53 results about "Machine recognition" patented technology

A system and methods are provided for area surveillance. The system is configured to acquire multiple video streams, each video stream associated with a geo-location indicator of the given device and to apply machine recognition by multiple specialized agents to images of the multiple respective video streams to generate indications of potential threats. Each Al agent is further configured with alternative early inference options. The one or more indications are then processed by a large language model (ELM) configured to determine an inference plan for processing the multiple respective video streams by an alternate set of Al agents, including alternate early inference options, to improve accuracy and efficiency of the processing by the Al agents, which are then triggered to operate according to the inference plan to generate a revised indication of a potential threat and to issue a corresponding alert.

The invention discloses a method and system for processing a circular weld joint based on point cloud data, and belongs to the technical field of machine identification of weld joints, and the method comprises the following steps: collecting the point cloud data of the surface of a workpiece through a visual sensor; preprocessing the collected point cloud data; calculating a normal vector of each point in the point cloud data; segmenting the workpiece surface point cloud and the reflective interference points based on a normal vector, and filtering the interference points deviating from a preset range in the normal direction to obtain a target point cloud and a reference point cloud; projecting the target point cloud to a reference plane obtained by fitting the reference point cloud, generating a target projection point cloud, detecting boundary points of the target projection point cloud, screening candidate weld joints based on distance distribution from the boundary points to the center of mass, and extracting weld information. According to the method, the workpiece made of the material with high light reflection is treated, the finally obtained result can still keep good precision, and the quality of the workpiece is improved.

The invention relates to the technical field of intelligent sorting, in particular to a full-automatic rice impurity removing method and device based on machine recognition, and the method comprises the following steps: extracting rice particle reflection intensity and analyzing a boundary trajectory, screening a mutation region to generate a recognition section, judging the consistency of recognition quality and weight error directions, and generating an error sample set. And performing statistics on disturbance frequency to generate a deformation mark, comparing numbers, extracting coordinates to generate a list, calculating a response difference, and outputting a rejection identification sequence. According to the invention, through comparison of reflection boundary trajectory abrupt change, focusing of a spectrum abnormal section, improvement of particle difference positioning precision, combination of consistency judgment of image identification errors and weight deviation, enhancement of abnormal particle screening accuracy, tracking of contour disturbance frequency identification of deformation behaviors, and enhancement of dynamic feature identification, the accuracy of abnormal particle identification is improved. An operation target is positioned based on multi-source feature intersection, orbital offset and air injection response time sequence calibration are combined, it is guaranteed that removal actions are coordinated and consistent, and efficient recognition and accurate removal of the mixed rice particles are achieved.

The invention discloses a welding seam track identification method for impeller wrap angle welding based on point cloud data, and belongs to the field of machine welding seam identification, and the method comprises the following steps: carrying out filtering, down-sampling and clustering on original scanning point cloud; classifying the point cloud data through a region growing clustering algorithm; determining an optimal interception plane based on principal component analysis, and performing curve fitting on section data to obtain a preliminary main welding seam track; the main welding seam track is projected to a local approximate plane in a segmented mode, and accurate extraction of the main welding seam track is completed; and performing plane fitting and thickness calculation on the wrap angle surface, solving an intersection line in combination with an adjacent local approximate plane, calculating an intersection point of the intersection line and the main welding seam track, and finally generating a wrap angle welding seam track in combination with the direction of the intersection line and the wrap angle thickness. Through cooperative calculation of local plane projection and spatial intersection line solving, the whole process operation from recognition, planning to welding can be automatically completed, and unmanned intelligent welding in the true sense is achieved.

The invention relates to a defect analysis and process optimization method based on artificial intelligence multi-mode fusion, and the method comprises the steps: respectively inputting virtual data of a chip or device design layout, an expected process flow and parameters, and an expected electrical performance curve into an image encoder, a process encoder, and an electrical encoder, extracting a corresponding feature vector; performing cross-modal information interaction on the corresponding feature vectors according to a cross attention fusion mechanism; determining a representation image, a defect category, a root cause category and a process parameter correction suggestion; and based on a manufacturing execution system (MES), an electronic design automation (EDA) tool or an equipment interface, process parameters are corrected, and a closed-loop control result from defect analysis to process optimization is obtained. Therefore, the problems of single data dimension, mutual coupling of various defects, influence of background noise, limited machine identification accuracy, overlarge workload of artificial expert judgment and low efficiency are solved.

