86 results about "Pedestrian recognition" patented technology

The invention relates to the technical field of computer vision and cross-visual-angle pedestrian recognition, in particular to a cross-visual-angle pedestrian re-recognition method based on visual language prompt learning, and the method comprises the steps: obtaining a target image; the target image is input into a preset pedestrian recognition model, a pedestrian re-recognition result is output, the pedestrian recognition model is obtained by training a visual language pre-training model CLIP through a prompt learning mechanism and a two-stage training strategy, the prompt learning mechanism is used for modeling visual angle deviation, and the two-stage training strategy is used for training visual angle deviation. The double-stage training strategy is used for realizing cross-modal semantic alignment. According to the invention, the accuracy and robustness of cross-view identification can be significantly improved.

The invention relates to the technical field of automatic driving, in particular to an automatic driving system based on pedestrian motivation and a control method, and the system comprises a data acquisition module, a pedestrian detection module, a pedestrian intention reasoning module and a vehicle control module. The data acquisition module is used for acquiring a video sequence and environment information of a traffic intersection when the intelligent driving vehicle drives to the intersection without the signal lamp. The pedestrian detection module detects pedestrians through video processing and a pedestrian recognition strategy and outputs pedestrian inter-frame images. The pedestrian intention reasoning module classifies pedestrian crossing motivations based on the trajectory and attitude information of pedestrians. The vehicle control module dynamically adjusts the vehicle speed according to the pedestrian motivation type and automatically brakes when necessary so as to ensure that pedestrians safely pass through the intersection. According to the system, self-adaptive comity of the intelligent driving vehicle to pedestrians at the intersection without signal lamps is realized, and the traffic safety and the adaptive capacity of intelligent driving are effectively improved.

The invention belongs to the technical field of traffic safety pre-warning, and provides a pedestrian zebra crossing pre-warning system device which comprises a mounting stand column, a monitoring assembly and a pre-warning assembly, a connecting base is arranged at the top of the front side of the mounting stand column, the monitoring assembly is arranged on the connecting base, and the monitoring assembly comprises an AI camera, a laser radar and a far infrared sensor. The early warning assembly comprises an electronic warning board, a laser transmitter and a laser spotlight; by comprehensively using various sensors such as the AI camera, the infrared sensor and the laser radar, pedestrians can be monitored and recognized from different angles and different modes, the accuracy of pedestrian recognition is improved, and the method is particularly suitable for pedestrian detection under complex weather and illumination conditions and has a wide application prospect. Through the synergistic effect of the laser spotlight, the laser curtain wall, the electronic warning board and the warning sound, early warning is performed on drivers and pedestrians from multiple aspects of visual sense and auditory sense, the early warning effect is enhanced, and the safety of the pedestrians crossing the road is improved.

The invention discloses a multi-sensor fused intelligent AGV path planning method and system, and the method comprises the steps: enabling an AGV to carry a 360-degree laser radar sensor, a binocular vision sensor, an infrared thermal imaging sensor, an ultrasonic array sensor and other sensors, processing data through a low-rank multi-source feature fusion module LMF, and constructing a high-precision two-dimensional grid map through a VF-Cart algorithm; the global path planning uses an improved MRRT * algorithm, and the algorithm combines the dynamic data processing capability of a Markov decision process MDP and the efficient search capability of RRT * to ensure the path optimality; local path optimization adopts an FPA * algorithm fusing an artificial potential field method APF and a dynamic heuristic search algorithm DHPA *, local optimum is avoided, and intelligent obstacle avoidance is realized in combination with deep reinforcement learning DRL; according to the method, centimeter-level obstacle detection and dynamic pedestrian recognition are realized in a complex dynamic environment, the path planning efficiency and the motion stability of the AGV are remarkably improved, and the scheduling time is shortened.

The application discloses a low-light target detection method based on a double-layer optimization theory, which significantly improves the enhancement quality and detection accuracy. The method first adopts a double-layer learning architecture to bridge the enhancement and detection tasks, and organically combines the low-light image enhancement task and the target detection task. Secondly, the enhancement network parameters are taken as the main optimization target, and the detection network parameters are taken as the constraint condition. The enhancement network generates images that are conducive to machine perception. Finally, in order to improve the training efficiency, a method based on the implicit function theory is proposed to approximately solve the corresponding gradient. The training cost is effectively reduced, and the training speed is improved. The application significantly improves the target detection accuracy in weak light scenes, and can be applied to various target detection tasks in weak light scenes, such as night pedestrian recognition, automatic driving, security monitoring and the like.

