43 results about "Visual projection" patented technology

The invention provides a construction and training method and device for an eye fundus image multi-task model, and belongs to the field of image processing, and the method comprises the steps: S1, collecting and sorting a public eye fundus image data set, constructing an image text pair according to a real label, and carrying out the two-stage training of a multi-modal large language model, the multi-mode large language model comprises a visual encoder, a visual projector and a large language model; s2, inputting the image data # imgabs0 # into a visual encoder in the trained multi-modal large language model to obtain an enhanced visual feature # imgabs1 #, and extracting a visual feature # imgabs3 # from the # imgabs2 # through a visual projector; and S3, embedding the text input # imgabs4 # to obtain a text feature # imgabs5 #, splicing the text feature # imgabs5 # with the visual feature # imgabs6 #, and inputting the spliced text feature # imgabs5 # and the visual feature # imgabs6 # into a large language model to generate a prediction text A. According to the method, a wide range of fundus image data is collected for training, multilevel lesion features in the fundus image are fully utilized, and the performance of the model for executing a fundus disease auxiliary diagnosis task can be effectively improved.

The invention provides a multi-modal fusion and semantic enhancement train positioning method and system, and belongs to the technical field of rail transit, and the method comprises the steps: carrying out the time-space alignment of data, obtaining a dense point cloud, constructing a dense semantic point cloud, and dynamically estimating the confidence coefficient weight of each type of sensors; a set residual error and a Manhattan structure constraint residual error of a plane are constructed, laser radar point cloud parameters are obtained, and visual projection constraints are constructed at the same time; constructing a comprehensive degradation scoring function to carry out degradation judgment on the current environment; when the degradation result is yes, introducing a structure and motion information independent of an external environment, maintaining trajectory estimation, and constructing a compensation constraint; introducing a prior semantic constraint and a large model semantic factor constraint; and constructing a global optimization objective function, dynamically adjusting the weight of each modal factor, obtaining an optimal estimation state, and outputting a high-precision train positioning result. According to the method, high-precision and robust track estimation in an extreme scene is realized, so that the continuity, safety and intelligence of train positioning are guaranteed.

A pallet loading system may comprise a processor, and memory with instructions stored thereon that, when executed by the processor, cause the processor to receive package loading data, the package loading data including characteristics of a package to be placed on the pallet and characteristics of a loading operator. A placement location for the package on the pallet may be determined using an artificial intelligence (AI) or machine learning (ML) algorithm based on the characteristics of the package and the characteristics of the loading operator and a visual marking (such as a visual projection) may be displayed at the placement location. The system may output an instruction to the loading operator to place the package at the displayed placement location.

The embodiment of the invention provides a cluster collaborative navigation method, a control system and a storage medium. The method comprises but is not limited to the technical field of navigation. The method comprises the following steps: in an upper layer module of a gene regulation and control network, determining a form boundary curve according to first position information of first execution equipment, second position information of second execution equipment and third position information of an obstacle; in a lower layer module of the gene regulation and control network, determining a tangential propulsion speed component and an offset correction speed component according to the form boundary curve and the first position information; and in a lower layer module of the gene regulation and control network, according to a visual adjacent distance regulation speed component, a tangential propulsion speed component and an offset correction speed component of the visual projection field, determining a target linear speed and an angle parameter. According to the embodiment of the invention, collaborative navigation and dynamic form maintenance of the cluster system can be realized.

An electronic device, computer program product, and method provide autonomous projecting of a selected content to a second display. The device is configured to, in response to a trigger to transmit first content of the electronic device to a connected second electronic device for presenting on a second display: (i) determine whether second display content is currently being shared with at least one third electronic device; and (ii) in response to determining that the second display content is currently being shared with at least one third electronic device: (a) withhold automatic rendering of the first content on the second display; and (b) generate and output a notification presented on at least the first display, the notification informing a user of at least one of the electronic device and the second electronic device that the content presented on the second display is being shared.

