123results about How to "Solve misjudgment" patented technology

The invention discloses a battery production defect inspection method, system, medium and equipment, and relates to the technical field of battery defect inspection. The method comprises the following steps: acquiring a video stream corresponding to each battery production line, and extracting a key frame sequence corresponding to each video stream; performing fusion analysis on spatial features and time sequence features of the key frame sequences to obtain abnormal state features; performing multi-stage screening on each key frame sequence based on the abnormal state features to obtain a target key frame with defects; performing space-time mapping on the target key frame, and determining an associated station corresponding to the abnormal production line; and according to the process parameters of the associated stations, carrying out grading evaluation on the abnormal state features, and generating abnormal early warning information. By implementing the technical scheme provided by the invention, misjudgment or missed judgment of abnormal conditions on the battery production line can be effectively avoided, so that the production quality of battery products is improved.

The invention relates to an MPPT (Maximum Power Point Tracking) control method based on dynamic optimization of a lightweight convolutional neural network, and belongs to the field of artificial intelligence control power tracking. The method comprises the following steps: S0, constructing a convolution-fuzzy neural network model which at least comprises an input layer, a convolution layer, a fuzzy control layer, a deep reinforcement learning layer and an output layer; s1, key parameters of photovoltaic equipment are collected in real time to serve as input data of an input layer; s2, performing convolution operation on the input data by using a convolution layer, extracting a feature vector, and transmitting the feature vector to a fuzzy control layer; s3, combining a fuzzy control layer with a deep reinforcement learning layer, and dynamically adjusting an optimization process of a fuzzy rule through a reward signal; and S4, the fuzzy control layer generates a control signal, and the control signal is transmitted to a DC / DC converter of the photovoltaic system through the output layer. The control method is suitable for a dynamic environment of a photovoltaic power generation system, can be expanded to a grid-connected system, and provides a new thought and technical support for efficient operation of photovoltaic power generation.

The invention discloses a steel plate butt weld pseudo defect ultrasonic detection method, system and device and a medium, and relates to the technical field of weld detection, the method comprises the following steps: setting a plurality of detection areas on a steel plate butt weld based on acquired structure parameters and surface characteristic data of a to-be-detected steel plate; performing multi-frequency and multi-angle ultrasonic scanning according to a preset first detection parameter by taking each detection area as a basic unit to obtain corresponding echo signal data; analyzing and identifying potential pseudo-defect signals, and calculating corresponding pseudo-defect characteristic values according to the pseudo-defect signals; taking each detection area as a basic unit, performing pseudo defect feature analysis according to the second detection parameters, and determining reliability evaluation data of pseudo defects; based on the pseudo-defect reliability evaluation data and the pseudo-defect characteristic value corresponding to each detection area, the pseudo-defect condition of the steel plate butt weld is analyzed, and a final weld quality comprehensive evaluation result is generated. According to the invention, the detection accuracy of steel plate butt weld detection is improved.

The invention relates to the technical field of coal mill powder pipes, in particular to a powder leakage defect automatic identification method based on digital twinning and AI video analysis, and the method comprises the steps: obtaining operation parameters and environmental parameters of a coal mill powder pipe, calculating a powder leakage risk characterization coefficient, and judging whether there is a potential risk of powder leakage; image data of the surface and the surrounding area of the powder pipe in the initial starting period are collected through an AI video analysis module, powder leakage characteristic parameters are extracted, a characteristic change curve segment is drawn, similarity comparison is conducted on the characteristic change curve segment and the reference characteristic curve segment, and whether a suspected powder leakage area exists or not is judged; actual operation parameters and model prediction parameters of the powder pipe in a stable operation period are obtained based on a digital twin model, the fluctuation degree of parameter deviation is calculated, and whether the powder leakage defect exists or not is comprehensively judged by combining a suspected powder leakage area judgment result; the early-stage accurate identification of the powder leakage defect is realized, the detection efficiency and precision are improved, and the operation and maintenance cost and the safety risk are reduced.

