29 results about "Manual interpretation" patented technology

The present disclosure provides an intracranial artery stenosis detection method and system. The present disclosure obtains artery stenosis detection results based on a first maximum intensity projection (MIP) image and a second MIP image obtained by preprocessing a medical image by adopting a detection model based on an adaptive triplet attention module and generates an auxiliary report and visualization results according to target category information in the artery stenosis detection results. Therefore, the problem that existing manual interpretation methods are easily affected by the subjective experience of doctors and are time-consuming and laborious can be solved, thus improving accuracy and efficiency of intracranial artery stenosis detection. Moreover, by inserting the adaptive triplet attention module into a backbone network of YOLOv4, the present disclosure can realize focus on key regions of high-dimensional features, reduce focus on irrelevant features, and provide characterization ability of the detection model.

The invention relates to a syphilis non-specific antibody detection result automatic interpretation method and system, and belongs to the technical field of biomedical detection. The method comprises the following steps: acquiring a sample image after agglutination reaction through shooting equipment, and transmitting the sample image to an automatic interpretation system through a wireless network; the system preprocesses the image, extracts morphological features, distribution features and texture features of agglutination particles by using a pre-trained AI model, and fuses metadata such as patient age and detection time to generate a fusion feature vector; performing similarity matching with a historical sample database based on the feature vector, and automatically interpreting a result through a weighting rule engine; and the interpretation result is output in a differentiated visual form, and the subjectivity of manual interpretation is eliminated and the problem of inconsistent results is solved through a unified image preprocessing and feature extraction method.

This invention relates to the field of measurement data analysis technology, specifically to an engineering measurement data analysis method and system based on image recognition. The method includes the following steps: using an industrial camera to identify structural edges, measurement markers, and calibration rulers to form a reference set of measurement point coordinates; establishing corresponding quantities of directional scales based on the calibration ruler direction; combining the displacement relationship of the measurement points to form a measurement point interval characterization set; determining the edge deviation reference rate; and finally forming a boundary positioning association result. This invention uses grayscale abrupt changes to define edges and construct measurement point positioning references; calculates the lateral and longitudinal displacements of the calibration ruler to extract corresponding directional scales; combines measurement point displacement and edge direction to form interval characterizations; analyzes the normal span and surface deviation state to form edge references; and combines the boundary extension orientation to define the associated segments of the arrangement trajectory. This reduces interference from manual interpretation, maintains consistency of multi-directional conversion benchmarks, controls the error accumulation process, and improves the coherence and stability of engineering monitoring data processing.

The present invention belongs to the field of exploration geophysics technology and relates to a method for producing seismic data labels for deep learning. The method comprises the following steps: Step 1. Loading seismic data and its existing interpretation results, and cutting and processing the seismic data; Step 2. Determining the position in the seismic data according to the interpretation results and assigning and labeling; Step 3. Slicing the seismic data. Slicing the seismic data, performing local transformation on the sliced label data body, and realizing the expansion of the training samples. The production method of the present invention can effectively improve the efficiency of label production with the help of existing manual interpretation results. The slicing operation can reduce redundant background information, and the local transformation can realize sample expansion. The method of the present invention can realize efficient labeling of seismic labels.

The invention discloses an ocean early warning algorithm generation method based on a large model, relates to the technical field of ocean supervision, and aims to solve the problems of low timeliness, poor accuracy and weak expandability of a traditional manual interpretation and manual coding mode. The method comprises the following steps: 1, acquiring a marine policy text through a distributed crawler, and performing cleaning and format standardization processing to obtain plain text policy data and metadata; 2, analyzing a text by using a fine-tuned large model in the field of marine policies, extracting an entity and semantic relationship, and converting the entity and semantic relationship into a structured rule containing a trigger condition, a judgment logic and an execution action; 3, storing rules and metadata according to a mixed storage scheme, and recording version history; and 4, mapping the rule to a preset algorithm template, filling service parameters to generate an executable code, and deploying the executable code. According to the method, automatic conversion from policies to early warning algorithms is realized, the supervision instantaneity and accuracy are improved, the cost is reduced, and smart ocean construction is supported.