A computer-implemented method in a hardware-secured mission-critical system governs text outputs produced by a transformer-based language model. A first transformer-based model generates candidate text and machine-identified propositions. A timing subsystem synchronized to PTP, GPS, or IMU clocks timestamps propositions and retrieved reports, logs, and sensor products, and admits only evidence within a configured skew window. Each proposition is segmented into semantic units under a constrained domain grammar that distinguishes subjects, predicates, modifiers, and domain-specific clauses. A second, architecturally isolated transformer-based model receives adversarial prompts built from propositions and summaries of admitted evidence and outputs structured challenge records with challenge types and severities. From evidence associations and challenge records the system constructs a semantic support mask and generates a conservative rewrite that preserves, in order, only tokens mapped to units that satisfy governance thresholds, introduces no new tokens, and passes a grammar-based integrity check before release as governed text output.

The invention discloses a road quality automatic intelligent detection method and system based on machine recognition, and relates to the technical field of road detection.The detection method comprises the steps that after gray normalization operation is executed according to a captured road surface original image, a standard image matrix is constructed; after directional gray scale difference calculation is executed, an abnormal response matrix is constructed, candidate connected regions are screened out through an eight-neighborhood connectivity judgment method, comparative analysis is carried out on the region areas of all the candidate connected regions, abnormal crack regions are screened out, and crack region feature data are constructed; in combination with an arc tangent function, analyzing the surface structure response degree of each crack communication region, determining the structure response coefficient of each crack communication region, and in combination with the extracted abnormal texture response value of each candidate pixel point in each crack communication region, determining a pavement abnormal strength index; whether the current pavement structure is abnormal or not is judged, and a corresponding grade quality detection result is sent out.

A method in a computing device includes: capturing, via a sensor, an image of an indicium; determining a confidence level associated with machine recognition of the indicium; based on the confidence level, selecting a feedback attribute; prior to outputting content encoded in the indicium, generating, via an output device, a feedback signal having the feedback attribute, wherein the attribute is configured to indicate a likelihood of success of a decode event to decode the content encoded in the indicium.

The utility model discloses carton feeding and conveying equipment based on machine recognition, and relates to the technical field of carton feeding, the carton feeding and conveying equipment comprises a conveying table, one end of the conveying table is provided with a sorting mechanism, the sorting mechanism comprises a second electric driving rod, the top end of the second electric driving rod is fixedly connected with a fixing seat, the bottom end of the fixing seat is fixedly connected with a second connecting rod, and the second connecting rod is fixedly connected with a conveying belt. A first rotating rod is fixedly connected to the top end of the second connecting rod, a bottom plate is movably connected to the outer side of the first rotating rod, a second rotating rod is movably connected to the top end of the bottom plate, a first rotating gear is fixedly connected to the outer side of the first rotating rod, and the bottom plate is located below the first rotating gear. According to the carton feeding and conveying equipment based on machine recognition, carton conveying is effectively achieved, the carton feeding efficiency is improved after feeding is completed, automation of carton feeding is achieved, and continuity of carton feeding is kept.

The application provides a signal plane diagram signal machine recognition method and system based on deep learning, which comprises the following steps: obtaining a signal plane diagram to be recognized; using a signal machine recognition network to recognize the signal machine in the signal plane diagram; and the signal machine recognition network is a neural network based on YOLOv5-s. The signal plane diagram signal machine recognition method and system based on deep learning take YOLOv5-s network as the basis, propose a Signal-Net recognition network suitable for signal plane diagrams, and can accurately detect the signal machine target in a high-resolution drawing. While ensuring the recognition accuracy of the network for the signal machine, the model parameter amount is moderate, and the real-time operation requirement on the edge computing device is met.

The invention relates to a defect analysis and process optimization method based on artificial intelligence multi-modal fusion, and the method comprises the steps: respectively inputting a representation image, an electrical curve and process parameters in a reasoning stage into an image encoder, an electrical encoder and a process encoder, so as to extract corresponding feature vectors; performing cross-modal information interaction on the corresponding feature vectors according to a cross attention fusion mechanism, and quantifying the responsibility degree of the process parameters to obtain fusion features; determining a defect category, a defect root cause, a process parameter correction amount and a process parameter correction suggestion; and based on a manufacturing execution system (MES), an electronic design automation (EDA) tool or an equipment interface, process parameters are corrected, and a closed-loop control result from defect analysis to process optimization is obtained. Therefore, the problems of single data dimension, mutual coupling of various defects, influence of background noise, limited machine identification accuracy, overlarge workload of artificial expert judgment and low efficiency are solved.