The invention discloses a cross-modal data retrieval method, system and equipment based on a multi-modal knowledge graph, and the method comprises the steps: carrying out the feature decoupling based on a visual feature vector and a structural feature vector, and determining a target visual projection vector; constructing a path constraint contrast learning loss function based on the target visual projection vector and the text feature vector; constructing a multi-modal knowledge graph, and mining effective paths among entities in the multi-modal knowledge graph; coding the effective path into a path coding feature, and constructing a multi-scale path perception rejection loss function based on the path coding feature; joint optimization is carried out on the path constraint contrast learning loss function and the multi-scale path perception rejection loss function, and a target path coding feature subset is determined; fusing the text feature vector, the target visual projection vector and the target path coding feature subset to obtain a joint vector; and performing cross-modal data retrieval based on the joint vector to obtain a retrieval result. The image-text retrieval method and device can improve the accuracy of image-text retrieval.

The invention discloses a curved surface deviation detection method based on virtual-real superposition and multi-user cooperation, and relates to the technical field of deviation detection. Comprising the steps of calculating point cloud data to obtain a curvature feature matrix; the reverse model data and the real object point cloud data are aligned, curved surface deviation detection operation is executed, a deviation field is coded, and preliminary visual display is carried out in a VR / AR environment; a plurality of users carry out collaborative recheck based on the deviation field, dynamic visual projection is carried out on annotation space coordinates according to curved surface curvature features, meanwhile, the annotation space coordinates, user gesture tracks and text keywords are jointly coded, a two-dimensional R tree annotation index is established, and annotation data and a deviation field result are spatially correlated; and dynamically updating the visual content and correcting the deviation according to the spatial association. According to the method, the problems of inaccurate alignment, low rechecking efficiency, difficulty in coverage of a blind area and the like in curved surface deviation detection are effectively solved, and the accuracy, interpretability and team cooperation efficiency of a detection result are improved.

The invention discloses an aircraft cluster control method and system based on a bionic visual projection field and identity balance, and belongs to the technical field of unmanned aerial vehicle clusters. Aiming at the problems that a traditional method depends on global planning and is poor in robustness in a communication limited and dynamic complex scene, two innovations are provided: one is a bionic visual projection field, and a dynamic environment model is generated through local sensing, so that an unmanned aerial vehicle autonomously avoids obstacles and cooperatively moves; and 2, a dynamic identity balancing strategy is adopted, leader or follower roles are allocated to cluster members according to tasks and positions, role adaptive switching is supported, and task fault tolerance is enhanced. The system is integrated with a multi-module cooperation mechanism, efficient cluster crossing is achieved in complex obstacle environments such as ruins and forests through local information interaction and rule simplification, and motion consistency and safety are guaranteed. The method is low in calculation requirement, high in adaptability and suitable for communication limited scenes such as post-disaster rescue, and the autonomous cooperation capability of the cluster is remarkably improved.

The invention relates to the technical field of medical instruments and clinical nursing, in particular to a radial artery puncture navigation system based on multi-modal sensing and digital projection, which comprises the following steps: a multi-modal blood vessel signal acquisition module, which is used for acquiring a blood vessel pulsation signal on the wrist of a patient through a photoplethysmography sensor array, the strongest pulsation line is positioned through the pressure sensor array, and the blood vessel depth and artery and vein distinguishing information are obtained through the bioelectrical impedance sensor array. According to the invention, through multi-modal blood vessel signal acquisition, blood vessel dynamic deformation modeling, personalized digital twinning construction and dynamic updating, projection path generation and control, and projection output and real-time error compensation, real-time visual projection of a blood vessel center line and local deformation is realized; therefore, the problem that puncture positioning is not accurate due to the fact that dynamic changes of blood vessels cannot be accurately reflected due to the fact that most traditional radial artery puncture adopts a hand feeling positioning method is solved.

