851 results about "Traffic scene" patented technology

The invention relates to the technical field of intelligent driving, in particular to a method for generating an obstacle avoidance and planning cooperative path in an urban complex environment, which comprises the following steps: acquiring environment real-time sensing data through a vehicle-mounted multi-sensor, and constructing a dynamic semantic traffic matrix in combination with high-precision map static semantic information; based on the matrix, a multi-objective optimization algorithm is adopted to calculate the security cost, the efficiency cost and the rule conformity cost of the path, and a global optimization path is generated; inputting the global optimization path and the dynamic obstacle motion vector into an intention prediction model, generating dynamic obstacle future trajectory probability distribution and interactive intention classification, and further generating an avoidance strategy and adjusting a local path in real time; and inputting the adjusted local path into a kinematics model to carry out kinematics feasibility verification, and outputting an executable track or path re-planning. According to the method, the integration of dynamic environment understanding, path planning and obstacle avoidance strategies is realized, and the method is suitable for the path generation task of an automatic driving system in an urban complex traffic scene.

The invention discloses a unified framework for solving automatic driving track prediction and planning consistency based on a world model. According to the method, through cooperative work of the automatic driving domain controller and the vehicle-mounted sensing system, end-to-end joint optimization of track prediction and planning in a complex traffic scene is realized, time sequence dependence and interaction dynamics among intelligent agents are accurately captured, and the prediction capability and robustness of a model are remarkably improved. The method comprises the following specific steps: firstly, constructing a generative world model, and generating potential future state representation by utilizing a behavior conditional and backtracking expansion technology; secondly, in combination with global modeling and a local convolutional network, multi-scale features are extracted, adaptive fusion is carried out, and a multi-modal prediction trajectory is generated; then, a multi-target planning model is adopted to integrate various driving indexes, and a track with the minimum loss function is generated; finally, path planning parameters are dynamically optimized through real-time environment perception and decision feedback, and the problems of prediction uncertainty and planning consistency of the automatic driving track are effectively solved.

The invention discloses an intention recognition method based on cross attention and multi-scale uncertainty. The intention recognition method comprises the following steps: preprocessing multi-modal data; parallel multi-modal feature coding oriented to intention recognition; the invention relates to multi-scale uncertainty perception decoding. According to the method, a parallelized multi-modal feature extraction path is constructed, and a hierarchical fusion mechanism based on cross attention is designed, so that deep semantic alignment and complementary enhancement of four types of heterogeneous information including the posture, the motion track, the global scene and the local vision of a rider are realized; the problems of incomplete feature representation and insufficient cross-modal correlation modeling caused by dependence on a single information source or adoption of a shallow fusion strategy in a traditional method are solved, so that the accuracy and robustness of intention recognition in a complex traffic scene are remarkably improved. According to the method, a multi-scale uncertainty perception decoding framework is introduced, risk early warning or context auxiliary verification is carried out on a low-confidence identification result, and the reliability of an automatic driving system in a safety critical scene is improved.

The invention discloses an intelligent ship autonomous collision avoidance method based on COLREGs and a DDPG algorithm, and relates to the technical field of an intelligent ship technology and an autonomous collision avoidance algorithm, and the method comprises the steps: building an intelligent ship kinematics model based on the motion parameters of a ship in a north-east coordinate system; aIS, radar and visual information are fused, COLREGs rule constraints are embedded, an environment model is constructed, and a ship size safety threshold value is calculated. According to the method, COLREGs and a DDPG algorithm are combined, a high-precision environment model is constructed by utilizing multi-source sensing data, and an algorithm structure is optimized aiming at a typical collision avoidance scene, so that the intelligent ship can automatically identify the meeting situation type, determine the way-giving responsibility or the direct navigation obligation and generate an optimal collision avoidance path in a complex marine environment; compared with a traditional collision avoidance method depending on manual driving and fixed rules, the autonomous collision avoidance capability of the ship in dynamic and complex ocean traffic scenes is remarkably improved, and the collision risk caused by human factors is reduced.