This application relates to the field of geophysical exploration technology and discloses an exploration method for identifying coal mine goaf areas. The method includes acquiring multi-source geophysical data of the exploration area; spatially aligning and gridding the data; and using a deep autoencoder network for data-driven feature fusion to construct a unified feature space. Then, a structure-coupled collaborative inversion is performed, and by constructing and minimizing a joint objective function containing cross-gradient structure coupling terms, a multi-physical attribute model that maintains spatial consistency is obtained. Furthermore, a three-dimensional convolutional neural network is used to perform voxel-level classification of the inversion results, automatically interpreting and generating a three-dimensional geological classification model indicating the location and attributes of the goaf. This invention, through data-driven fusion, structure-coupled inversion, and intelligent interpretation, effectively reduces the ambiguity of inversion, replaces the subjective manual interpretation process, and significantly improves the accuracy, efficiency, and objectivity of goaf detection.

The application provides an object posture adjustment method and device. The object posture adjustment method comprises the following steps: collecting an object image; determining the marker point three-dimensional coordinates of each target marker point according to the marker point pixel coordinates of each target marker point in the object image, and determining the tool three-dimensional coordinates of each tool target ball according to the target ball pixel coordinates in the object image; determining the bush three-dimensional coordinates corresponding to each bush according to the positional relationship between each target marker point and at least three bushes on the object to be adjusted and the three-dimensional coordinates of each marker point; determining the current posture information of the object to be adjusted according to the three-dimensional coordinates of each bush and the three-dimensional coordinates of each tool, and adjusting the posture of the object to be adjusted according to the current posture information and target posture information. The method provided in the application converts the traditional posture adjustment process which depends on manual interpretation and single measurement into a digital and synchronous measurement mode. The posture adjustment efficiency is significantly improved, and reliable technical support is provided for the precise assembly of spacecraft structures.

The invention provides a carbonate rock hard bottom surface rapid identification method and device and a storage medium, and belongs to the technical field of oil and gas reservoir exploration and development. The method comprises the following steps: acquiring core image data; preprocessing the core image to obtain a processed core image; performing color identification according to the processed core image, and marking biological disturbance in the core image; image segmentation is carried out according to the hard bottom surface features and the core image marked with the biological disturbance, and a hard bottom surface area in the core image is marked. The core image is used for color identification, the corresponding biological disturbance is marked from the core image, and the biological disturbance in the core image is identified and segmented based on the hard bottom surface features, so that the hard bottom surface area in the core image is marked, manual identification is not needed, the hard bottom surface identification precision and efficiency are improved, and the identification efficiency is improved. Hard bottom distribution can be quickly and accurately positioned, and reliable data support is provided for reservoir modeling, well drilling design and development planning; and the subjectivity of manual interpretation is reduced.

This invention belongs to the field of automated bioinformatics analysis technology, and discloses an automatic interpretation and analysis method, device, equipment, and storage medium for sequencing data. The method includes: automatically interpreting sequencing data according to rule expressions set by quality control rules to obtain quality control results; if the quality control results are all passed, displaying a sample list and the target variants obtained after double-layer filtering of selected samples in the sample list by rule expressions set by data filtering rules and report rules; extracting the elements required for the report; and automatically filling in and generating a report for the target batch of samples based on the elements. Thus, by customizing quality control rule settings, filtering rule settings, and report rule settings, automatic interpretation of quality control and automatic screening of variants can be achieved, which can improve the degree of automated interpretation, improve analysis efficiency, reduce reliance on manual interpretation, and reduce the error rate of report interpretation analysis.