The invention discloses a movable road disaster night visual projection early warning method and device. The device comprises an electric carrying platform, a fixing device is arranged on the electric carrying platform, a telescopic device is arranged on the fixing device, a rotating device is arranged at the top end of the telescopic device, and a projection device is arranged on the rotating device; the telescopic device and the rotating device are combined for use, so that a projection picture is allowed to be accurately projected to a preset area, the projection accuracy is ensured, and the overall early warning efficiency of the early warning device is remarkably improved; the projection device enables an early warning sign and a signal to be clearly projected to the ground at night, so that a running vehicle can more strikingly receive early warning information, a driver is prompted to take prevention measures such as deceleration or parking, potential traffic accidents are effectively avoided, and the driving safety is improved. The technical problem that the early warning effect of an early warning device in the prior art is not obvious is solved.

The invention relates to intelligent sensing for tray loading and unloading. The tray loading system may include a processor and a memory having instructions stored thereon that, when executed by the processor, cause the processor to receive parcel loading data that includes characteristics of a parcel to be placed on the tray and characteristics of a loading operator. A placement location of the parcel on the tray may be determined using an artificial intelligence (AI) or machine learning (ML) algorithm based on characteristics of the parcel and characteristics of the loading operator, and a visual marker, such as a visual projection, may be displayed at the placement location. The system may output instructions to the loading operator to place the parcel at the displayed placement location.

The invention relates to an indoor three-dimensional reconstruction method based on SLAM and 3DGS technologies, and the method comprises the following steps: S1, obtaining indoor high-definition image information through a binocular camera, and transmitting the indoor high-definition image information to an ORB-SLAM3 module in real time; s2, performing indoor image downsampling, extracting and matching ORB feature points, performing camera pose estimation and other processing on the ORB-SLAM3; s3, seamlessly transmitting three-dimensional sparse data generated by the ORB-SLAN3 module into a 3DGS module, initializing SFM point cloud into a three-dimensional Gaussian ellipsoid set, visually projecting the three-dimensional Gaussian ellipsoid set to a rasterization plane under a world coordinate system, and performing rapid micro-rasterization rendering and the like; and S4, displaying a reconstruction result of the rendered three-dimensional model, and optimizing the model. According to the invention, portable room structure information, accurate space dimension measurement and flexible furniture layout adjustment can be provided for a user, and the convenience and practicability of home design are remarkably improved.

The invention relates to the technical field of remote sensing vision, and particularly discloses a remote sensing multi-modal reasoning method and system based on a geospatial thinking chain, and the method comprises the steps: constructing a geospatial thinking chain data set which comprises structured reasoning data organized according to a planning-positioning-synthesizing three-section cognitive architecture, the association module is used for establishing verifiable association between visual evidence in a remote sensing image and a text conclusion; constructing a remote sensing visual language basic model, wherein the model comprises a visual encoder, a language decoder and a visual projection layer connecting the visual encoder and the language decoder; according to the method, remote sensing analysis is modeled into a verifiable multi-step reasoning process, so that the model provides a verifiable analysis track while outputting a final answer, the problem that output cannot be verified due to traditional end-to-end training is solved, the performance of a complex analysis task is remarkably improved, and meanwhile, the accuracy of the result is improved. And the conversion from opaque perception to structured and verifiable reasoning is realized.

The invention discloses a robot interaction intention projection method based on environmental perception, and belongs to the technical field of robot detection and human-computer interaction. The method comprises the following steps: S1, acquiring environment sensing data, robot state data and projection effect influence factors; s2, carrying out intelligent processing on the environment perception data, the robot state data and the projection effect influence factors, and generating an optimal projection decision; and S3, projecting a visual symbol according to the optimal projection decision, and automatically adjusting the projection effect based on the projection effect influence factors. According to the robot interaction intention projection method based on environmental perception, dynamic intention visual projection is generated by fusing the internal intention and the external environment information of the robot, and adaptive adjustment is performed on the visual projection image, so that efficient, safe and clear man-machine interaction is realized.

The invention provides a lip language recognition model construction method fusing external information, and the method comprises the steps: constructing an initial lip language recognition model which comprises a visual modal data processing module, a pre-trained visual encoder, a visual projector, a text embedding module and a first large language model; a data set is constructed, and the initial data set takes each visual sequence image and corresponding external information as samples and takes a real speaking text and a correct reasoning process description text as labels; and based on the constructed training set, taking the visual sequence image and the corresponding external information as input, and taking a predicted speaking text and an actual reasoning process description text as output, carrying out iterative training on the initial lip language recognition model until convergence so as to obtain a final lip language recognition model, in the training process, a preset cross entropy loss function is adopted to calculate loss, and parameters of the first large language model, the visual projector and the pre-trained visual encoder are updated based on the calculated loss.

