72 results about "Detection rate" patented technology

The invention relates to the technical field of artificial intelligence multi-agents, in particular to a proposal review method and system based on evidence verification and multi-agent cooperation. A cross validation mechanism based on a review information pool is introduced. The cross validation mechanism is a multi-agent round table cooperation mechanism, and a preliminary review result generated by an agent is not directly output, but passes multiple rounds of mutual verification, supplementation and correction until a preset convergence condition is met. According to the evidence verification process, cognitive deviation and knowledge blind areas possibly existing in a single agent are effectively eliminated, and the omission ratio and the misjudgment rate of the review result are remarkably reduced, so that the accuracy, the integrity and the reliability of a final review report are ensured; the technical problems of low efficiency and low accuracy caused by insufficient review depth, single angle and lack of verification in the prior art are effectively solved, and the review efficiency and decision quality of enterprise project proposals are remarkably improved.

The invention discloses a double-channel confrontation and progressive self-training driven cross-working-condition open set domain adaptive fault diagnosis method. In order to suppress an interactive negative migration effect, a three-stage two-way adversarial progressive self-training network DAPN is designed, separation of unknown target samples and feature aggregation of known samples are realized through cooperation of a gradient inversion layer and a two-way adversarial discriminator, and inter-class discrimination and intra-class aggregation are obviously enhanced. When the performance of the DAPN is evaluated in two open set domain adaptive scenes, for open set proportions of different intensity domain offsets and changes, the method always keeps high classification accuracy and unknown class detection rate, and the performance is obviously superior to that of a comparison baseline. According to the method, two-way confrontation and a progressive self-training mechanism are combined, so that accurate classification of known faults and effective identification of unknown faults are realized.

The invention discloses a target detection method and device based on domain self-adaption, equipment and a medium. The target detection method based on domain adaptation is realized through a target detection model, the target detection model comprises a dynamic domain adaptation module, and the dynamic domain adaptation module comprises an image level domain classifier with a first dynamic confrontation gradient inversion layer and an object level domain classifier with a second dynamic confrontation gradient inversion layer. In the back propagation of the training process of the target detection model, the gradient inversion intensity of the first dynamic confrontation gradient inversion layer is dynamically adjusted according to the image level domain classification loss, and the gradient inversion intensity of the second dynamic confrontation gradient inversion layer in the back propagation is dynamically adjusted according to the object level domain classification loss; according to the method, the target detection precision under the severe weather condition can be improved, the target missing detection rate and the target false detection rate under the severe weather condition are reduced, and then the safety and the reliability of the automatic driving system are improved.

The invention relates to a malicious software detection method based on multi-feature fusion and interpretability analysis. According to the technical scheme, static features are extracted through'macroscopic-microscopic 'double paths, and a dynamic behavior knowledge graph is constructed in combination with sandbox monitoring, stain analysis and semantic abstraction and converted into feature vectors through graph embedding; multi-modal feature deep interaction and accurate classification are realized through a double-end cross attention-triple fusion-deep classification architecture; a hierarchical interpretable framework is constructed based on cross-modal causal alignment, case reasoning and anti-fact analysis, and a complete decision evidence chain is generated. The method has the advantages that comprehensive representation of the form, semantics and intention of malicious software is achieved, the detection precision and robustness are remarkably improved, meanwhile, the'black box 'dilemma of the model is solved, credible explanation is provided for security analysis, the method is suitable for complex network threat detection scenes, and it is verified through experiments that the method has good application prospects. According to the method, the detection rate of the Windows malicious software is remarkably increased, and the false alarm rate is effectively reduced.

