and a strategy is achieved; a Nash Q-learning algorithm with a KL constraint is utilized to carry out joint optimization; and inputting the optimized strategy model into a behavior simulator, driving a generated prediction trajectory by an initial state, comparing the prediction trajectory with a real trajectory, and evaluating a prediction result based on a specified index. According to the method, the performance of the model in the aspects of prediction precision, behavior consistency and environment generalization ability can be effectively improved, the stability of a reward function and a strategy network is further improved, the generalization ability of the model in different traffic scenes subsequently is enhanced, and the learned strategy has better migration potential.

The invention discloses a linkage calling passing inspection method of a customs clearance camera, which is used for controlling calling of a first camera and a plurality of second cameras, the first camera is used for shooting an image of a pedestrian at a first visual angle, and the second cameras are used for shooting an image of the pedestrian at a second visual angle; the method comprises the following steps: a first camera tracks and records a plurality of first view angle images when a pedestrian approaches a gate; forming a pedestrian prediction track in a preset detection area of the first camera according to the first view angle images at the multiple different moments, and calling at least one second camera on the pedestrian prediction track; a second camera obtains a second visual angle image of the pedestrian and recognizes the second visual angle image and preset face information, and if recognition succeeds, a preset gate is controlled to be opened; and if the identification is wrong, closing the gate, and collecting the biological information of the pedestrian for verification. According to the linkage calling passing checking method, the calling efficiency of the camera module is effectively improved, and the passing efficiency of pedestrians is improved.

The invention relates to the technical field of traffic management, in particular to a school entrance and exit pedestrian traffic management system and method based on the Internet of Things, and the method comprises the steps: obtaining the road traffic monitoring data of a school entrance and exit in a school leaving period; the road congestion degree is evaluated by extracting vehicle distribution data; evaluating the traffic interference degree by identifying the vehicle driving direction and the mixed driving behavior of the vehicle and the pedestrian; and evaluating the school leaving speed of the students according to the road congestion degree and the traffic interference degree, and adjusting the school leaving strategy according to the school leaving speed of the students. The school leaving speed of the students is evaluated by quantifying the road congestion degree and the traffic interference degree of the school entrances and exits, the school leaving efficiency of the students is improved, and meanwhile the road traffic safety risk of the school entrances and exits is reduced.

The invention belongs to the field of computer vision, and particularly relates to an unsupervised visible light infrared pedestrian re-identification method. According to the method, a ResNet50 neural network is used as a feature extraction backbone network, a dual-module collaborative framework composed of a boundary pseudo-tag correction module and a distribution perception soft alignment module is constructed, the boundary pseudo-tag correction module is used as a basic component, and a core sample and a boundary sample are explicitly distinguished and clustered through a contour coefficient analysis strategy, so that the feature extraction accuracy is improved. And a dynamic evaluation mechanism in a Jaccard distance space is adopted to process samples with different confidence coefficients, so that the problem of disordered feature space structure caused by wrong distribution of boundary samples is effectively relieved.

The invention discloses a multi-camera collaborative pedestrian and vehicle cross-view tracking method and system, and relates to the technical field of intelligent monitoring. The method comprises the following steps: carrying out video stream acquisition on a monitoring area to obtain a multi-view video data set; multi-target detection of pedestrians and vehicles is carried out, tracking is carried out according to a plurality of detection targets, and a local tracking trajectory set of a plurality of cameras is generated; carrying out collaborative analysis on the plurality of cameras, constructing a multi-camera collaborative strategy to carry out cross-view trajectory matching on the local tracking trajectory set, and generating a global tracking trajectory set; and updating target states of the plurality of detection targets according to the global tracking trajectory set, and generating a multi-target cross-view tracking result of the pedestrian and the vehicle. The technical problem that in the prior art, information of different visual angles is difficult to integrate effectively in a multi-camera scene, so that target loss and track discontinuity easily occur in cross-camera tracking of pedestrians and vehicles is solved, and the technical effect of improving the continuity and accuracy of cross-camera tracking is achieved.