The present invention discloses an ultrasonic localization imaging detection method, device and medium for defects on the outer surface of a storage tank, which relates to the technical field of localization detection. By acquiring the probe movement trajectory and ultrasonic echo signals, the two-dimensional position data of the probe after data alignment is obtained, and further three-dimensional coordinates are obtained. Based on the three-dimensional coordinates, a three-dimensional projection image of the defect is generated to provide a three-dimensional visual projection of the defect, intuitively display the defect characteristics, and assist subsequent intelligent recognition and diagnosis. Then, feature data is extracted from the three-dimensional projection image and ultrasonic echo signals to construct a defect classification model and train the defect classification model; based on the trained defect classification model, the probability distribution of the defect type is output; a defect diagnosis report is generated based on the probability distribution of the defect type, which can realize the intuitive visualization of the defect position, distribution and characteristics, improve the detection efficiency, reduce missed detections and false detections, and realize the automatic storage of detection data.

The application discloses an intelligent visual projection method and system suitable for optical imaging, wherein the system comprises a visual stimulation control module for setting a visual stimulation control signal; an image output module for generating full-spectrum visual stimulation and light intensity according to the visual stimulation control signal; a spectrum switching module for receiving the full-spectrum visual stimulation and the light intensity and outputting first specific spectrum visual stimulation; an infrared light source module for outputting infrared spectrum visual stimulation; a spectrum processing module for filtering preset green spectrum signals of the first specific spectrum visual stimulation and integrating the infrared spectrum visual stimulation to output second specific spectrum visual stimulation; and an image stimulation focusing module for fine-tuning the projection area and light intensity of the second specific spectrum visual stimulation and projecting the second specific spectrum visual stimulation to a visual tissue region. The application can solve the problem that the prior art cannot meet the research demand of visual neural optical signals and has the characteristics of intelligence, automation, precision and no optical pollution.

The invention relates to a method for enabling an intelligent robot with a body to look around an aerial view and a storage medium, and the method comprises the steps: reading a feature A4, a feature A3 and a feature A2 through a position encoder, reading internal parameter matrixes of n cameras and an external parameter matrix relative to a main body, and obtaining corresponding BEV information through visual projection transformation; performing feature extraction on a result generated by the position encoder by using two dual-parameter convolution residual modules, and generating corresponding object information and 3D position information according to an extracted feature result; carrying out loss calculation on the corresponding object information and 3D position information and a true value, and carrying out back propagation operation through re-parameter convolution before re-parameter; and carrying out re-parameterization on all the re-parameter convolution to obtain a model for finally generating BEV features. According to the method, coding calculation does not need to be carried out through a transformer technology, and due to the fact that no transformer exists, the method can be deployed on all end-side devices rapidly and universally.

A carwash system and method of operating the carwash system for washing a vehicle is provided. The carwash system comprises a carwash bay, a visual projection assembly to project images and video, an audio output system to emit sound, a user interface having a content menu with a plurality of content selections, a washing unit to conduct a cleaning cycle, and a control unit. The control unit is operatively connected to the visual projection assembly, audio output system, user interface, and washing unit. The control unit synchronizes the visual projection assembly and audio output system with the cleaning cycle conducted by the washing unit based on a specific content selection received from a user input on the user interface.

The invention relates to the technical field of data processing, in particular to a surgical operation information data management system and method based on the mobile internet, and the method comprises the following steps: obtaining first surgical operation information data through a medical high-definition camera, a sound recorder and medical equipment; performing data self-adaption on the first surgical operation information data by utilizing an artificial intelligence algorithm to generate a second surgical operation information data set; performing visual projection on the second surgical operation information data set by using a matrix decomposition method to generate a surgical operation characteristic matrix projection drawing; performing data visualization processing on the surgical operation information matrix decomposition graph by using a deep learning algorithm to generate a surgical operation feature interactive view; performing homomorphic encryption on the surgical operation convolutional feature model by using a homomorphic encryption algorithm; uploading data of the surgical operation homomorphic encryption model to a surgical operation information data management system by using a 5G technology; according to the invention, accurate and orderly management of surgical operation information data is realized.