The utility model discloses a sampling device for artificial intelligence industrial quality inspection, which belongs to the technical field of industrial quality inspection and comprises a horizontal base, a conveying device and a bottom detection device. The conveying device is mounted on the horizontal base through a bracket; the bottom detection device is mounted at the upper part of the conveying device; the bottom detection device comprises a lifting seat, a clamping plate positioned at the lower part of the lifting seat, and a camera mounting seat which is mounted on the lifting seat through a hinge seat and can rotate in a vertical plane, and a first camera is mounted on the camera mounting seat; the bottom detection device is fixed to the upper portion of the conveying device support through a portal frame. The lifting base is fixed to the bottom of the portal frame beam through a hydraulic cylinder and telescopic guide rods vertically arranged on the two sides. According to the utility model, the technical problems that the judgment standards of different quality inspectors are possibly different, fatigue is easily caused by long-time repeated work, and the missing detection rate and the false detection rate are increased can be solved, and the detection precision and the detection efficiency are improved.

Systems and methods for learning from mistakes to improve detection rates of Machine Learning (ML) models. The systems and methods including receiving data with labels; running the data through a trained ML model for predictions; identifying errors in the predictions based on the labels received with the data; adjusting weights associated with samples in the data based on the identified errors; and retraining the ML model with the adjusted weights.

The application provides a kind of adaptive tracking type nondestructive testing method and system for special equipment inspection, belongs to detection technical field, method includes gathering equipment historical operation and detection data to construct multidimensional time series dataset, trains intelligent model combined with deep reinforcement learning and convolutional neural network;The model is deployed to edge computing system, and the initial reference baseline of multidimension is established on site, and the current state is compared with baseline in continuous detection, through the two-way decision mechanism of intelligent model and preset logic threshold, dynamically drive mechanical arm to adjust probe pose and synchronously optimize ultrasonic excitation parameters, to actively compensate the change of coupling state, maintain the optimal detection condition, and accurately identify and evaluate defects in stable state;The system includes defect accurate identification and evaluation module, etc.;The application realizes the dynamic self-optimization of detection condition under non-stop complex working condition, improves the defect detection rate, quantitative accuracy and repeatability of detection results.

The application discloses a synthetic aperture radar image target detection method, equipment and medium, and belongs to the technical field of SAR image target detection. The synthetic aperture radar image target detection method comprises the following steps: S1: inputting a SAR image; S2: constructing a feature separation module; S3: performing feature extraction; S4: constructing an adaptive feature aggregation network; S5: obtaining a target detection result; and S6: optimizing model parameters. The application solves the problem that the prior art is prone to causing high false detection rate and missed detection rate of multi-scale target detection. The application separates noise features and adaptively aggregates multi-scale features, reduces the interference of a complex background, improves the influence of a noise environment and multi-scale targets on detection, directly restricts the position and aspect ratio of a prediction box by using a regression loss, optimizes the multi-scale target detection effect of a model, reduces the false detection rate and missed detection rate of SAR image target detection, and improves the robustness of the model.

The present application relates to the technical field of defect detection, in particular to a kind of generator rotor forge internal defect ultrasonic and electromagnetic composite detection method, first to the surface of forge is adaptively path electromagnetic scanning, utilize multi-frequency eddy current technology and intelligent signal analysis identification and parameterization mark near-surface suspicious area;Subsequently, based on suspicious area information, through dynamic trigger mechanism executes global scanning and local directional precision scanning, generates the defect point cloud dataset containing multidimensional attribute;Finally, through three-level fusion architecture, electromagnetic and ultrasonic data are spatio-temporal alignment and feature-level correlation, and the type of defect is accurately determined and visualized output;The present application realizes the composite detection of the internal defect of generator rotor forge, can simultaneously improve the detection rate, positioning accuracy and type identification ability of defect.

The invention provides an urban building deformation monitoring leak detection point identification method and system, and belongs to the technical field of urban building health monitoring. According to the method, the isotropic hypothesis of a traditional two-dimensional plane buffer area is abandoned, a parameterized three-dimensional confidence space conforming to an SAR side-looking imaging physical mechanism is additionally constructed on the basis of contour strong judgment, multiple factors such as the radar sight line direction, the incident angle, the building height and the shielding effect are fully considered, misjudgment of non-target scattering points is effectively avoided, and the detection accuracy is improved. The spatial affiliation relationship between the PS point and the building is more accurate, and the reliability of omission ratio evaluation is remarkably improved.