The invention relates to the technical field of computer vision, in particular to a multi-feature fusion visible light-infrared pedestrian re-identification method, which comprises the following steps of: respectively extracting low-layer specific features of two modes of an input image through independent visible light and infrared branches, and inputting the low-layer specific features into a shared deep network; a wavelet enhancement multi-feature generation module WMFGM and a coordinate attention fusion module CAFblock are introduced into the modal sharing network; a statistical normalized attention prototype module SNAP is introduced; the statistical normalization attention prototype module strengthens response of a target area through a statistical significance attention mechanism and performs prototype aggregation in a normalized semantic similarity space, so that local feature information with consistent semantics is extracted from visible light and infrared modalities; local feature information and global feature information are used for pedestrian distinguishing, various loss functions are used for guidance during model training, and cross-modal and intra-modal differences are effectively relieved.

The invention discloses a fine-grained unsupervised cross-modal pedestrian re-identification method based on large model semantic driving, and relates to a computer vision and mode identification technology. The method comprises the following steps: inputting an unlabeled visible light-infrared pedestrian data set, generating an image text description by using a vision-language model, and analyzing the image text description into a structured semantic attribute vector through a large language model; extracting visual features of the image and semantic query embedding corresponding to attributes, and generating fine-grained features of semantic alignment through a multi-head cross attention mechanism; fusing the visual similarity and the attribute similarity to generate a cross-modal pseudo tag; combining attribute-visual alignment loss and inter-attribute decoupling loss to optimize features; in the test stage, cross-modal matching retrieval is completed based on the optimized features. Through semantic analysis and enhancement, the problem that cross-modal feature alignment is insufficient in an unsupervised scene is solved, and experiments show that compared with a mainstream method, the model performance is improved, and the method can be applied to the fields of intelligent monitoring, cross-modal retrieval and the like.

An intelligent testing method for autonomous driving based on interaction of multiple background traffic participants, which includes the following steps: obtaining a test scenario and path information of traffic interactors in the test scenario, wherein the traffic interactors include a vehicle under test and background traffic participants, and the background traffic participants include background vehicles and background pedestrians; constructing a kinematic model of the background traffic participants based on the path information; dynamically constructing safety constraints between any two traffic interactors based on the path information; controlling the background traffic participants to conflict with the vehicle under test based on the kinematic model and the safety constraint, and constructing model prediction control to obtain a reference speed of the background traffic participants under collaborative interaction; and collecting real driving data of the vehicle under test and the background traffic participants in real time.

The invention discloses an intelligent alarm system based on a real-time video stream of a platform, and relates to the technical field of video processing early warning, and the intelligent alarm system comprises a data collection module which carries out the real-time monitoring based on a monitoring camera, reads the image data of a point position camera, carries out the standardization and normalization of the data, and carries out the unified input of the image data; and the target discovery module is used for inputting the preprocessed image frame and the monitored area, identifying and positioning pedestrians in the image frame by adopting an algorithm based on target detection, performing label processing on the identified pedestrians based on a detected tracking algorithm byterack, and outputting the residence time of the current frame target in the monitored area. According to the platform safety management system, various advantages are brought to platform safety management, the alarm threshold value can be flexibly adjusted, decision support is provided, management personnel are helped to analyze passenger flow, optimize deployment and troubleshoot hidden dangers by recording related data, a'early warning-processing-optimization 'closed loop is formed, the safety management and control level is improved, and safety accidents are effectively prevented.

The invention discloses a cross-modal pedestrian re-identification method and system based on a multi-scale joint learning network, and relates to the technical field of pedestrian re-identification, and the method comprises a four-flow network architecture which enables a model to extract diversified semantic features through a mode of separating an application data enhancement branch from an original branch. Random channel selection and self-adaptive graying are respectively applied to the data enhancement branch, so that the robustness of the model to color change and the adaptive capacity of the model to different thermal imaging conditions are improved. Important channels are enhanced through a channel attention mechanism, irrelevant channels are inhibited, and two substreams are guided to internally enhance modal specific features. Richer semantic information is reserved through features extracted by a four-flow network, a joint semantic learning module is designed, a group of learnable vectors are defined and spatial position codes are added, global features of original branches and color invariant features of channel data enhancement are fully learned under the guidance of loss, and the overall feature of the original branches is optimized; and the features among different modes have higher semantic consistency.