The invention discloses a medical examination report analysis system and method based on big data, and relates to the technical field of medical data processing, the system comprises a data acquisition module, a preprocessing module, a storage module, an index correlation analysis module, an abnormity early warning module, a result generation module and a user interaction module, and the method comprises corresponding data processing and analysis steps. Multi-source data such as inspection reports, historical diagnosis and treatment, clinical cases and the like are collected through a standardized interface, are subjected to cleaning, standardization and structural conversion, and are accessed by means of a distributed storage architecture; an Apriori association rule algorithm is adopted to mine index association, a random forest and a neural network model are combined to realize risk assessment, and a three-level grading early warning mechanism is matched to push accurate suggestions. According to the method, traditional manual interpretation is replaced, the analysis efficiency and accuracy are improved, the problems of insufficient associated information mining, non-uniform diagnosis standards and the like are solved, medical institution scenes at all levels are adapted, interactive optimization is supported, clinical precise diagnosis is assisted, the burden of doctors is relieved, and the method has important popularization value for intelligent application of medical data.

The invention relates to the technical field of material fatigue performance evaluation and structural integrity detection, in particular to a multi-stage fatigue crack propagation demarcation point identification and evaluation method and system, electronic equipment and a medium. Comprising the following steps: acquiring test data and constructing a curve, calculating a derivative by adopting a sliding window and identifying candidate demarcation points, and determining the demarcation points in combination with material behavior verification; the invention further discloses electronic equipment and a computer readable storage medium applying the method. The problems that in the prior art, a crack propagation stage division method depends on manual interpretation, the subjectivity of an identification result is high, and unified standards are lacked are solved.

The invention discloses a ground penetrating radar reinforcing steel bar positioning and parameter measuring method based on a dual detection network. The method comprises the following steps: step 1, preprocessing collected reinforcing steel bar data; step 2, intercepting an image with a reinforcing steel bar; step 3, marking a reinforcing steel bar area; step 4, intercepting the marked area; step 5, performing enhancement; step 6, carrying out yolov8 training to obtain two data sets; and step 7, obtaining steel bar target positioning and parameter measurement. According to the method disclosed by the invention, the efficiency of automatic processing is improved by 80% compared with that of manual interpretation, the method is suitable for large-scale engineering detection, the precision is optimized, noise interference is reduced by adopting dynamic region screening, and the accuracy of a dual-detection network in a complex scene is improved to 95% or above; the method expands the application range, supports double-layer reinforcing mesh recognition, and fills the blank in the prior art.

The invention provides geotechnical engineering in-situ investigation equipment based on three-dimensional scanning and a visual investigation method. The geotechnical engineering in-situ investigation equipment comprises an intelligent drilling unit, a data relay drill rod and a ground control platform, the intelligent drilling unit is mounted at the front end of the data relay drill rod, is coaxially positioned at the rear part of a drill bit of the data relay drill rod and continuously drills into rock soil along with the drill bit; the ground control platform is arranged on the ground on the outer side of the drilling area, and the ground control platform is connected with the intelligent drilling unit and the data relay drill rod. The invention relates to the technical field of geotechnical engineering investigation, and can solve the problems that in the prior art, spatial representation of a geological structure is discontinuous, geological information is interpreted depending on manual work, and a drilling camera cannot construct a three-dimensional model.

The invention discloses a supervision submission variable quantity data generation method and system. The data generation method comprises the steps that AI-driven data preprocessing and intelligent cleaning are carried out; a multi-table linkage recursive marking engine; predicting a change hotspot table preloading mechanism driven by the AI; parameterizing the common variation engine; and generating an NLP intelligent abstract and a risk report. Structured variable quantity data is automatically converted into a natural language report, and a risk level is attached, so that the manual interpretation cost is greatly reduced.

The application provides an examination site and probe posture recognition method and system based on an ultrasonic body mark label diagram, and a storage medium. The method is divided into a training stage and an inference and automatic labeling stage. The training stage includes constructing a multi-vendor ultrasonic body mark dataset, designing and training a "body mark positioning and anatomy classification integrated model" and a "probe posture key point regression model", and model joint evaluation and deployment. The inference and automatic labeling stage includes image acquisition and preprocessing, serial execution of two pre-trained models, completion of body mark positioning, anatomy site classification and probe posture key point extraction, body mark pose analysis and standard view label generation, and structured data generation and storage. The application realizes full-link automation of body mark positioning, anatomy classification and probe posture analysis of ultrasonic images through the construction of a double model joint inference, breaks through the limitation of body mark information being used only for manual interpretation, and realizes the automatic value conversion of ultrasonic body mark information.