The application discloses a kind of based on three modal consistency constraint cross-modal pedestrian re-identification method and system, the method includes: constructing pedestrian identification model;Fusion color and infrared image, produce enhanced sample;Color, infrared and enhanced image input model, extract image feature map;Implementation pooling operation obtains the overall and local features of image, input classifier completes classification task;Construct tri-center loss, reduce the feature difference between modal;Compact modal constraint is built, reduce the feature difference within modal;Joint cross-entropy, tri-center loss and compact modal constraint, optimize pedestrian identification model;Test application stage, the overall and local features of pedestrian are extracted, and it is concatenated to carry out pedestrian matching, and recognition result is obtained.The application makes full use of infrared and color image information, produces enhanced image to reduce the difference between modal, constructs loss to solve the problem that modal and modal difference is big, improves the matching accuracy of cross-modal pedestrian re-identification.

This application relates to the field of image detection technology, specifically to a method, device, and system for pedestrian trajectory tracking across cameras. Specifically, it includes: dividing a surveillance image into far-field, mid-field, and near-field regions; analyzing the imaging characteristics of moving pedestrians in different regions to construct regional change coefficients for far-field and near-field regions at different time periods; constructing motion change influence coefficients for each region by analyzing the pixel feature changes of clusters within the surveillance image during pedestrian movement and the grayscale changes of each pixel over time; determining the tracking region in each far-field and near-field region based on the regional change coefficients and motion change influence coefficients; reducing the hard threshold for denoising in the tracking region and mid-field region to retain more image details, thereby performing pedestrian trajectory tracking; avoiding the problem that traditional denoising algorithms use the same denoising parameters for the entire image, which may smooth the pedestrian's features during denoising, affecting the accuracy of pedestrian recognition.

The invention relates to a multi-spectral image and point cloud fusion pedestrian identification and safety protection method based on cross-modal feature cooperative constraint, and belongs to the technical field of target identification. The method comprises the following steps: firstly, carrying out space-time synchronization on a point cloud acquired by a laser radar and an image acquired by a multispectral camera, and then preprocessing synchronized point cloud data and image data; in the segmentation stage, two paths of complementary image segmentation paths are adopted to respectively extract pedestrian and dangerous object areas, and two paths of segmentation results are fused and complemented, so that the stability and accuracy of pedestrian identification are improved. And finally, performing morphological optimization on the segmentation mask, fusing with the point cloud data, and realizing calculation of the shortest distance between the pedestrian and the dangerous object in combination with a convex hull grid projection algorithm of spatial index acceleration so as to support personnel safety protection and target early warning. According to the invention, the identification degree of pedestrians and dangerous objects in the image can be improved, and an active safe distance protection function is realized.

The application discloses a kind of identification methods of pedestrian on crosswalk, the method utilizes vehicle front camera to obtain the image corresponding to vehicle front scene, obtains the color block in image and carries out screening, and then obtains the regression straight line of color block, utilizes the characteristics of the longitudinal boundary line of white color block of crosswalk in the image of front camera compared to a point to the regression straight line in image is screened and numbered, and the regression straight line screened is the longitudinal boundary line of white color block of crosswalk.By the change of regression straight line number in detection process, the shielding condition of pedestrian to the white color block of crosswalk is reflected, and then the identification of pedestrian on crosswalk is realized.

The invention relates to the technical field of image data processing, in particular to a vehicle-mounted thermal imaging target AI identification method supporting aerial imaging display. The method comprises the following steps: acquiring a thermal imaging video stream and vehicle speed data shot in a vehicle driving process; identifying each suspected pedestrian area of adjacent image frames in the thermal imaging video stream, and calculating the area similarity of each corresponding suspected pedestrian area of the adjacent image frames; performing picture consistency matching on the adjacent image frames to enable the adjacent image frames to complete image alignment processing; reflecting the change trend of the whole picture by using the vehicle speed data, and obtaining a pedestrian body region and a pedestrian limb region in the aligned picture by combining the region similarity of each suspected pedestrian region in the aligned picture; and determining whether pedestrian features exist in the aligned picture or not, and marking and displaying the pedestrian features when the pedestrian features exist in the corresponding region. Based on the technical scheme of the invention, accurate pedestrian recognition can be realized.