An electronic device, computer program product, and method provide autonomous projection of a selected viewable content to a second display. The device is configured to, in response to receiving a trigger to project selected viewable content to a selected second display: determine, by evaluating location information of the electronic device and location information of the selected second display, whether the selected second display is located within an acceptable range of and in a line-of-sight of a user of the electronic device; and in response to confirming that the second electronic device is within the acceptable range and in the line-of-sight, initiate a casting of at least the selected viewable content to the selected second display. The device is configured to withhold casting to the selected second display, pending confirmation by a user that the selected second display is the intended device to cast the selected viewable content.

The positioning method, device, medium and equipment for actively adjusting the field of view of visual inertial angle fusion are provided, the feature point relative inertial IMU external parameter rotation matrix in each frame target image of binocular camera calibration is calculated according to current angle information; based on the three-dimensional motion of the target image and the parameters measured by the inertial IMU, a local coordinate system with the initial position of the target image as the origin is established, the corresponding visual projection error vector and the IMU pre-integration residual error vector are calculated based on the feature points and the external parameter rotation matrix of the target image respectively, the factor graph is constructed according to the visual projection error vector and the IMU pre-integration residual error vector, and the state quantity in the factor graph is solved based on the sliding window, and the unmanned aerial vehicle position and the high-frequency unmanned aerial vehicle body pose are determined based on the state quantity; according to the distribution of the feature points in the camera field of view, the camera direction is automatically adjusted, and the positioning method for actively adjusting the field of view of visual inertial angle fusion is solved.

The invention belongs to the technical field of posture capture, and provides a space-time Transform-based multi-modal human body posture estimation model training method, which comprises the following steps that: a multi-modal human body posture estimation network comprises a shallow layer space-time Transform cascade network and a deep layer space-time Transform cascade network; obtaining a sample pair set; iterative training is carried out on the text feature extraction network and the multi-modal human body posture estimation network based on the sample pair set, and comparative learning is carried out on global posture features obtained by the shallow space-time Transform cascade network and global text features obtained by the text feature extraction network; optimizing network parameters of the shallow space-time Transform cascade network and the text feature extraction network based on comparison loss, and optimizing network parameters of the visual projection layer, the deep space-time Transform cascade network and the attitude output layer based on joint position errors; the invention further discloses a space-time Transform-based multi-modal human body posture estimation method, a computer program product and electronic equipment. The posture estimation accuracy is improved by the space-time Transform-based multi-modal human body posture estimation method and the space-time Transform-based multi-modal human body posture estimation device.

This utility model relates to a tool cutting edge image measurement and chamfering machining machine, including a V-shaped clamp, a clamping seat, a support base, a hand rest, a hand rest adjustment seat, an imaging module, an X-axis moving module, a Y-axis moving module, a Z-axis moving module, and a display. The V-shaped clamp is equipped with a clamping seat for stabilizing the tool. A hand rest is provided on one side of the V-shaped clamp and is mounted on the hand rest adjustment seat. An imaging module for visual projection and detection of the clamping process is located above the V-shaped clamp. The imaging module is mounted and fixed by the X-axis moving module, the Y-axis moving module, and the Z-axis moving module. The imaging module is connected to the display, which is placed on the side of the support base via a display bracket. An angle adjustment component for adjusting the angle of the V-shaped clamp is provided on the support base, and the V-shaped clamp is mounted on the angle adjustment component. This utility model can improve the accuracy of the angle, thereby achieving high-quality machining.