The application discloses a human posture detection method and system based on target identification. The application uses key parts as object targets, and establishes a connection relationship based on a limb heat map to express human posture features, thereby reducing the missed detection rate under the shielding condition. Moreover, the posture features used by the application already contain a human boundary box, so when human posture features, human target features and part target features are needed at the same time subsequently, an accurate and non-regression fitting prediction boundary box can be obtained without additionally loading a target detection algorithm, and the detection speed is improved.

The application provides a method for identifying risk points of construction quality of a multi-ribbed beam floor based on artificial intelligence; the method organizes three-stage inspection information of formwork installation, support erection and pre-pouring review into point records according to the axis network coordinates, extracts and splices inspection items, measured and photo video semantic features, inputs the point risk score and risk source process prompts in combination with the stage conditions; and performs risk joint checking expansion based on a support chain, a formwork row and a boundary reinforcing belt correlation diagram, generates a review list according to the joint checking score and the on-duty capacity, and forms a closed loop by writing back the review conclusion; the scheme realizes phased and traceable point risk identification and review priority sequencing, improves the risk positioning accuracy and the review efficiency, and reduces the missed detection rate.

The application discloses a multi-dimensional tensor-based test literacy cognitive level intelligent labeling method and system, in order to solve the problem of multi-dimensional literacy rating performance collapse caused by expert labeling scarcity, loss of mathematical symbol semantics and long tail distribution in the prior art, a three-dimensional tensor space is constructed by fusing course standards; the first data set is used as a few-sample example, and a diagnostic thinking chain is combined to drive a large language model to generate structured pseudo-labels for unlabeled test questions in the second data set; through the incremental pre-training of the mask language model by injecting mathematical symbol prior and combining the cost-sensitive weighting mechanism based on real slot statistics, the lightweight training of the high-dimensional multi-task joint rating network is realized; finally, the active backflow closed loop of low confidence samples is executed through confidence evaluation. The application greatly improves the detection rate and rating accuracy of high-order long tail literacy under very small samples, significantly reduces the labeling cost, and realizes the engineering intelligent evaluation of cross-version mass question banks.

The invention discloses an ADAS perception algorithm target detection rate evaluation method and device based on an image large model, and a storage medium, and belongs to the technical field of intelligent driving. The method comprises the following steps: acquiring a verification data packet containing an original video; reasoning the video data by using a pre-trained image large model, and automatically generating a truth value file containing a target detection frame, a category and a tracking ID; inputting the same video data into a to-be-evaluated ADAS perception algorithm for recharging to obtain an output result file of the to-be-evaluated ADAS perception algorithm; through an automatic comparison tool, according to an intersection-to-union ratio (IoU) matching rule and a continuous frame statistical rule based on a target tracking ID, an output result is compared with a truth value, and key performance indexes such as a positive detection rate, a false detection rate and a missing detection rate are obtained through calculation. According to the method, a large model is utilized to automatically generate a high-quality truth value, a traditional manual annotation or expensive truth value system is replaced, high-efficiency, low-cost and standard-unified ADAS sensing algorithm performance evaluation and comparison are realized, and the algorithm development iteration efficiency is remarkably improved.

The application discloses an online detection system and method for a digital printing film switch, and relates to the technical field of machine vision detection. The method comprises the following steps: acquiring an initial image of a product to be detected; processing the initial image to obtain a second image; detecting the second image to obtain a size defect result, a position defect result, a color difference defect result, a bubble defect result and a crack defect result of the product to be detected; obtaining a total defect detection result of the product to be detected according to the size defect result, the position defect result, the color difference defect result, the bubble defect result and the crack defect result; and sorting the product to be detected according to the total defect detection result of the product to be detected. The method improves detection precision and production efficiency, and effectively reduces the missing detection rate and the false detection rate.