The invention relates to the technical field related to automatic driving, in particular to a trajectory prediction method based on adaptive interaction and dynamic intention, which comprises the following steps: firstly, constructing a heterogeneous interaction map, dividing a traffic scene into a vehicle grid, an environment grid and a non-driving area grid, and embedding multi-dimensional dynamic features; then dynamically adjusting a region of interest based on the behavior intention of the target vehicle, and extracting a high-correlation interaction subnet; modeling an interaction relationship by adopting a heterogeneous graph convolutional network, and processing the motion characteristics of the target vehicle and the neighbor vehicle through a sub-channel coding strategy; further realizing dynamic intention perception through a double-branch parallel attention architecture, and fusing macroscopic intention and dynamic intention information; and finally, iteratively generating a future trajectory prediction result based on a decoding architecture of a message passing mechanism. The method can effectively improve the long-term prediction performance in a lane changing scene, adaptively captures a dynamic interaction relationship, and improves the adaptability of a prediction system to the behavior intention change of a driver.

The invention discloses a real-time monitoring system based on highway traffic flow monitoring, and the system specifically comprises a video collection and self-calibration module which obtains calibration video data with an aligned visual angle; the edge consistency adaptive enhancement module outputs an enhanced video frame, an edge intensity graph and a noise risk graph, and constructs an edge confidence graph; the improved segmentation SAM module is used for carrying out traffic scene domain adaptation training on the improved segmentation SAM model by adopting Adapter to generate an initial multi-target segmentation mask; the deformation optimization module is used for calculating a corresponding boundary stability index; the traffic flow index extraction module is used for generating a corrected tracking result; the abnormal event candidate recognition module is used for generating an event confidence score for each traffic flow abnormal event candidate; and the abnormal event output module is used for outputting graded abnormal event alarm information. According to the method, the problems of boundary edge eating, missing detection and inter-frame drifting are remarkably reduced, and high-time-space-consistency segmentation is realized.

The invention discloses a vehicle trajectory planning method and device and a vehicle, and relates to the technical field of intelligent driving. The method comprises the following steps: obtaining vehicle state information of a target vehicle, a reference trajectory and environment perception data containing available traffic width, and performing trajectory optimization processing according to the vehicle state information, the reference trajectory and the environment perception data to obtain a trajectory optimization result of the reference trajectory; in the trajectory optimization processing process, taking a first passing cost required for minimizing a trajectory optimization result as a target; a passing width cost item in the first passing cost is a power function taking the residual passing width as an independent variable, and the passing width cost item is increased along with the reduction of the residual passing width; the remaining passing width is the difference between the available passing width and the vehicle width of the target vehicle. Therefore, in a narrow traffic scene, the traffic width cost item is increased in a super-linear manner, the constraint strength on the transverse position of the track is enhanced, the generation of a high-risk track excessively close to an obstacle is effectively avoided, the planning interruption frequency is reduced, and the trafficability of the vehicle in the narrow scene is remarkably improved.

The invention relates to a method and system for identifying a highway event by using a video large model, and the method comprises the steps: employing a three-stage processing architecture, firstly carrying out the real-time target detection and preliminary event judgment of a highway monitoring video stream through employing a YOLO algorithm, and generating an event candidate set; inputting the candidate events and the video clips thereof into a specially trained visual large model for deep semantic analysis and secondary reasoning; and finally, a reasoning result is rechecked through a rule engine, and false alarms are filtered by applying illusion suppression and a space-time association rule. According to the method, the real-time performance of traditional target detection and the deep reasoning capability of a visual large model are fused, so that the problems of high false alarm rate and high missing report rate of a traditional method are effectively solved, the accuracy and reliability of event identification in a complex traffic scene are remarkably improved, and meanwhile, the real-time processing capability of a system on multiple paths of high-definition video streams is ensured.