The invention discloses a sandstone type uranium mine lithology identification method based on an improved random forest algorithm, and relates to the technical field of lithology identification. Comprising the following steps: screening original logging data, and screening out logging data corresponding to lithologic labels; dividing the screened logging data into a core feature set and an auxiliary feature set, and preprocessing the screened logging data to obtain preprocessed data; training a random forest model by using the preprocessed data and the known lithology label to form an improved random forest lithology identification model; and inputting the preprocessed data of the to-be-identified well section into the improved random forest lithology identification model to obtain a lithology identification result. According to the method, the data input quality is improved, so that the accuracy, robustness and practicability of the random forest model in a complex sandstone type uranium mine lithology recognition task are remarkably improved, and the defects of a traditional manual interpretation and standard machine learning method are overcome.

The invention discloses a river section data intelligent quality inspection method and system based on multi-modal data fusion, a river section data set to be subjected to quality inspection and reference data are received, and the method comprises the following steps: 1, data preprocessing: carrying out coordinate unified processing, topological relation reconstruction processing and data association processing on the river section data and the reference data; 2, inputting the preprocessed river section data and reference data into the constructed multi-modal rule engine to carry out spatial distribution verification, geometric morphology verification, section logic verification and data integrity verification; 3, according to a verification result, carrying out classification, grading and spatial positioning on discovered problems; and 4, generating a visual intelligent quality inspection report containing problem positioning information. Compared with the prior art, the method has the advantages that full-process automatic intelligent verification is realized by constructing the multi-modal rule engine, the huge workload of traditional manual interpretation is compressed to the minute level, and professional manpower is greatly liberated.

The present application relates to the technical field of coal mine geology engineering, in particular to a coal seam roof and floor equivalent lithology determination method based on drilling columnar chart information, which can extract all straight line objects and text object data in the drilling columnar chart in a software manner, match the text and rock stratum information by classifying the text objects into the areas divided by the corresponding vertical straight lines according to the horizontal positions, construct a table framework combined with the straight line objects, extract the sequence number, dip angle, true thickness and lithology data of each rock stratum, determine the coal seam roof and floor equivalent lithology based on the lithology description recognition, lithology classification rules and algorithms, and finally output the structured rock stratum data and equivalent lithology results. The present application has the beneficial effects of greatly improving the efficiency of the exploration drilling columnar chart information reading and processing, significantly reducing the subjective error of manual interpretation, realizing the accurate analysis of the rock stratum data, ensuring the accuracy and consistency of the coal seam roof and floor equivalent lithology determination, and laying an important foundation for the mine gas accuracy prediction.

The invention provides a horizontal well area high-precision three-dimensional geological modeling method which comprises the steps that on the basis of manual interpretation of geological stratification, a horizontal section to be subjected to geological stratification treatment of a horizontal well in a target area is selected, and stratification treatment is conducted on the horizontal section of the horizontal well; based on the vertical well, the horizontal well and the seismic data of the target area, establishing a three-dimensional grid structure model of the target area; and establishing a three-dimensional geological attribute model of the target area. According to the method, the three-dimensional geological modeling effect can be improved, and the modeling precision is improved.

The application discloses a kind of medical scientific research literature intelligent interpretation methods based on large language model, it is related to the cross technical field of natural language processing and medical informatics.The method is supported by three kinds of input modes of PMCid, XML full text and webpage URL through multi-source literature acquisition module, after extracting standardization field and metadata by literature structured analysis module, by the core intelligent interpretation engine based on "thirty-six strategies-logic methodology of medical scientific research", through eight sub-engines and multi-node collaborative reasoning chain architecture, realize the deep structured interpretation of medical scientific research literature, finally aggregate and output eight chapters standardization documents.The present application systematically embeds professional medical scientific research methodology into large language model reasoning process, solves the problems of low efficiency of traditional manual interpretation, existing AI tool interpretation is shallow and not standardized, shortens the depth interpretation time of single document from several hours to several minutes, guarantees the professionalism, standardization and repeatability of interpretation result.