The application discloses a kind of stealthy photographing behavior detection method, device, electronic equipment and storage medium, the present application first collects the pedestrian image in target area, then, through machine recognition technology, to identify the arm posture of each pedestrian in pedestrian image, and determine the stealthy photographing suspect object according to arm posture;Then, again through the hand-held object type of identifying the stealthy photographing suspect object, and the dress type of pedestrian in its corresponding stealthy photographing image area, to finally determine whether the stealthy photographing suspect object exists stealthy photographing behavior;Finally, after determining that there is stealthy photographing behavior, then the interception of stealthy photographing image can be carried out, and sent to monitoring terminal, to realize the retention of evidence and secondary detection of stealthy photographing behavior;Therefore, the present application uses machine recognition technology to realize stealthy photographing behavior detection, compared with artificial patrol, efficiency is higher, detection accuracy is higher, effectiveness is stronger, can produce greater deterrent to lawbreaker, can fully guarantee the privacy security of passenger.

This invention relates to the field of machine recognition technology, and particularly to a multi-task-based method and device for recognizing multi-pointer instruments in substations. This invention proposes a highly efficient multi-task instrument recognition network that can simultaneously handle three key tasks: target detection, instrument centering, and pointer segmentation of substation instruments. This significantly improves the detection speed while greatly enhancing the accuracy of instrument pointer detection. The algorithm proposed in this invention has higher universality, enabling the recognition of multiple instruments and multiple pointers, effectively solving the problem of using a different recognition algorithm for each instrument. The dataset for this invention is entirely derived from actual substations. Even in outdoor scenes and under conditions of poor image quality, it achieves accurate recognition of single and double-needle instruments in substations, including instruments in switchgear, discharge counters, main transformer temperature instruments, current transformers, voltage transformers, and surge arresters, with an accuracy approaching 100%. The algorithm is flexible in configuration, easy to set up, and reliable in operation.

The invention discloses a method for identifying an undamaged area of an optical transparent protective material, which effectively avoids transmission white edge errors and machine identification misleading caused by refractive index differences of transparent materials through multi-angle image acquisition and registration, and improves the identification accuracy of the undamaged area of the optical transparent protective material by combining morphological reconstruction and hole filling technologies. The accurate segmentation of the damaged area and the non-damaged area is realized; according to the method, the multi-view features are integrated by adopting the Poisson fusion algorithm, the accuracy and global consistency of damage identification are remarkably improved, meanwhile, the residual visible area of the transparent protective material can be quickly and reliably evaluated through pixel-level quantitative calculation, and objective maintenance and replacement basis is provided for a high-precision application scene of the optical transparent protective material.

The present application belongs to the technical field of data extraction, and discloses an information extraction method and device, an electronic device and a storage medium. The present application adopts a machine recognition and processing mode to automatically extract the content in pictures and documents, and performs error correction processing on the extracted content after extraction, so as to obtain an information keyword set with high accuracy. Compared with manual extraction, the present method is not only efficient, but also reduces labor costs, and is suitable for large-scale application and promotion in the field of information extraction technology.

An example system includes an analysis engine to detect a first set of significant portions of a target. The system includes an event engine to detect a set of potential events in a physiological signal and identify a second set of significant portions of the target based on the set of potential events. The system includes a comparison engine to identify a difference between the first set of significant portions and the second set of significant portions.

This application discloses a batch calibration method for multi-workpiece weld seam scanning trajectories, belonging to the field of machine recognition weld seam technology. The method includes the following steps: acquiring the original point cloud data of all workpieces through a line laser sensor; processing the original point cloud to obtain workpiece point cloud data, and performing filtering and downsampling processing; performing coarse registration to obtain the initial rigid body transformation relationship; performing fine registration to obtain the optimal rigid body transformation matrix; mapping the predefined weld seam scanning path in the template coordinate system to the actual workpiece coordinate system to generate a working trajectory that can drive the execution device; immediately starting the welding operation process of the first workpiece after generating the scanning path of the first workpiece; simultaneously performing the registration and path calibration process sequentially on the subsequent workpiece point clouds, realizing the parallelization of processing and calculation, and effectively solving the core pain point of batch calibration of weld seam scanning paths for similar workpieces.

An operation management device includes a machine tool detection unit which detects industrial machines from videos from cameras based on registered feature points of the industrial machines, a varying point detection unit which detects, from the videos, time-varying feature points of the detected industrial machines and change times, and an individual machine recognition unit which recognizes the individual industrial machines based on the time-varying feature points and change times. The operation management device also includes a camera parameter calculation unit which calculates camera parameters of each camera, a machine distance calculation unit which calculates the distances between the feature points and the cameras based on the camera parameters, and a layout diagram creation unit which creates a layout diagram of icons of the industrial machines from the calculated distances and locations and three-dimensional coordinates of the center of the lens of each camera.