The invention discloses an intelligent driving scene understanding and decision-making method and system based on a multi-modal large language model, and relates to the technical field of intelligent driving scene understanding and decision-making, and the method comprises the steps: collecting the visual, radar, laser radar, Internet of Vehicles, voice and vehicle state data of the surrounding environment of a vehicle, and forming a multi-modal original input set; performing feature extraction and semantic coding on various data in the multi-modal original input set to generate semantic feature vectors of corresponding modals; unified space mapping is carried out on the semantic feature vectors through a cross-modal alignment mechanism, multi-modal fusion processing is carried out based on an alignment result, and comprehensive semantic representation is generated; analyzing a social interaction relationship in the traffic scene based on the comprehensive semantic representation, identifying action modes and behavior tendencies of surrounding traffic participants, and generating social intention description information; and generating a scene query request according to the comprehensive semantic representation, and matching related traffic rules and driving experience in a pre-constructed driving common knowledge base.

The invention provides a passable area reasoning method and system based on a visual language model, and the method comprises the steps: obtaining the multi-modal data of a vehicle and the current position information of the vehicle; analyzing the multi-modal data, and determining visual features and traffic symbol features; performing spatial position coding on the visual object and the traffic symbol elements, and determining aerial view angle coordinate information; performing semantic alignment on the visual features and the traffic symbol features, and determining a shared embedding representation; constructing a traffic semantic map by fusing, sharing and embedding representation based on a graph neural network and bird's-eye view coordinate information of a visual object and a traffic symbol element; and according to the current position information of the vehicle, the traffic semantic map and a preset traffic rule, generating a bird's-eye view semantic map including a passable area, a no-pass area and a semantic association relationship. According to the method and the device, semantic alignment and consistency expression of visual perception and traffic symbol recognition are realized, and further feasible region reasoning of a complex traffic scene is realized.

The invention discloses a transmitting power determination method and device, electronic equipment, a medium and a product, and relates to the technical field of radars. The transmitting power determination method comprises the following steps: acquiring historical time sequence characteristic information and environmental point cloud data acquired by a radar sensor, and determining target point cloud data and traffic scene perception characteristics according to the environmental point cloud data; determining a motion mode of a traffic participant in the environmental point cloud data according to the traffic scene perception feature and the historical time sequence feature information; determining participant categories of traffic participants in the target point cloud data based on a preset target classification model and the motion mode, and constructing a radar scene model according to the traffic scene perception features, the historical time sequence feature information and the participant categories; and obtaining target scene data of the radar scene model, and determining the target transmitting power of the radar sensor through the preset power determination model and the target scene data, thereby realizing adaptive determination of the target transmitting power according to the traffic scene perception characteristics and historical time sequence characteristic information.

The invention discloses an intelligent network connection automobile active safety teaching control method based on digital twinning. The method comprises the steps that S1, multi-source sensor data are collected and preprocessed to generate a driving state data set; s2, inputting the state data into a digital twin model to drive a traffic scene and outputting a synchronous state; s3, based on Dueling-DDQN, executing strategy learning to generate an active safety control instruction; s4, the instruction response effect is simulated and verified in the virtual environment; s5, collecting driver operation behaviors, inputting the improved MHA-BiLSTM model to extract time sequence features, and outputting behavior features; and S6, comparing the driving behavior with a standard instruction item by item, calculating an operation deviation and a response difference, and generating a personalized active safety teaching task. According to the invention, efficient comparison and teaching feedback of the driving behavior and the active control strategy can be realized, and the intelligent level of driving training is improved.