The invention relates to the technical field of medical instrument detection, in particular to a method and a system for detecting coating uniformity of a medical catheter, and solves the technical problem of low detection accuracy in the prior art. The method comprises the following steps: acquiring liquid-solid three-phase line contour information of each measurement point in a preset grid on the surface of a medical catheter sample and a corresponding initial contact angle; the preset grid is composed of a plurality of longitudinal measurement point sets; according to the liquid-solid three-phase line contour information of each measurement point in the preset grid and the corresponding initial contact angle, carrying out hydrophilic change trend analysis, and determining a local morphology disturbance degree corresponding to each measurement point and a comparability coefficient between different longitudinal measurement point sets; performing spatial interpolation correction on the initial contact angle based on the local morphology disturbance degree and the comparability coefficient to obtain a corrected contact angle of each measurement point; and performing coating uniformity detection on the medical catheter according to the corrected contact angle of each measuring point.

The invention discloses an unmanned express delivery vehicle control system and method based on operation state perception, and relates to the field of unmanned express delivery vehicle control systems. The method comprises the following steps: acquiring vehicle-mounted data in a driving period through a vehicle-mounted sensor array, performing time alignment and validity marking on the vehicle-mounted data to obtain a data quality label, constructing an evidence item set based on the data quality label, generating a sub-portrait label set according to the evidence item set, and displaying the sub-portrait label set in the driving period. And generating a portrait parameter set based on the sub-portrait label set through a preset priority rule and a persistence judgment strategy. Although the scheme has excellent performance in the aspects of improving system adaptability, safety, dynamic decision-making capability, reviewable performance and resource utilization efficiency, the problems of sensor faults, data loss, real-time data processing bottleneck, label interpretation errors and the like may influence the real-time response and decision-making precision of the system in a complex or extreme environment. Therefore, the overall performance and stability are reduced, especially under the condition of high load or low signal.

The invention provides a communication connection adjustment method and device, electronic equipment and a storage medium, and the method comprises the steps that first communication equipment determines a received signal strength indication of a communication signal received from second communication equipment, and determines a data packet error rate of the communication signal received from the second communication equipment; determining a target communication parameter according to the received signal strength indication and the data packet error rate; and adjusting the communication connection between the first communication device and the second communication device according to the target communication parameter. According to the embodiment of the invention, the PER is used as a core link quality evaluation index, a cooperative discrimination mechanism is formed with the RSSI, two typical link states of high signal strength but obvious interference and low signal strength but stable transmission are effectively distinguished, the problem of misjudgment caused by only depending on RSSI evaluation in the prior art is solved, and the accuracy of rate switching decision and the environmental adaptability are improved.

The invention discloses a semantic structure feature-based ancient book content integrity detection method and system, and relates to the technical field of information processing, the system comprises a data acquisition unit, a chapter integrity analysis unit, a paragraph integrity analysis unit, an integrity comprehensive analysis unit, a result feedback unit and a display terminal; structural features of chapters and paragraphs are accurately extracted by utilizing a semantic structure recognition technology, and the content integrity of the ancient books is comprehensively detected from the aspects of chapters and paragraphs in combination with an integrity analysis model, so that the defect that a traditional means depends on manual proofreading or simple page number comparison and keyword extraction, and the efficiency is high is effectively overcome. According to the method, the problems of typesetting confusion, chapter page missing, quotation omission and fuzzy paraphrasing which are common in the ancient books are difficult to deal with, automatic and multi-level analysis of the structural content and semantic level content integrity of the ancient books is realized, and the detection and recognition efficiency of the content integrity of the ancient books is improved.