The embodiment of the invention discloses a pedestrian trajectory prediction method based on an interactive perception diffusion model, and the method comprises the steps: constructing a forward diffusion module, and obtaining a noise-added future trajectory of a target pedestrian; constructing a reverse denoising module, regarding social and scene interaction features as conditions of a parameter-containing reverse Markov chain, and performing denoising operation on a noise-added future trajectory of a target pedestrian; constructing a loss calculation module, and optimizing a pedestrian trajectory prediction model composed of the above modules by using the loss value to obtain an optimal pedestrian trajectory prediction model; and constructing a test denoising module, and denoising test initial random noise based on the optimal pedestrian trajectory prediction model to obtain a pedestrian trajectory prediction result. According to the method, the social encoder and the scene encoder are designed to extract the social interaction features and the scene interaction features respectively, so that the scene space information is reserved, the double guidance decoders are designed to strengthen the cooperative guidance of the two interaction features, and the accuracy and rationality of pedestrian trajectory prediction are further improved.

Provided is a vehicle control device capable of preventing unnecessary operations in a driving assistance function for avoiding a collision between an own vehicle and a pedestrian at an intersection, for example. There is provided an operation-suppression determination unit adapted to determine whether to change the result of collision determination (whether to suppress an unnecessary operation), based on the change (temporal change) of the lap rate between a target object and an own vehicle, and based on the change (temporal change) of the lateral speed generated in the target object with respect to the own vehicle, if a collision determination unit determines a collision. Further, there is provided a control-intervention determination unit adapted to determine control intervention in the own vehicle, based on results from the collision determination unit and the operation-suppression determination unit.

The invention relates to a vehicle-pedestrian safety interaction control method and system based on light blanket projection. The safety interaction control method comprises the steps that motion state information of pedestrians is acquired; identifying and classifying intentions of the pedestrians according to the motion state information of the pedestrians; constructing an interaction mode of each security level based on the type of the pedestrian intention and the motion state of the vehicle, wherein the interaction mode comprises the following contents: generating different light blanket patterns and projecting the light blanket patterns to the road surface; the system can recognize the position, posture and intention of a pedestrian in real time, can project dynamic interactive light blanket patterns with clear semantics to the pedestrian, integrates multiple interactive modes, and can intelligently switch and arbitrate control logic according to scenes.

The present application relates to the technical field of intelligent city infrastructure safety monitoring, and discloses a real-time response system for safety events of intelligent pedestrian tiles. The system comprises a real-time state monitoring module, a dynamic event detection module, a multi-dimensional event difference analysis module, a safety area correlation positioning module and a self-adaptive response generation module. The sensor array collects tile pressure distribution, surface humidity and temperature fluctuation data, and combines historical event data to construct a dynamic prediction model to output potential safety event probability values. A multi-dimensional difference analysis generates an event-level difference feature matrix, and a spatial topology network and environmental parameters are used to generate a safety event heat distribution map and locate high-risk areas. Finally, response parameters are adaptively configured. The system realizes real-time sensing, accurate positioning and dynamic response of safety events in urban pedestrian areas.

The invention belongs to the field of computer vision, particularly relates to an air-ground cross-view-angle pedestrian re-identification method based on multi-scale and view angle perception double flow, aims to match the same pedestrian individual under different view angles, and is applied to the industries of intelligent monitoring, public safety and the like. The system comprises two core branches, the multi-scale flow captures feature representations of different granularities, multi-scale semantic features are grouped and aggregated, and a feature representation which keeps consistency in a multi-scale space is learned, so that the model has robustness. According to the visual angle perception flow, a dynamic training strategy is adopted for different visual angle pairs, difficult samples are mined, the model can process visual angle changes more intelligently, and the cross-visual-angle matching capacity is remarkably improved. According to the method, feature representation with higher robustness and discrimination capability can be learned, and the Re-ID performance in a complex air-ground scene is remarkably improved.

The invention discloses an indoor positioning method and system based on 5G channel characteristics and pedestrian dead reckoning, and belongs to the technical field of indoor positioning. The method comprises the following steps: firstly, converting input complex 5G channel impulse response (CIR) data into a two-dimensional characteristic matrix; inputting the two-dimensional feature matrix into a deep convolutional neural network to perform feature transformation and obtain a 5G positioning position track, and then performing time sequence smooth optimization on the 5G positioning position track by using a Kalman filter to obtain an optimized 5G position track; according to the invention, a deep convolutional network is adopted to fully extract amplitude-phase composite features of 5G channel impact response CIR, and machine learning step length measurement and calculation of multi-dimensional gait features are combined to accurately position an indoor position track, so that the utilization efficiency of positioning data is improved; and intelligent complementation can be carried out between the optimized 5G position trajectory and the dead reckoning trajectory, and the indoor positioning precision and positioning stability in a complex indoor environment are improved.