The invention relates to an intelligent driving control method of a vehicle and the vehicle, and belongs to the technical field of intelligent driving, and the method comprises the following steps: obtaining an original data stream of a vehicle sensor network, and carrying out feature extraction on a dynamic feature vector; state vectors of the traffic participants are extracted according to the dynamic feature vectors, edges between the two traffic participants are constructed, a traffic participant interaction graph is constructed, and an asymmetric factor matrix is calculated; acquiring a historical scene data set of the vehicle, calculating correction similarity, and calculating a weight coefficient; performing weighted fusion on the weight coefficient and the historical scene data set to obtain an enhanced data set; and performing updating training on the pre-trained intelligent driving decision model according to the enhanced data set to obtain an updated intelligent driving decision model, generating an intelligent driving decision, and controlling vehicle operation, thereby realizing more accurate characterization of a dynamic game relationship in a complex traffic scene, and improving the accuracy of the dynamic game relationship. And the intelligent driving decision model can continuously adapt to an asymmetric interaction mode effect in a real scene.

The invention relates to a non-linear feature extraction-based roadside end radar and camera fused three-dimensional target detection method and device, and a medium. The three-dimensional target detection method comprises the steps of synchronously completing data enhancement and downsampling of an image and a point cloud; extracting high-dimensional nonlinear features by using an encoder improved by a Kolmogov-Arnod network, and projecting the high-dimensional nonlinear features to a unified aerial view space; establishing cross-modal dependence by using multi-head cross attention; carrying out weight fusion by using nonlinear convolution to generate an integrated aerial view feature map; and the decoder and the detection head output the three-dimensional coordinate, the size and the category of the target. Compared with the prior art, the Kolmogorov-Arnod network is introduced to perform nonlinear enhancement on the image and the point cloud encoder, the cross-modal weight is dynamically calculated through multi-head cross attention in the aerial view space, and finally, the weighted fusion is completed by using the KANs convolution. Therefore, the three-dimensional detection precision in a complex traffic scene is remarkably improved while the consistency of the receptive field is ensured.

The invention belongs to the field of artificial intelligence and intelligent traffic systems, particularly relates to a path control system combining multi-modal perception and dynamic trajectory prediction, and aims to solve the problems of insufficient perception fusion, low trajectory prediction precision and control response lag in a complex dynamic environment. The system comprises a multi-modal perception fusion unit, a dynamic interaction modeling unit, a space-time coupling prediction unit, a risk field construction unit and an adaptive path generation unit. Multi-source sensor data are fused through confidence coefficient weighting, an interaction weight matrix under an attention mechanism is constructed, a future trajectory is predicted in combination with individual dynamics and a social force model, a four-dimensional space-time risk field is generated, and a minimum risk path is solved based on an improved fast marching algorithm. The system realizes sensing-prediction-control closed-loop cooperation, the end-to-end delay is less than 250 milliseconds, and the driving safety and comfort in a complex traffic scene are improved.

The invention relates to a traffic scene vehicle and event identification method based on cooperation of an edge small model and a cloud large model, and belongs to the technical field of intelligent traffic. Aiming at the problems of low recognition precision, dependence on a large amount of labeled data, incapability of recognizing unknown categories and the like in a complex environment in the prior art, the method provides a multi-modal traffic visual perception coding system, a large model enhanced small sample cooperative training algorithm and a dynamic trigger type double-model reasoning framework. Small target feature expression is enhanced through semantic and visual joint coding, dependence of a small model on annotation data is reduced by using a large model pseudo tag and knowledge distillation, and a cloud large model is dynamically called according to confidence and scene complexity for secondary discrimination. According to the method, the recognition robustness of the system under severe conditions is effectively improved, and the balance between open vocabulary perception and low-resource efficient deployment is realized.

The invention relates to the technical field of intelligent traffic control, and discloses a road traffic AI adaptive edge computing server. The server comprises a traffic situation sensing module, a traffic flow intention analysis module, a control strategy construction module, a control scheme generation module and an efficiency rolling optimization module. The server synchronously receives the original data flow of the heterogeneous traffic sensor, and extracts and generates a microscopic traffic behavior sequence after timestamp alignment and cleaning. A behavior sequence is matched with a historical scheme library, a control strategy knowledge graph based on a traffic entity relationship is constructed, and a candidate control scheme set is reasoned according to the control strategy knowledge graph. And performing conflict detection and rolling optimization on the candidate schemes based on a short-time traffic flow prediction result, and outputting a final adaptive control instruction set. According to the invention, deep understanding and foresight control of a complex traffic scene are realized, and the intersection passing efficiency and the control adaptive capability are improved.