The present invention discloses a method and system for intelligent analysis, prediction and early warning of the health status of chronic kidney disease, and relates to the technical field of medical data analysis and prediction. The method for intelligent analysis, prediction and early warning of the health status of chronic kidney disease obtains several historical health measurement data of the patient, and the historical health measurement data includes historical blood data, historical urine data, and historical physiological data; the historical health measurement data of each time of the patient is evaluated and analyzed respectively. The present invention integrates and dynamically analyzes the historical blood, urine and physiological data to construct a health assessment index for each time, and combines these indexes to form a comprehensive view of the health status of the kidney. Compared with traditional methods, this method can more accurately predict changes in renal function and potential lesions by considering multiple health dimensions and using trend analysis of historical data, and solves the problem of misjudgment that may be caused by a single data point in the prior art.

This invention provides a label defect detection method, apparatus, electronic device, and storage medium, belonging to the field of machine vision technology. The method includes: acquiring a registered and aligned binarized image of a template edge and a binarized image of the edge to be tested; performing dilation processing on the template edge binarized image and finding its intersection with the binarized image of the edge to be tested; subtracting this intersection from the binarized image of the edge to be tested to obtain a first defect region with redundant features; performing dilation processing on the binarized image of the edge to be tested and finding its intersection with the template edge binarized image; subtracting this intersection from the template edge binarized image to obtain a second defect region with missing features; and determining the defect detection result based on the aforementioned defect regions. This invention, through a bidirectional dilation tolerance differential mechanism, effectively accommodates local deformation caused by label wrinkles and positional offset of edge pixels, solving the problem of numerous false positives caused by local deformation, and significantly reducing the false alarm rate while ensuring a high defect detection rate.

The application provides an intelligent cruise garbage pickup and classification vehicle, method, equipment and medium to solve the problem that the existing laser radar can only obtain two-dimensional point cloud data and cannot identify garbage material, volume and other characteristics. The garbage pickup and classification vehicle comprises the following components: an upper computer, a lower computer, a laser radar, a depth camera, an odometer, an IMU, a mechanical arm and a wireless communication port. Through the heterogeneous data cooperation of the laser radar and the depth camera, the limitations of a single sensor are effectively eliminated. The laser radar provides high-precision two-dimensional spatial information, and the depth camera supplements three-dimensional geometric and visual semantic features, significantly enhancing the comprehensive perception ability of the garbage material, shape and volume. This fusion not only solves the problem of garbage material misjudgment in the traditional scheme, but also supports the adaptive grabbing strategy of the mechanical arm, reducing the risk of grabbing failure caused by environmental complexity. A double-bus communication redundancy design is adopted to ensure the stable transmission of key signals.

The present application relates to the technical field of automatic control and machine vision, and particularly relates to a path planning and blind area elimination method for a crawler-type cleaning robot, which comprises the following steps: collecting a photovoltaic module surface image through a vision sensor arranged on a robot body, extracting photovoltaic cell grid line features and establishing a local coordinate system; generating a path reference line based on the grid line direction, and calculating a lateral deviation and an angle deviation between a current position of the robot and the path reference line in real time; dynamically adjusting a robot walking path through a fuzzy control method; dynamically adjusting a cleaning interval by calculating a current cleaning width and an expected overlap rate; and executing an obstacle crossing or board crossing decision based on an inclination angle data when sensor data conflicts. A multi-sensor data fusion technology improves the anti-interference capability of the robot, so that the robot can stably operate in a strong light, a reflective light and the like.

The invention discloses a pipe blockage monitoring device, an application type medicine infusion device and a monitoring method, and relates to the technical field of medical instruments, in particular to a pipe blockage monitoring device comprising a Hall encoder used for detecting the rotation angle of a ratchet wheel in a ratchet wheel driving assembly and comprising a code disc and a Hall element, the code disc and the ratchet wheel are coaxially and fixedly installed, and the Hall element is used for detecting the rotation angle of the ratchet wheel in the ratchet wheel driving assembly; the Hall element and the medicine storage bin are oppositely and fixedly arranged; according to the tube blockage monitoring device, the application type medicine infusion device and the monitoring method, the rotation angle of the ratchet wheel is monitored through the non-contact code disc and the Hall element in the Hall encoder, then the problems of poor contact and increase of the rotation resistance of the ratchet wheel are avoided, and meanwhile the counting module is used for counting the action times of the driving shifting piece; the difference value between the rotation angle of the ratchet wheel and the action times of the driving shifting piece is calculated, when the difference value is larger than a set value, it is indicated that the infusion device has the ratchet wheel slipping problem caused by pipe blockage, the monitoring result is accurate, limitation of the length of a medicine storage bin is avoided, and applicability is high.

