40 results about "Machine recognition" patented technology

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 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.

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 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.

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.

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.

The invention discloses a welding seam scanning path calibration method based on point cloud registration, and belongs to the technical field of machine identification welding seam, and the method comprises the following steps: carrying out filtering and voxel downsampling on workpiece scanning point cloud data and digital-analog point cloud data; adopting a robustness plane fitting algorithm to obtain a main body bottom surface point cloud set; calculating the curvature of points in the point cloud set of the bottom surface of the main body and the feature descriptor of the fast point feature histogram, and identifying a low feature richness region; further sampling is carried out, and the number proportion of the data in the overall scanning point cloud data is reduced; adopting a principal component analysis method to obtain a coarse registration pose; adopting a point-to-surface ICP (Inductively Coupled Plasma) algorithm to carry out iterative precise registration to obtain an optimal registration result; and generating a high-precision welding seam scanning path mapped to an actual workpiece coordinate system according to the spatial transformation relation corresponding to the optimal registration result, thereby realizing full-process automation from point cloud preprocessing, weight division to coarse and fine registration, and remarkably improving operation efficiency and consistency.

This disclosure is directed generally to an automatic intelligent electronic data processing system, platform, and method for computerized multi-facet data pattern recognition and ranking, and particularly to intelligently personalizing recommendation of data items for consumption by a particular entity based on past data consumption history of the entity and / or other entities via machine recognition of intra and / or inter-entity data item selection correlations. Such personalized recommendation may be based on a multi-facet ranking of the data items by integrating various intra-entity and inter-entity correlations and patterns in data item consumption into a quantifiable entity-specific ranking score for each data item that may potentially be selected for consumption by a particular entity.

The invention discloses a fruit and vegetable shape sorting method and system, and relates to machine recognition. The method comprises the following steps: carrying out foreground and background segmentation on image data; according to the external contour coordinates of the fruits and vegetables in the foreground image, through an ellipse fitting algorithm, calculating semi-major axes and semi-minor axes corresponding to the N angle images respectively, and calculating eccentricity corresponding to the N angle images respectively; sequencing the semi-major axes, taking the sequenced maximum semi-major axis as a reference, judging the difference values between the rest semi-major axes and the maximum semi-major axis, obtaining images of which the difference values are smaller than a preset first threshold value, and obtaining an effective image set; when the number of the images in the effective image set is greater than a preset number threshold value, obtaining a maximum eccentricity and a minimum eccentricity corresponding to a major axis in the effective image set; when the difference value between the maximum eccentricity and the minimum eccentricity is larger than a preset second threshold value, the corresponding fruits and vegetables are judged to be flat fruits, and otherwise, the corresponding fruits and vegetables are judged to be non-flat fruits. The purpose of improving the flat fruit recognition accuracy is achieved.

This invention discloses a seal removal method based on a Seal-Inception GAN network, comprising the steps of: constructing and training a Seal-Inception GAN network; inputting an image containing a seal into the trained Seal-Inception GAN network for seal removal, and outputting the processed image. This invention utilizes the advantages of Cycle GAN's recurrent generation structure and unsupervised training, innovatively combining the Inception structure with the classic ResNet structure in the generator. Based on the characteristics of seal removal, the sampling structure and training dataset in the generator are adjusted to effectively preserve the handwritten characters in the seal portion while removing the seal from approval documents, indirectly improving the accuracy and efficiency of machine recognition.

The application discloses a kind of based on machine identification's road quality automation intelligent detection method and system, it is related to road detection technical field, the detection method steps include, according to the original image of captured road surface, after executing gray normalization operation, standard image matrix is constructed;After executing directional gray difference value calculation, build exception response matrix, using eight neighborhood connectivity determination method, candidate connected region is screened out, the area of each candidate connected region is compared and analyzed, and abnormal crack region is screened out, and crack region feature data is constructed;In combination with inverse tangent function, the surface structure response degree of each crack connected region is analyzed, the structure response coefficient of each crack connected region is determined, in combination with the abnormal texture response value of each candidate pixel point in each crack connected region extracted, determine the road surface abnormal intensity index;Judge whether the current road surface structure exists exception, and issue corresponding grade quality detection result.

A computer-implemented method includes receiving, by one or more processors, a set of structured records containing clinical information associated with one or more patients; extracting, by the one or more processors, a set of medical entities from the clinical information; determining, by the one or more processors, relationship strengths between pairs of the medical entities; and configuring, by the one or more processors, a computer-readable memory with a knowledge graph data structure in which nodes of the knowledge graph data structure represent the medical entities and the nodes are related to one another in the knowledge graph data structure based on the relationship strengths.

This invention discloses a web crawler detection system based on application scenarios, including a crawler detection platform. The crawler detection platform comprises a user analysis unit, a human-machine recognition unit, a secondary verification unit, a space development unit, and a reality integration unit. This invention relates to the field of web crawler detection technology. This application scenario-based web crawler detection system divides the enterprise network into different application scenarios according to their uses. Based on these application scenarios, it performs triple malicious crawler identification, achieving a more comprehensive screening of malicious crawlers. Simultaneously, it provides registered users with personal spaces, offering convenience and effectively avoiding false positives on genuine users, thus improving user experience. Furthermore, by combining network information related to the application scenarios to manage personal spaces, it greatly enhances the flexibility of malicious crawler detection while effectively locating abnormal parts of the enterprise network, providing powerful assistance for the operation and management of the enterprise network.