The invention provides a dynamic traffic scene-oriented unmanned aerial vehicle charging base station site selection and scheduling collaborative optimization method. The method comprises the following steps of: 1, establishing a dynamic traffic flow space-time distribution model and an emergency event probability model based on a GIS (Geographic Information System) platform and traffic monitoring data; step 2, constructing a multi-target dynamic optimization model; 3, solving the multi-target dynamic optimization model by adopting an improved multi-target genetic algorithm; 4, a site selection and scheduling double-layer optimization structure is constructed, and unmanned aerial vehicle task allocation and base station utilization rate balance is realized through an unmanned aerial vehicle charging task scheduling model; and 5, realizing dynamic deployment and scheduling optimization closed loop of the base station through real-time data feedback. The method has remarkable advantages in the aspects of coverage rate, energy consumption, response time delay, system stability and the like. The emergency response speed and the energy utilization efficiency are remarkably improved, the system operation and maintenance cost is reduced, and the method is suitable for scenes such as intelligent traffic and emergency communication.

The invention discloses an end-to-end automatic driving method based on hierarchical uncertainty estimation and dynamic safety response, and belongs to the technical field of intelligent driving. The method comprises the steps of environment observation data acquisition and preprocessing, perception information enhancement, adaptive space-time attention fusion and trajectory prediction, hierarchical uncertainty estimation and dynamic safety response, experience pool storage and end-to-end deep learning optimization. By dynamically integrating the multi-modal features of the RGB image and the LiDAR point cloud, a space-time dependency relationship is captured, and high-precision trajectory prediction is realized; meanwhile, layering quantification cognition and random and time sequence uncertainty are carried out, and a five-level safety response strategy is triggered in combination with an environment self-adaptive threshold value; and carrying out reinforcement learning by adopting priority experience playback and a multi-task loss function. According to the method, the limitation of traditional fixed weight fusion is overcome, the prediction accuracy and the system safety are improved, the average displacement error (ADE) can be reduced in a complex traffic scene, the uncertainty calibration degree is improved, and the method is suitable for real-time decision making of an automatic driving vehicle.

The invention relates to a traffic flow prediction method and system based on a dynamic perception expert network, and the method comprises the steps: obtaining traffic observation data, constructing a traffic network diagram, carrying out the feature embedding, and generating an initial spatial-temporal feature vector; inputting the initial spatial-temporal feature vector into a double-path time encoder, and respectively extracting a personalized time sequence feature vector and a time sequence dynamic feature vector through parallel channel independent paths and channel mixed paths; carrying out vector fusion through a gating mechanism to obtain a time context feature vector, inputting the time context feature vector into a multi-scale hybrid expert model, activating a plurality of most relevant time scale experts, and generating a corresponding routing weight; and generating a spatial dependency graph through a scale condition dynamic graph generator, performing graph convolution to extract a plurality of spatio-temporal feature vectors, performing weighted aggregation, sending the spatio-temporal feature vectors into a prediction model, and generating a traffic prediction value. Compared with the prior art, the model constructed by the method is relatively high in prediction precision and relatively high in robustness in a complex traffic scene.

The embodiment of the invention provides a trajectory prediction method and device and a vehicle. The method comprises the steps that environment information around a vehicle, pedestrian information, intention probability and scene probability are acquired, and the intention probability is used for representing the probability that pedestrians execute each predefined pedestrian behavior mode in each predefined typical traffic scene; the scene probability is used for representing the probability that the traffic scene where the vehicle is located currently is the typical traffic scene; and performing trajectory prediction on the pedestrian based on the environment information, the pedestrian information, the intention probability, the scene probability and a pre-constructed space-time prediction network. According to the method, the scene probability and the intention probability are utilized to guide the space-time prediction network to learn a pedestrian motion mode under scene constraint and intention guidance, and the trajectory prediction accuracy can be improved.