A method in a computing device includes capturing, via a sensor, an image of a marker; determining a confidence level associated with the machine identification of the marker; selecting a feedback attribute based on the confidence level; prior to outputting the content encoded in the flag, a feedback signal having a feedback attribute is generated via the output device, where the attribute is configured to indicate a likelihood of success of a decoding event for decoding the content encoded in the flag.

The application relates to the field of machine recognition, and discloses a belt roller deviation detection method and device, computer equipment and a storage medium, the method comprises the following steps: acquiring a belt roller image, and detecting all edge line segments in the belt roller image through a line segment detection model; screening all the edge line segments, and determining a left edge line segment, a right edge line segment and a lower edge line segment of the belt roller from the remaining edge line segments; and determining whether the belt roller deviates according to the positions of the left edge line segment, the right edge line segment and the lower edge line segment. The belt roller deviation detection is fully automated, the detection cost is reduced, and the production efficiency is improved.

A method includes receiving, a plurality of records containing clinical information associated with one or more patients; extracting, using a Natural Language Processing (NLP) model, a plurality of medical entities from the clinical information to generate a first data set that contains the plurality of medical entities; denoising, the first dataset to generate a second data set by: determining relationship strengths between pairs of respective ones of the medical entities; identifying a subset of the pairs of the respective ones of the plurality of medical entities that satisfy a relationship strength threshold; generating uncommonality scores for one or both of a first and a second medical entity in each of the subset of pairs, the uncommonality score for the first medical entity being indicative of a frequency that the first medical entity occurs with the second medical entity across an entire set of instances of the second medical entity in the clinical information, the uncommonality score for the second medical entity being indicative of a frequency that the second medical entity occurs with the first medical entity across an entire set of instances of the first medical entity in the clinical information; and generating a relevance score for each of the subset of pairs based on one or both of the uncommonality scores for the first and second ones of the medical entities included in the respective pair and a frequency of occurrence of the respective pair in the clinical information; and generating a knowledge graph data structure representing ones of the subset of pairs having relevance scores, respectively, that satisfy a relevance threshold.

The invention discloses a bone nail recognition system and method based on machine recognition and artificial intelligence, and relates to the technical field of bone nail intelligent recognition. Bone nail tray partition images are collected through an image collection module, and the image quality of a bone nail tray is optimized by automatically adjusting shooting parameters; a camera of the image acquisition module moves in a partition above the tray through a three-axis movement mechanism, and then an image splicing and fusion unit of the image processing and recognition module splices and fuses partition images of the bone nail tray to obtain a complete image of the bone nail tray; and the deep learning recognition unit inputs the complete image of the bone nail tray into a pre-trained deep learning model, and outputs to obtain a bone nail information recognition result, so that the accuracy and efficiency of bone nail recognition can be improved.

The invention relates to the technical field of machine recognition, and discloses a substation pointer instrument scale digit anti-shielding reading recognition method and system, and the recognition method comprises the following steps: obtaining a current field image through a camera disposed right in front of an instrument; the camera is arranged in the heat preservation shell, transparent protective glass is arranged at the front end of the camera, the outer surface of the protective glass is coated with a super-hydrophobic transparent coating, an arc-shaped rain baffle is arranged above the protective glass, a rubber mechanical brush capable of moving up and down in a reciprocating mode is arranged at the bottom of the protective glass, and the protective glass is low-power transparent heating glass internally provided with an electric heating module. And inputting the acquired image into a pre-trained YOLOv5 target detection model which is improved for a small target, and detecting and outputting a position bounding box of each scale number in the image and a corresponding numerical value label thereof. According to the method, physical protection and cleaning are utilized, and the neural network and machine vision technologies are combined to solve the problem that instrument scale numbers are shielded and cannot be recognized. The method has the advantages of robustness, real-time performance and rapidness.

The invention relates to the technical field of machine identification, in particular to production line sorting equipment based on visual identification technology, which comprises a processing mechanism, the whole body is composed of a COREXY structure, and the COREXY structure is composed of a stepping motor, a belt, a synchronizing wheel and a toothed synchronizing wheel; the storage mechanism is composed of U-shaped channel steel and a funnel, the sorting mechanism is composed of a sliding rail and a mechanical arm, the whole sorting mechanism adopts a linear sliding rail structure and is connected with the sliding rail through a third sliding block, and the mechanical arm adopts a Cartesian coordinate system for path planning; the conveying mechanism is composed of a first roller, a second roller and a conveying belt. By means of machine sorting, high efficiency is achieved; the sorting accuracy of the machine is high, and the stable working quality is kept; the requirement for manual labor force is reduced through machine sorting, and automatic equipment can be more efficient in large-scale production or sorting tasks; the machine can normally work for a long time in a relatively severe or monotonous environment; machine sorting systems are generally easier to integrate and expand.