The application provides a kind of automobile crane personnel intrusion monitoring method and device, belong to engineering machinery safety monitoring technical field.The method includes: synchronous acquisition contains the RGB image of pedestrian information, disparity map and three-dimensional point cloud data, and time-space registration is carried out and is preprocessed to establish training set;Personnel intrusion monitoring model of automobile crane is constructed, including feature extraction module, boom prediction module, pedestrian position prediction module and time-space registration module;Personnel intrusion monitoring model of automobile crane is trained using training set data;Collect construction site RGB image, disparity map and three-dimensional point cloud data, after time-space registration, input personnel intrusion monitoring model of automobile crane completed training according to the relationship between the position of pedestrian, motion direction and fence, determine intrusion risk.The application improves the pedestrian recognition accuracy in complex environments such as rain, fog and occlusion through the time-space attention feature fusion mechanism, and can provide early warning for potential intrusion behavior.

The invention relates to a method for controlling a driver assistance system in a road vehicle (22) having an electronically controllable drive motor (24) and an electronically controllable service brake system having wheel brake cylinders (30, 32, 34, 36) which can be actuated with pressure medium, and having a camera-based vehicle recognition, pedestrian recognition and traffic sign recognition, as well as a speed regulation function and an emergency brake function, wherein, when the road vehicle (22) approaches an identified bus stop or tram stop (2), the controller (26) of the driver assistance system, in addition to generating and outputting a warning signal to the driver of the road vehicle (22), also carries out at least one further measure to avoid a collision with a pedestrian (14.1, 14.2, 14.3). As an additional measure, according to the invention, the wheel brake cylinders (30, 32, 34, 36) of the service brake system of the road vehicle (22) are pre-filled with a predetermined fluid pressure before reaching the bus stop or tram stop (2) in order to prepare for a possible emergency brake. The invention also relates to a device (40) for carrying out the method.

The present invention provides a method and device for pedestrian recognition. The method comprises: after determining that the tracking identifier trackID of a target person is lost during the tracking process of a frame image in a target video, obtaining a predicted trajectory of the target person based on a target candidate sequence and a historical position sequence; and determining whether the target person is occluded or has left a boundary; performing different processing on the target candidate sequence based on different states of the target person; and identifying the target person using a REID algorithm and face recognition based on the target candidate sequence during the period when the trackID is lost. The present invention processes the target candidate sequence by predicting the pedestrian's motion trajectory during the period when the pedestrian tracking identifier trackID is lost in the video frame image, thereby improving the stability and continuity of target tracking in the occluded state. In the state when the pedestrian has left a boundary, video splicing technology is used to verify the correctness of the pedestrian's motion trajectory, thereby improving the accuracy of pedestrian recognition.

The application discloses a light neural network model edge end cross identification vehicle courtesy pedestrian method and system, belongs to the technical field of identifying vehicle courtesy pedestrians, and comprises the following steps: S1: a vehicle-mounted road surface video collector of a bus video acquisition system collects three groups of picture data of no zebra crossing picture D1, no pedestrian on the zebra crossing D2 and pedestrian on the zebra crossing D3 in a driving process; S2: the collected picture data is subjected to image preprocessing; S3: a zebra crossing identification model is trained by using negative samples (D2+D3) and positive samples D1, and a pedestrian identification model on the zebra crossing is trained by using positive samples D3 and negative samples D2; and S4: when a camera in the vehicle does not detect a zebra crossing on the road, only the zebra crossing identification model is kept listening, and when the zebra crossing on the road is detected; listening data is selectively collected to consider a multidimensional interaction state of a vehicle and a target traffic light area road condition, and a vehicle courtesy pedestrian behavior is quantified in a scoring mode.

This disclosure relates to the field of pedestrian recognition technology, and provides a method and apparatus for identifying people traveling together. The method includes: constructing an encoder using multiple fully connected layers; constructing a self-optimizing clustering module based on a clustering algorithm; constructing a people traveling together recognition model using a graph attention network, the encoder, and the self-optimizing clustering module; training the people traveling together recognition model using a reconstruction loss function, a mapping loss function, and a clustering loss; acquiring pedestrian images; extracting pedestrian relationship features and pedestrian attribute features from the pedestrian images; inputting the pedestrian relationship features and pedestrian attribute features into the trained people traveling together recognition model; and outputting the recognition result of whether the pedestrian in the pedestrian image is a person traveling together. By employing the above technical means, the problem of high computational load and low efficiency in existing people traveling together recognition methods is solved.