The invention relates to a vehicle control method and device, a vehicle and a computer readable storage medium, and the method comprises the steps: determining a target input sequence based on image information and vehicle state information of a target vehicle in a driving process; the target input sequence comprises a visual token sequence; performing planning prediction processing on the target input sequence by using a planning decision module to obtain initial driving information of the target vehicle; performing confrontation reasoning on the initial driving information and the visual token sequence by using a safety review module to obtain risk warning information of the target vehicle; and performing arbitration processing on the initial driving information and the risk warning information by using an arbitration module to obtain a control instruction of the target vehicle, so that the target vehicle runs according to the control instruction. According to the invention, the path decision security in a complex traffic scene in the automatic driving process of the vehicle can be improved.

The invention discloses a vehicle speed anomaly detection method and electronic equipment. The method comprises the following steps: acquiring trajectory data of a plurality of vehicles; one or more feature values of each piece of trajectory data are constructed, the trajectory data comprise continuous multi-frame vehicle data, the feature values at least comprise speed feature values and partition codes, and the partition codes are codes of areas where geographic positions of vehicles in the trajectory data are located; the characteristic values of the trajectory data with the same partition codes are input into a speed anomaly detection model, abnormal trajectory data output by the speed anomaly detection model are obtained, and the speed anomaly detection model is obtained through training of an isolated forest algorithm. The method can adapt to different traffic scenes and abnormal types, and the false alarm rate and the missing report rate are reduced. In addition, an isolated forest algorithm is used for training a speed anomaly detection model, and the requirement for real-time anomaly detection is met. Meanwhile, the algorithm does not depend on specific distribution of data, and traffic data in non-normal distribution can be processed.

The invention discloses an automatic driving decision-making method and system for a port mixed traffic intersection based on deep reinforcement learning, and relates to the technical field of automatic driving. Comprising the following steps: acquiring real-time state information of automatic driving transportation equipment in a port mixed traffic intersection environment, and fusing the real-time state information into an automatic driving transportation equipment state vector in a heterogeneous manner; constructing a reinforcement learning decision model based on the DDPG; building a port traffic environment simulation platform; the reinforcement learning decision model based on the DDPG is trained; and solidifying and deploying the trained reinforcement learning decision model based on the DDPG in a vehicle-mounted calculation unit of the automatic driving transportation equipment, and controlling the driving of the automatic driving transportation equipment. The traffic efficiency, stability and safety of the automatic driving transportation equipment at the port intersection can be improved, the unmanned transportation process of the port mixed traffic scene is promoted, and the wharf transformation and operation cost is reduced.

The invention discloses a vehicle-road cooperation virtual-real fusion simulation test method, device, equipment and medium, and the method comprises the steps: setting a test event in a virtual simulation environment, generating virtual sensor detection data, sending the data to a cloud control platform, carrying out the fusion processing, generating a standard vehicle-road cooperation message, and issuing the message to a real road side unit, and the real road side unit forwards the message to the real vehicle-mounted unit, the real vehicle-mounted unit analyzes the message, generates a vehicle control instruction and returns the vehicle control instruction to the virtual simulation environment, and the state of the virtual vehicle model is updated according to the vehicle control instruction in the virtual simulation environment and response verification is completed. According to the method, a virtual-real combined vehicle-road cooperation test link is constructed under a unified framework, and a complete closed loop from virtual sensor sensing, cloud fusion processing, roadside forwarding, vehicle-mounted analysis to virtual vehicle response is realized, so that the test coverage rate and controllability of a vehicle-road cooperation function in a complex traffic scene are effectively improved; and the test efficiency and the verification precision are improved.

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.