The application discloses a kind of multi-stage progressive exclusion and verification robust sparse sample mixed detection method and system, belong to popular disease early infection screening technical field.The method includes: according to the mixed detection parameter of screening scene setting;Subsequently enter multi-stage detection process, according to the current sample quantity to be detected in each stage, dynamically adjust mixed pool scale and detection times, and according to the detection result, sample is continuously classified and screened, and gradually exclude definite negative sample, while directly lock part of definite positive sample by analyzing the remaining sample in positive pool;Finally, the remaining sample is detected separately, and the identification of all positive samples is completed.The present application effectively copes with the uncertainty in actual detection through the robust multi-stage progressive design and dynamic adjustment mechanism, while ensuring the detection efficiency, significantly improves the detection rate of low viral load samples, has the characteristics of strong implementability and wide adaptability, and is suitable for large-scale sparse sample detection scene.

To provide an information processor, an information processing method and an information processing program for suppressing the deterioration of reading efficiency to a plurality of medical images photographed in one examination.SOLUTION: The information processing apparatus acquires a detection result of a lesion for a plurality of medical images obtained by imaging the same part at different timings in one examination, derives a detection rate of the lesion according to the number of medical images in which the same lesion is detected among the plurality of medical images, and performs control to display the detection result by changing a display mode of the detection result according to the detection rate or performs control to display the detection result and the detection rate.SELECTED DRAWING: Figure 6

The present application relates to the technical field of biological information analysis, and particularly relates to a construction method of a POLE mutation discrimination model based on second-generation sequencing results and application thereof, the construction method comprising S1, preprocessing of second-generation sequencing data, S2, sequence alignment, variation interpretation and mutation filtering, S3, analysis of filtered mutation data and POLE mutation discrimination. The POLE mutation discrimination model constructed by the construction method of the present application does not need to explicitly set POLE mutation as an amplification marker, sets a certain number of mutations and a maximum mutation abundance threshold, filters false positives caused by too few mutation sites, and combines POLE mutation discrimination score standards to assist in the determination of POLE mutation and improve the detection rate of true positives. Meanwhile, the mutation detection range of the present application covers 680 cancer-related genes, and the larger detection coverage makes the given score standard closer to the estimated hypothesis, and thus is more reliable.

This invention relates to a method and apparatus for object detection and determining training samples for object detection. The method, when dividing positive and negative samples, determines positive samples of target boxes based on the aspect ratio of the target boxes in the training image. Furthermore, when the aspect ratio of a target box is large, a first preset threshold inversely proportional to the aspect ratio is set. This results in a smaller first preset threshold when the aspect ratio of the target box is large. Using the intersection-union ratio (IUU) of candidate boxes and target boxes, along with the first preset threshold, can also effectively determine whether a candidate box is a positive sample, avoiding missed detections. This lays the foundation for training an object detection model based on positive and negative samples of target boxes in the training image. The trained object detection model can better learn the position and feature information of the object to be detected, reducing the missed detection rate of objects with large aspect ratios, thereby improving the accuracy of object detection.

The invention discloses a multi-lead electrocardio classification method and system based on a multi-scale residual network and dual calibration, and the method comprises the steps: obtaining original multi-lead electrocardio data, carrying out the data preprocessing, constructing an improved deep residual network model, constructing a joint objective function containing a Logit adjustment item and an asymmetric loss item, and carrying out the classification of the multi-lead electrocardio data. Training a deep residual network model by using the joint objective function; and performing classification prediction on the to-be-detected electrocardiogram data by using the calibrated model parameters, and outputting a final diagnosis result. The system comprises a preprocessing module, a feature extraction and classification module, a model training module and a calibration and reasoning module. The method effectively overcomes the limitation that a single receptive field is difficult to consider the instantaneous steep waveform and the local form contour at the same time; the problem that sparse abnormal features are easily ignored by global average is solved; the detection rate of rare arrhythmia under long tail distribution is obviously improved; and the reliability and the practicability of the auxiliary diagnosis system are greatly improved.