The invention discloses an OCT (optical coherence tomography) image-oriented intelligent medical report generation method and system. Comprising an OCT image preprocessing and layer structure calibration module, a multi-level feature extraction and quantitative analysis module, a knowledge graph-based diagnostic reasoning module, a structured report template dynamic adaptation module, a clinical multi-dimensional data fusion optimization module and a report interactive auditing and iterative updating module. According to the method, the process from preprocessing to report output of the OCT image is automated, and the splitting property of manual interpretation and writing is avoided; based on a reasoning mechanism of the knowledge graph, the incidence relation between the features and the conclusion is clarified, quantitative indexes provide objective basis for a report, and clinical persuasion is enhanced; through template dynamic adaptation and multi-dimensional data fusion, the report can be matched with requirements of different departments and individual backgrounds of patients, and clinical pertinence is improved.

The invention relates to the technical field of syphilis detection, and discloses a syphilis detection result interpretation method, an interpretation device and electronic equipment. The syphilis detection result interpretation method comprises the following steps: acquiring a diluted reaction card image; determining a reaction card interpretation result under the 2n-time titer based on the reaction card image under the 2n-time titer and the reaction card image under the previous titer; wherein n is a positive integer greater than 0; and under the condition that the reaction card interpretation result is positive, obtaining a reaction card image under the latter titer, and repeating the steps until the reaction card interpretation result under the latter titer is negative. Unnecessary dilution and detection steps can be avoided, and consumption of reagents, reaction cards and time is remarkably reduced. As manual interpretation is not needed, the problems of high subjectivity, limited accuracy and low efficiency of manual interpretation of the RPR detection result by inspection personnel in the prior art are solved, and the accuracy of interpretation of the detection result is improved.

The present application relates to the technical field of industrial visual detection and pressure pipeline weld defect identification, and provides a pressure pipeline weld defect real-time detection method and system, comprising: acquiring a pressure pipeline weld image and constructing a data set containing crack defects, pore defects and spatter defects; constructing a hierarchical adaptive enhancement detection Transformer model based on an RT-DETR architecture, the model comprising a partial reparameterization enhancement module, a deep hierarchical adaptive feature fusion module and a global encoder; training the model based on the data set to obtain a trained defect detection model; inputting a to-be-detected image into the trained model to output a defect category and a positioning bounding box. The method can improve multi-scale defect recognition capability, anti-interference capability in complex background and detection efficiency, realize real-time detection of pressure pipeline weld defects, and reduce result differences caused by manual interpretation.

The application discloses a fault surface characterization and extraction method and system based on fault combination relationship judgment. The method comprises the following steps: enhancing the interpretability of target seismic data to obtain an enhanced data volume; establishing a relative isochronous model according to the enhanced data volume to obtain a target horizon; obtaining a sensitive attribute volume according to the enhanced data volume; obtaining an along-layer slice based on the sensitive attribute volume according to the target horizon; obtaining a fault strike according to the along-layer slice, rotating a seismic survey network direction to be perpendicular to the fault strike, and obtaining a fault feature enhanced display profile; judging a fault combination relationship according to the along-layer slice and the fault feature enhanced display profile; and obtaining a three-dimensional fault surface based on the along-layer slice and the fault feature enhanced display profile according to the relationship. The system comprises various functional modules for realizing the above steps. According to the application, the problems of poor fault surface effect generated by an existing post-stack attribute analysis based fault surface generation method and low efficiency of a manual interpretation fault surface method can be solved.