The invention relates to the field of traffic video data processing, and more specifically relates to a traffic video data processing method and system based on artificial intelligence, a data acquisition module is adopted, the data acquisition module obtains traffic video data recorded by a monitoring module, the monitoring module comprises a monitoring camera, and the monitoring camera is connected with the data acquisition module. The geomagnetic sensor and the meteorological sensor acquire vehicle passing information, and the result output and application module receives processed traffic data, generates a traffic report and provides decision support for an intelligent traffic system. The data acquisition module acquires mass traffic video data, the data preprocessing module performs format standardization, noise removal and image enhancement processing on the data, the problem of processing errors caused by old data is avoided, in addition, the target detection and recognition module recognizes the color of a vehicle based on a deep learning computing platform, and the accuracy of the vehicle color recognition is improved. Identifying the posture of the pedestrian; and the content of the traffic sign is accurately judged, so that the effect of accurate analysis is achieved.

The application relates to a pedestrian attribute recognition method based on deep learning and belongs to the technical field of pedestrian recognition. The recognition method comprises the following steps: constructing a pedestrian attribute recognition network; the pedestrian attribute recognition network comprises a backbone network, a first pooling layer and a full connection layer, and an attention module is introduced between the backbone network and the first pooling layer; training the pedestrian attribute recognition network; and performing pedestrian attribute recognition according to the trained pedestrian attribute recognition network. The pedestrian attribute recognition network for performing pedestrian attribute recognition introduces an attention module between the backbone network and the first pooling layer, the importance of high-level semantic information is distinguished through the attention module, and then the high-level semantic information is purposefully selected and processed by the first pooling layer, the global features corresponding to the high-level semantic information are emphasized, and the recognition accuracy is improved. The pedestrian attribute recognition network structure is simple, the complexity of the network structure is avoided from being greatly increased, and the recognition efficiency is improved.

The embodiment of the application belongs to the field of multi-camera systems, and relates to a multi-target region dynamic vehicle and pedestrian identification method, which comprises the following steps: acquiring image data of a target region collected by multiple cameras; pre-processing the image data to extract features of the target region; performing target detection on the image data to identify vehicle and pedestrian targets in the image data and determine positions and features of the targets; preliminarily positioning each target and predicting a motion trajectory of the target through a Kalman filtering algorithm or a particle filtering algorithm; and performing space-time correlation on the targets identified by the multiple cameras based on a joint data correlation method of the positions, features and motion trajectories of the targets. The application also provides a multi-target region dynamic vehicle and pedestrian identification system. The application improves the coverage and accuracy of target identification.

The invention discloses a vehicle and pedestrian identification method, device and system, and the method comprises the steps: transmitting a control instruction to each signal transmitter, enabling each signal transmitter to transmit a respective detection signal according to the control instruction, and enabling a pedestrian terminal device to generate a feedback signal according to at least one detection signal after receiving the at least one detection signal; acquiring a feedback signal, and performing pedestrian attribute recognition and pedestrian position recognition according to the feedback signal to obtain a recognition result; and determining whether there is a collision risk between the vehicle and the pedestrian according to the identification result, and if yes, controlling the vehicle to execute an anti-collision measure. According to the invention, a parallel pedestrian recognition scheme can be provided when the existing pedestrian recognition scheme of the vehicle fails or even fails, and the pedestrian recognition accuracy of the vehicle is ensured.

The invention discloses a pedestrian re-identification method and system based on vision-language information interaction gain, and relates to the technical field of cross-modal pedestrian identification. The invention provides a data processing optimization and deep interactive fusion scheme which mainly comprises four processing stages: in the data processing stage, a thermodynamic diagram is generated through a saliency network, and non-uniform sampling is realized by combining a grid module to amplify key region features; in the attribute extraction stage, a Ground SAM model is adopted to generate a visual attribute mask, and meanwhile, a GPT2 model is utilized to analyze text attributes; in the feature fusion stage, the visual attribute masks and original image features are subjected to weighted fusion, and cross-modal coding is realized through a CLIP model. In the loss optimization stage, a hierarchical alignment loss function is designed, sample pairs are subdivided according to similarity, and difficult sample alignment is optimized. According to the method, key feature expression, bimodal attribute deep interaction and hierarchical loss optimization are enhanced through saliency sampling, so that the pedestrian re-recognition accuracy in a complex scene is remarkably improved.