The utility model discloses a device for inspecting body paper scratches and DPO surface smudginess in paper box assembly, which is characterized by comprising a platform, a fixed plate and a movable plate, the fixed plate and the movable plate are positioned on two sides above the platform, guide rods penetrating through an L-shaped plate are respectively arranged above the fixed plate and the movable plate, a lamp strip is arranged at the vertical part of the L-shaped plate, and a light source is arranged on the lamp strip. And a threaded rod used for height adjustment penetrates through the horizontal part of the L-shaped plate. The inspection qualified rate during manual visual inspection is improved, and the omission ratio is reduced.

The application relates to the field of artificial intelligence, and discloses a kind of artificial intelligence's slub yarn fabric visual inspection method and system, to solve the problems of low recognition rate of micro-defects, high false detection and missed detection rate, poor environmental adaptability and model updating lag in prior art under complex texture background.The method comprises: collecting fabric images through high-resolution line array camera and multispectral programmable light source; constructing a five-layer multiscale texture feature pyramid network, fusing local gradient and global semantic features; deploying a cloud dynamic defect discrimination model, combining defect prior knowledge to guide pixel-level segmentation and classification; establishing an adaptive parameter optimization engine and incremental learning mechanism, the application realizes high-precision automatic detection of slub morphology anomaly and yarn breakage defect through the above-mentioned technical cooperation, the detection rate reaches 99.7%, the false detection rate is less than 0.5%, the detection speed reaches 120 meters / minute, and the online quality control demand of high-speed production line is met.

The application discloses a power distribution network data distortion identification and repair method and system, relates to the technical field of power distribution network data quality management, and comprises the following steps: firstly, the time series of voltage, current, active power and reactive power are denoised, aligned and interpolated; distortion such as drift, spike and loss is identified by combining statistical screening, machine learning and deep network; the abnormal section is respectively reconstructed by using Bayesian, GDBT and time sequence neural network, and the repair value is obtained by adaptively weighting according to historical error, type adaptation, feature correlation and working condition adaptability; and the feasibility check is carried out by introducing constraints such as power flow consistency, voltage range and line thermal limit. The method can significantly improve the detection rate and positioning accuracy, inhibit false repair, ensure that the result meets the physical law and equipment boundary, is still stable in the scene of sparse observation, noise and working condition switching, reduces manual review and outage risk, and is convenient for online landing.

This invention provides a vehicle braking method and apparatus, relating to the field of active safety driving technology. The method includes: calculating a longitudinal collision time based on the longitudinal position of an obstacle relative to the vehicle; determining whether the obstacle is a collision object based on the longitudinal collision time and the lateral position of the obstacle relative to the vehicle; if so, determining the detection condition of the collision object using sensors; confirming a target missed detection rate and a target false detection rate from the sensor characteristics, operating conditions, and the distance between the collision object and the vehicle; and determining whether to brake the collision object based on the target missed detection rate and the target false detection rate. This invention, by considering the sensor's false detection rate, missed detection rate, and errors, can effectively reduce false braking and missed braking, thereby achieving more accurate braking and making driving safer.

The invention relates to a tower-shaped test and penetration test inspection method for a large-diameter tube blank. According to the method, the defects of different depths below the surface can be effectively evaluated, and the detection rate of micro hairlines (lt: 0.5 mm) is obviously improved; the distribution rule of the defects in the radial direction and the axial direction of the pipe blank is obtained, and a direction is provided for process improvement; according to the invention, by quantizing data, human experience misjudgment is avoided, and the reliability of quality control is improved; the method is particularly suitable for large pipe blanks, and fills the blank of precision defect detection of the size section.