The present invention relates to artificial intelligence technology and provides a method for generating a chart information report. First, a chart data model is formed; then, the chart data model is parsed to obtain key chart data, and the key chart data is processed into a chart data card through a model analysis system. Then, the chart data card is input into a preset chart parsing model with a unique identifier, and then the chart data card is parsed by the chart parsing model to generate a chart parsing text, and classification information corresponding to the unique identifier is annotated on the chart parsing text according to the category to which the unique identifier belongs to form a chart information report. Each data card has a unique identifier, so that in the process of processing the data card, the security and stability of the overall process are improved, and the layout is automatically performed to form a chart information report, so that a standardized chart information report can be formed without manual interpretation, thereby improving reporting efficiency, saving manpower, and reducing overall project time.

This invention discloses a method for detecting linear deformation of fixed ice based on spaceborne InSAR and deep learning, belonging to the field of sea ice remote sensing monitoring technology. To address the problem of existing technologies' difficulty in automatically identifying subtle linear deformation features before fixed ice fracturing, this invention uses Sentinel-1 image data for differential interferometry to generate coherence maps. A pixel-level labeled dataset is constructed through manual interpretation. An improved U-Net network framework, LDF-UNet, suitable for linear deformation feature detection, is built, and the Dice loss function is introduced for training. Linear deformation features are automatically detected from the coherence maps. This invention achieves large-scale mapping of linear deformation features of fixed ice at 60-meter resolution. The detected features are consistent with the height of the ice arch fracturing boundary and can serve as an early indicator signal of fixed ice instability.

The invention provides a seismic facies prediction method and system based on a mask technology, and relates to the technical field of seismic facies prediction. Comprising the following steps: acquiring a three-dimensional seismic data volume, respectively selecting seismic data along a line direction and / or an xline direction and along the line direction and / or the xline direction after deflection of a set angle, establishing a seismic data set, and acquiring labels matched with the seismic data to obtain a label data set; training a neural network model by using the seismic data set and the tag data set, setting a mask matrix based on whether the seismic data has a corresponding tag, and truncating the seismic data without the corresponding tag in the back propagation process of the neural network gradient to complete the parameter update of the neural network. And obtaining a trained neural network model. According to the invention, the degree of freedom of label acquisition is improved, and the extraction of global spatial features is truly realized; and a mask technology is adopted to perform corresponding mask on the output end of the seismic data network without artificial interpretation labels, so that the prediction precision is improved.

The invention discloses a coastline classification method based on land-sea overall planning, and the method comprises the steps: carrying out the aerial photographing of a target coastline region through an unmanned plane carrying an optical camera, obtaining remote sensing images of a sea area and a land area in a 2km range in a direction perpendicular to the coastline, and synchronously recording the positioning and attitude determination data; aerial triangulation processing is carried out on the aerial photography image, pre-arranged image control point space coordinates are imported to carry out stabbing point optimization, and a two-dimensional orthoimage is generated; performing automatic interpretation on coastal zone ground features through a supervised classification method based on the orthoimage; clustering processing is carried out on the initial classification result, classification plaques with poor spatial continuity are combined, and an optimized classification graph is generated; and verifying and optimizing the precision of the classification graph through manual visual interpretation, and if the difference between the classification result and the manual interpretation exceeds a threshold value, readjusting the training sample. According to the invention, the automation level and interpretation precision of coastline classification can be improved.

The invention discloses a journal blade boss verticality detection device and method, and belongs to the technical field of aero-engine blade manufacturing. The device comprises a bottom plate, and a sliding seat and a support which are arranged on the bottom plate, the support is provided with a positioning shaft sleeve for positioning a blade shaft neck and a quick clamping mechanism; a gauge stand is slidably installed on the sliding seat, and the gauge stand is hinged to a lever and provided with a dial indicator. A detection contact at one end of the lever is used for contacting a blade boss, and a transmission contact at the other end is used for contacting a dial indicator measuring head. During detection, after the blade journal is positioned and clamped, the gauge stand is slid to enable the lever detection contact to sweep over the surface of the boss, the height change of the boss is amplified and transmitted to the dial indicator through the lever mechanism, and the maximum jerk value of the gauge hand is the perpendicularity error. Complex space geometric measurement is converted into visual linear reading, operation is easy and fast, efficiency is high, cost is low, and the problems that three-coordinate detection is low in efficiency and manual interpretation is not accurate are effectively solved.