A high-order driving auxiliary lane changing trajectory planning method for an intelligent vehicle comprises the steps that laser radar and camera data are obtained, and feature vectors of obstacles are extracted after the data are processed by an ISP at the front end of a chip of the intelligent vehicle; constructing a spatio-temporal joint attention network model, inputting the feature vector of the obstacle into the spatio-temporal joint attention network model, calculating an interaction weight between the vehicle and the obstacle through a self-attention mechanism, and synchronously outputting predicted trajectory distribution of the obstacle and a candidate planning trajectory cluster of the vehicle; introducing a dynamic game in the game theory, and constructing a comprehensive cost function in combination with the interaction weight to obtain a screened track; and taking the screened trajectory as an initial value, inputting the initial value into a rear-end quadratic programming optimizer on the premise of meeting vehicle dynamics constraints, carrying out numerical optimization solution, searching an optimal programming sequence, and finally finishing lane changing trajectory planning. According to the invention, the interactive perception capability and riding comfort of the intelligent vehicle in a complex traffic scene can be improved.

The invention discloses a video anomaly detection method based on a traffic scene, and relates to the related field of computer vision, and the method comprises the steps: obtaining an input video clip which comprises a plurality of frames of images, traversing the plurality of frames of images, carrying out the selection through combining with a text prompt, determining a plurality of key frames, and enabling the text prompt to be input by a user; performing context generation based on the plurality of key frames to obtain position and time context information; the key frame, the position, the time context information and the text prompt are synchronized to a large visual language model for visual questions and answers of diversified traffic scenes, abnormal events are extracted, and the abnormal events comprise abnormal scores; and carrying out abnormal event detection analysis according to the abnormal score, and carrying out abnormal detection on the diversified traffic scenes. The technical problem that abnormal events in diversified traffic scenes are difficult to comprehensively and accurately detect in existing traffic scene-based video anomaly detection is solved, and the technical effect of improving the accuracy and generalization of traffic scene video anomaly detection is achieved.

The invention relates to the technical field of intelligent traffic, and discloses a vehicle trajectory prediction method and system based on time-space diagram attention fusion. The method comprises the steps of collecting historical trajectory data of a vehicle; the bidirectional long-short-term memory network is used as an encoder to carry out bidirectional encoding on the displacement sequence, and time sequence characteristics of the vehicle are obtained; introducing a graph attention network, regarding the vehicle as a node of a graph, constructing a sparse edge index based on a central position, and obtaining spatial features of the vehicle through multi-layer graph attention convolution and based on time sequence features of the vehicle; after the time sequence features and the spatial features of the vehicle are fused, the fused features are input into an encoder for global attention calculation; and carrying out decoding, injection modal embedding and position coding on the output characteristics of the encoder to generate multi-modal future trajectory prediction. According to the method, the vehicle trajectory can be predicted more accurately through cooperative combination of time and space processing, especially in a complex traffic scene.

The invention discloses a sequence-based vehicle type category correction method and device, and relates to the technical field of intelligent traffic systems, and the method comprises the steps: obtaining a vehicle detection model, and collecting a backflow video data set; traversing the set, detecting each frame of vehicle target by using the model, and associating cross-frame targets by using a target tracking technology to obtain a vehicle trajectory and form a plurality of vehicle target sequences; traversing the sequence to count vehicle types, and correcting the vehicle types according to a statistical result to obtain a plurality of corrected vehicle target sequences; identifying a missing detection frame based on an original sequence, predicting a missing detection target position, and obtaining a plurality of missing detection data sets; and iteratively updating the vehicle detection model in combination with the correction sequence, the missing detection data and the original sequence. According to the invention, the technical problems of vehicle type judgment errors and target detection deficiency due to the fact that multi-frame associated information is not combined in a traffic scene in traditional vehicle detection are solved, and the technical effects of improving the vehicle type detection accuracy and integrity and realizing continuous optimization of the detection capability are achieved.