A pedestrian detection device for blind spots on buses with risk avoidance function includes a powered telescopic frame, a pedestrian recognition module, an audible and visual alarm, an environmental recognition module, a geographic location recognition module, an emergency avoidance module, and an embedded microcontroller. The powered telescopic frame is installed on the front roof of the bus. The pedestrian recognition module, environmental recognition module, and emergency avoidance module are all mounted on the powered telescopic frame. The audible and visual alarm and the embedded microcontroller are both installed on the interior ceiling of the bus, directly below the powered telescopic frame. The geographic location recognition module is mounted on the embedded microcontroller. The embedded microcontroller is connected to the powered telescopic frame, pedestrian recognition module, audible and visual alarm, environmental recognition module, geographic location recognition module, and emergency avoidance module via signals. This invention can effectively identify pedestrians in the blind spots of buses, alert the driver to driving safety through the audible and visual alarm to avoid traffic accidents, and also has risk avoidance and emergency avoidance functions.

The invention relates to an intelligent control method and system for an electric sliding door, and relates to the technical field of intelligent doors and windows, and the method comprises the steps: obtaining an infrared detection state of a detection area in front of a door; acquiring an infrared identification position when the infrared detection state is consistent with the effective identification state, and calling a preset image shooting device according to the infrared identification position to acquire a region detection image; pixel points corresponding to the infrared identification positions are defined as pedestrian identification points in the area detection image; comparing the area detection image with a preset original image in front of a door to determine distinguishing feature points; when the distinguishing feature points are adjacent to the pedestrian recognition points, the distinguishing feature points are automatically updated to the pedestrian recognition points, and after all the distinguishing feature points are analyzed, the pedestrian passing width is determined according to all the pedestrian recognition points; and according to the pedestrian passing width and the preset surplus width, the reasonable opening width is determined through calculation, and the sliding door is controlled to be movably opened according to the reasonable opening width. The application has the effect of facilitating the use of the sliding door.

The application provides a pedestrian recognition method, system and device based on radio frequency waves and a storage medium. The method comprises the following steps: acquiring a radio frequency thermal map group in a target area, acquiring pedestrian image data captured in the target area according to a preset capturing condition, and extracting pedestrian capturing information from the pedestrian image data; mapping the pedestrian capturing information to the radio frequency thermal map group to cut out a pedestrian radio frequency thermal image group from the radio frequency thermal map group; extracting pedestrian features from the pedestrian radio frequency thermal image group; and comparing the extracted pedestrian features with pedestrian features stored in a database to identify whether the captured pedestrian is a warehouse entry personnel. The method based on RF-PersonReID can effectively manage warehouse entry and exit personnel, complete the statistics of warehouse entry and exit personnel information and warehouse residence time, and effectively solve the visual interference caused by personal dress change or the same dress.

The invention relates to the technical field of traffic control, in particular to an intersection traffic control method and system based on vulnerable group passage optimization. When the traffic light is in a green light state, vehicles pass, pedestrians wait at the two sides of the pedestrian path, the identification module obtains human body characteristics and modes of the pedestrians at the two sides of the path at the moment to identify whether vulnerable pedestrians exist or not, and if the vulnerable pedestrians exist, the identification module marks the vulnerable pedestrians as first vulnerable pedestrians and determines vulnerability labels and levels of the vulnerable pedestrians; and the control module determines a delay coefficient according to the delay coefficient, and prolongs the duration of the traffic light in the red light state next time according to the delay coefficient, so that the first vulnerable pedestrian has more passing duration. Besides, in order to avoid corresponding traffic flow congestion caused by red light delay, cloud and dynamic adjustment can be carried out on flow dredging of the original traffic flow based on flow dredging compensation delay increment of the downstream intersection, a more humanized traffic environment can be constructed in an aging structure, and the method has remarkable social benefits.