The embodiment of the invention provides a visual inertia angle fusion positioning method and device for actively adjusting a field of view, a medium and equipment. The method comprises the following steps: calculating an external parameter rotation matrix of feature points in each frame of target image calibrated by a binocular camera relative to an inertia IMU according to current angle information; the method comprises the following steps: establishing a local coordinate system taking an initial position of a target image as an original point based on three-dimensional motion of the target image and parameters measured by an inertial IMU, and respectively calculating a corresponding visual projection error vector and an IMU pre-integration residual vector based on a feature point rotation matrix and an external parameter rotation matrix of the target image; constructing a factor graph according to the visual projection error vector and the IMU pre-integration residual vector, solving the state quantity in the factor graph based on a sliding window, and determining the position of the unmanned aerial vehicle and the high-frequency body pose of the unmanned aerial vehicle based on the state quantity; the orientation of the camera is automatically adjusted according to the distribution of the feature points in the field of view of the camera, and the positioning method for actively adjusting visual inertia angle fusion of the field of view is achieved.

This application provides a method, device, medium, program product, and terminal for autonomous driving eye monitoring based on facial projection transformation and visual projection. By acquiring driver image data and driving operation data in real time, and combining facial projection transformation and 3D model construction, the accuracy and reliability of driver state monitoring are significantly improved. It effectively corrects image displacement and distortion caused by driver head movement and posture changes, ensuring accurate facial information recognition. The coarse and fine eye positioning coordinates generated based on facial projection enable accurate monitoring of eye opening and closing in complex situations, thereby identifying driver fatigue. Furthermore, by constructing a 3D head model, the driver's posture at different angles can be analyzed, further enhancing monitoring accuracy. Intelligent warning commands generated by combining driving operation data and eye state can promptly alert the driver to potential safety risks, thereby effectively improving driving safety.

The application discloses a cross-modal data retrieval method, system and device based on a multi-modal knowledge graph. The method decouples features based on a visual feature vector and a structural feature vector, and determines a target visual projection vector. A path constraint contrast learning loss function is constructed based on the target visual projection vector and a text feature vector. A multi-modal knowledge graph is constructed, and effective paths between entities in the multi-modal knowledge graph are mined. The effective paths are encoded into path encoding features, and a multi-scale path perception repulsion loss function is constructed based on the path encoding features. The path constraint contrast learning loss function and the multi-scale path perception repulsion loss function are jointly optimized to determine a target path encoding feature subset. The text feature vector, the target visual projection vector and the target path encoding feature subset are fused to obtain a joint vector. Cross-modal data retrieval is performed based on the joint vector to obtain a retrieval result. The application can improve the accuracy of image-text retrieval.

The invention belongs to the technical field of intelligent measurement and machine vision, and provides a container top surface inclination monitoring method based on a camera visual projection proportion, which realizes lightweight inclination angle calculation without depending on a large-scale training sample and a deep learning model by using top surface projection proportion characteristics under a camera visual angle. Compared with side face inclination measurement in the prior art, the top face inclination measurement method is innovatively provided, accurate monitoring of container top face inclination is achieved through the camera visual projection proportion, and therefore batch recognition and calculation of the top face inclination states of a plurality of containers can be achieved in the mode that the whole container area image is collected; and the monitoring efficiency is greatly improved.

The invention provides an automatic driving data normalization method and system based on topological structure recognition and parameter semantic analysis coupling, and is applied to the technical field of data processing. According to the method, data topological structure reconstruction is completed through directory depth, extended name entropy and regular feature clustering, a logic index is established, and a data role is anchored; extracting calibration parameter metadata, and identifying parameter features and generating analysis codes and operators through large model agent small sample thinking chain reasoning; then, a data processing assembly line is built, an analytic operator is dynamically injected to achieve algorithm coupling, automatic verification is completed through a visual projection alignment effect, relevant rules, logics and operators are iteratively optimized according to a verification result, closed-loop tuning is conducted, and finally, whole-process normalization is conducted on original data, so that the data processing efficiency is improved. And outputting a standardized automatic driving data set with a unified structure and clear parameter semantics.

1. Name of the product in this design: Exercise Bike (Audio-Visual Projection Version). 2. Purpose of this design: a stationary bike for fitness. 3. The key point of the design of this product lies in its shape. 4. The picture or photo that best illustrates the design points: Stereoscopic drawing 1.