The embodiments of this specification provide a training method and device for a representation model used for human-machine recognition in barrier-free scenarios, and a human-machine recognition method and device. The training method includes: receiving data generated by the user's behavior and collected by the sensor on the terminal device side; performing at least two modification calculations on the data received from the same sensor; wherein the magnitude of the modification calculation for each modification is less than a preset magnitude threshold; using the representation model to be trained to obtain each representation of the data after at least two modification calculations; calculating the loss based on each representation of the data after at least two modification calculations; using the loss to adjust the parameters of the representation model to be trained until the loss function converges; wherein, when the distance between each position of each representation of the data after at least two modification calculations in the feature space of the representation model is less than a first preset distance threshold, the loss function converges. The embodiments of this specification can more effectively perform human-machine recognition in barrier-free scenarios.

A kind of machine vision rate distortion encoding method of license plate recognition task, by obtaining license plate image data set, extracting saliency feature, modulation adaptive quantization parameter, rate distortion optimization and fast division, license plate recognition step composition.The present application extracts background, vehicle, license plate three-level area feature using target detection network, calculates normalized perception intensity, uses hyperbolic tangent function to dynamically adjust the quantization parameter of coding unit, and optimizes the coding division process by combining subsection depth constraint strategy.Solves the technical problems of license plate character target blur and machine recognition rate decline under the conditions of application of video coding on unmanned aerial vehicle and limited code rate, reduces the encoding code rate and computational complexity, preserves the high-frequency microscopic details of license plate characters, improves the recognition accuracy and robustness of machine vision.The present application has the advantages of balancing code rate, low computational complexity, high license plate recognition accuracy, and can be used for unmanned aerial vehicle long-distance license plate recognition.

This application discloses a defense method and related equipment based on human-machine recognition, relating to the field of data processing technology. The method includes: responding to a human-machine recognition command, acquiring user operation data; performing human-machine recognition based on a preset human-machine recognition strategy and the user operation data; and obtaining a human-machine recognition result. The human-machine recognition result includes manual input and script input. The preset human-machine recognition strategy is constructed based on a preset lightweight model. If the human-machine recognition result is script input, defense is implemented. This application leverages the different trajectory features corresponding to manual input and script input, using trajectory features extracted from user operation data and a preset lightweight model to accurately determine whether the trajectory during the verification process is script input, thereby flexibly implementing defense.

The invention relates to the technical field of license plate recognition, in particular to a machine recognition method for a smart parking lot, electronic equipment, a medium and a product, and the method comprises the steps: obtaining an exit image, and recognizing exit vehicle information and exit time contained in the exit image, the exit vehicle information comprising an exit three-dimensional license plate image and an exit three-dimensional vehicle body image; matching corresponding entering vehicle information and entering time from an entering database based on the leaving vehicle information, wherein the entering vehicle information comprises an entering three-dimensional license plate image and an entering three-dimensional vehicle body image; and generating a to-be-paid order and a gate opening instruction based on the exit moment and the entrance moment, and feeding back the to-be-paid order. According to the invention, the precision of license plate recognition is improved, so that the accuracy and efficiency of automatic charging are improved.

A procedure in a computing device includes: capturing an image of a character via a sensor; determining a confidence level associated with a machine recognition of the character; selecting a feedback attribute based on the confidence level; and, prior to outputting content encoded in the character, generating a feedback signal with the feedback attribute via an output device, the attribute being configured to indicate a probability of success of a decoding event for decoding the content encoded in the character.

The invention relates to a computer-implemented method for more efficiently processing sensor data for an at least semi-autonomous driver assistance system of a vehicle, comprising the steps of: acquiring (110) sensor data by means of at least two different vehicle sensors, the acquired sensor data representing the surroundings of the vehicle; preparing (120) each piece of the sensor data to form a feature vector by means of a neural network; selecting (150) the relevant sensor data on the basis of at least some of the feature vectors representing the sensor data by means of a sparse gating network; determining (160) perception data on the basis of the relevant sensor data by means of at least one trained machine recognition method; and controlling (170) the vehicle on the basis of the determined perception data.