A method is provided for determining correctives to statistical identification variances of a biometric identification system using facial and / or pedestrian recognition. The method includes (a) defining a set of categories, each category having at least one corresponding physical appearance characteristic; (b) extracting, for each individual, at least one physical appearance characteristic from at least one image of the individual; (c) assigning, to each individual, at least one category on the basis of the extracted physical appearance characteristic; (d) distributing, for each category, each individual into two groups according to the failure or success of their identification by the system; and (e) calculating, for each category, a number of individuals to be selected in the group corresponding to the success of the identification such that the relative proportions of individuals in the group corresponding to the failure of the identification are substantially equal between all the criteria relating to said category.

The invention relates to the technical field of image recognition, in particular to a pedestrian target recognition method and system based on rapid image enhancement, and the method comprises the following steps: obtaining a pedestrian video; performing rapid enhancement processing on image frames in the pedestrian video to obtain an enhanced pedestrian video; a high-frame-rate pedestrian recognition model is formed by combining a YOLOv5 algorithm, a DeepSORT algorithm and a CBAM attention mechanism, and zoom intersection-union ratio loss is introduced for training; and performing pedestrian identification and tracking on the enhanced pedestrian video through a high-frame-rate pedestrian identification model. According to the method, the image frame detection and enhancement model is integrated as an image preprocessing mechanism, and the improved high-frame-rate pedestrian recognition model is combined, so that the algorithm structure is optimized, the recognition performance is improved from the data source, and efficient and accurate pedestrian multi-target recognition is realized.

The application discloses a neural network model training method and terminal equipment for pedestrian attribute recognition, belongs to the technical field of pedestrian recognition, and can solve the problem of poor effectiveness of a weighting method in existing multi-task model training. The method comprises the following steps: obtaining a forward propagation total loss of a neural network model according to a weighted loss of different attribute training tasks; weighting the gradient norm corresponding to the different attribute training tasks according to the sample number proportion of the different attribute training tasks to update the gradient norm; obtaining the relative back propagation speed of the different attribute training tasks, and obtaining a gradient norm difference total loss according to the updated gradient norm and the relative back propagation speed; updating the neural network model parameters by using the forward propagation total loss, and updating the weight of each training task by using the gradient norm difference total loss; performing normalization processing on the weight of the training task; and repeating the above steps until the training times reach a preset number of times. The application is used for training of a pedestrian attribute recognition model.

The invention relates to the technical field of blind area pedestrian recognition, in particular to an industrial equipment blind area pedestrian recognition and grading early warning method based on AI vision. The method comprises the following steps that a plurality of joint points of the mechanical arm are obtained and segmented into mechanical arm points, and then camera shooting equipment is triggered to obtain surrounding environment image data; pedestrians are recognized according to the surrounding environment image data, the shortest distances between the pedestrians and boundary line segments of the left side, the right side and the rear side of the rectangular range area are calculated, then historical data are obtained, different track route areas of the pedestrians are predicted, the different track route areas of the pedestrians and the rectangular range area of the operation track of the mechanical arm are intersected, and blind area areas are obtained. And then the height value from the point of the mechanical arm to the ground is calculated, it is judged that the pedestrian selects to pass through the blind area below the mechanical arm or bypass the blind area behind the mechanical arm, more calculation resources and monitoring attention can be focused on the blind area where the pedestrian possibly passes through actually through different track route areas of the pedestrian, and the whole monitoring efficiency is improved.

The invention discloses a pedestrian occlusion re-identification method based on attitude guidance, and aims to improve the pedestrian identification accuracy in a complex occlusion environment. Comprising the following steps: inputting a pedestrian image into an attitude estimation network, extracting pedestrian key points by using an HRNet attitude estimation model, and judging a shielding region; through symmetrical structure characteristics, mapping and repairing the characteristics of the unshielded region to the shielded region; constructing a feature restoration loss function to enhance the feature discrimination capability; subdividing the key point region into semantic groups, and performing feature aggregation based on a semantic relationship between the regions; introducing a global enhancement module to realize global feature extraction in combination with multi-scale pooling and attitude guidance; and finally fusing local and global features to carry out pedestrian representation. The method effectively solves the problems of feature missing and interference caused by shielding, has good recognition performance and universality, and is suitable for shielding pedestrian recognition tasks in actual scenes such as video monitoring and dangerous areas.