This invention first provides a method and system for evaluating whether paint will develop pinhole defects, comprising the following steps: storing multiple standard curves corresponding to standard paints with various different compositions; acquiring measured curves, randomly or according to instructions selecting multiple feature points and comparing the measured feature values ​​of the measured curves at the selected feature point positions; reading the standard feature values ​​corresponding to the selected feature points on the standard curves; and determining whether the paint will develop pinhole defects based on the deviation of the measured feature values ​​from the standard feature values. This invention utilizes surface tension curves to pre-assess whether the paint to be tested will develop pinhole defects, is simple and convenient to operate, cost-effective, and has a high detection rate.

This invention discloses a non-destructive testing method and system for track fasteners based on machine vision, belonging to the field of track inspection technology. By organizing the sleeper-beam nodes of the track network and creating inspection objects, a detection cloud center is built; associated sleeper-beam sets are extracted and the inspection objects and data clusters are labeled; a set of detection behavior features and anomaly feature sequences are generated; and the correlation of detection anomalies is analyzed using a sliding time window, achieving accurate inspection and maintenance of track fasteners. This invention solves the problems of low efficiency, high missed detection rate, and lack of correlation analysis in traditional inspection methods, enabling efficient and accurate inspection of track fastener conditions, reducing maintenance costs, and improving track operation safety. It is applicable to fastener inspection scenarios for various track lines, including high-speed rail and conventional rail.

The embodiment of the invention provides a gas leakage identification method, a storage medium and equipment, and belongs to the field of gas leakage detection. The method comprises the following steps: establishing a data set, wherein the data set comprises a gas leakage image and a non-leakage interference image; preprocessing the data to obtain a preprocessed data set; extracting gray level co-occurrence matrix features of the images in the preprocessed data set; performing dimension reduction processing on the gray level co-occurrence matrix features based on principal component analysis; and based on the data set and the corresponding dimension reduction processing characteristics, establishing a gas leakage identification model for identifying gas leakage. The method is used for improving the detection rate and the response speed of gas leakage detection.

The invention relates to the field of artificial intelligence, discloses an artificial intelligence slub yarn fabric visual detection method and system, and aims to solve the problems of low tiny defect recognition rate, high false detection and omission rate, poor environmental adaptability and model updating lag under a complex texture background in the prior art. The method comprises the following steps: cooperatively acquiring a fabric image through a high-resolution linear array camera and a multispectral programmable light source; constructing a five-layer multi-scale texture feature pyramid network, and fusing local gradient and global semantic features; deploying a cloud dynamic defect discrimination model, and realizing pixel-level segmentation and classification by combining defect priori knowledge guidance; according to the method, a self-adaptive parameter optimization engine and an incremental learning mechanism are established, high-precision full-automatic detection of abnormal bamboo joint shapes and yarn breakage defects is achieved through cooperation of the technologies, the detection rate reaches 99.7%, the false detection rate is lower than 0.5%, the detection speed reaches 120 m / min, and the online quality control requirement of a high-speed production line is met.

The application provides a film surface defect detection method and system, relates to the technical field of defect detection, and fundamentally improves detection performance by introducing a defect confidence evaluation mechanism based on morphology and energy distribution and constructing an intelligent secondary verification link. The mechanism can accurately distinguish real physical defects with regular morphology and concentrated energy from pseudo defects caused by electromagnetic interference, film instantaneous wrinkles and the like, effectively solves the high false alarm rate problem caused by the dependence of the prior art on single signal strength while retaining high detection rate for low contrast defects. In addition, the judgment model based on physical characteristics has natural robustness to background noise generated in high-speed motion, and the adjustable confidence threshold gives the system high practical flexibility, enabling it to adapt to complex industrial environments and different quality control standards, significantly enhancing the accuracy, reliability and intelligence level of the entire detection system.