This invention provides a test method for determining injection fusion in multiple injections of a high-pressure common rail system, comprising the following steps: S1. Constructing a high-pressure common rail system injection pattern test system to test the injection pattern of two injections: pre-injection + main injection; S2. Constructing a high-speed photography direct-view test system for the injection process of the high-pressure common rail system to test the spray development characteristics under injection intervals; S3. Processing the images of the spray development process of multiple injections of the high-pressure common rail system, and repeating the spray development characteristic test until the fuel sprays of the two injections merge. The test method for determining injection fusion in multiple injections of a high-pressure common rail system described in this invention employs a spray visualization test method, obtaining the actual development process of real fuel spray in the cylinder, which can accurately determine whether actual fuel spray fusion has occurred in the cylinder, solving the problem of misjudgment of injection fusion based on control signals and injection pattern curves.

This invention provides a method and system for constructing a negative obstacle risk map, relating to the field of robot environmental perception technology. The method includes: acquiring multimodal sensing data of a legged robot in the current operating environment to obtain the types of negative obstacles, multiple candidate regions, and the corresponding confidence and uncertainty of each candidate region; using a risk diffusion algorithm, performing ink blurring processing based on the type, confidence, and uncertainty to generate a continuous risk field corresponding to each candidate region; dividing the continuous risk field to generate a continuous risk gradient distribution corresponding to each candidate region; performing terrain modeling based on laser point clouds, updating the continuous risk gradient distribution using Bayesian fusion to obtain a multi-level grid map, and then performing tactile closed-loop correction on the multi-level grid map to generate a risk cost map. This invention improves the accuracy of constructing a negative obstacle risk map.

The invention discloses a blood oxygen signal denoising method and system based on adaptive filtering, and the method comprises the steps: synchronously collecting an original blood oxygen signal and a three-axis acceleration signal through a sensor module, and obtaining an original blood oxygen signal sequence and a three-axis acceleration signal sequence which are aligned in time; the three-axis acceleration signals are fused into one-dimensional motion artifact reference noise signals used for representing the limb motion state of the user; the method comprises the following steps: dynamically calculating an adaptive step length factor sequence through a nonlinear mapping function based on multilevel energy analysis of a motion artifact reference noise signal; constructing a time-varying step size adaptive filter, and calculating noise components in the original blood oxygen signal by using the motion artifact reference noise signal and an adaptive step size factor; and subtracting the calculation noise signal from the original blood oxygen signal, and reconstructing to obtain a pure blood oxygen signal without motion artifacts. According to the method, accurate calculation and efficient suppression of motion artifacts can be achieved, and the purification quality and processing efficiency of blood oxygen signals are remarkably improved.

The present application relates to a kind of detection method and system for air bag bag production, it is related to the field of automobile passive safety technology, it includes obtaining current process number;Determine current tool number;Call in report record;Call in tool use information;Analysis report record and tool use information, to output abnormal reason;Real-time acquisition global overhead view and air bag fabric side view;Determine air bag fabric fluctuation point;Frame select air bag fabric fluctuation point corresponding fluctuation area;Determine fluctuation area height feature;Determine fluctuation area texture feature;Determine coating thickness anomaly, and output defect air bag bag signal;When there is no coating thickness anomaly, determine wrinkle area;When there is wrinkle area, execute flattening operation, and re-execute step 9 to step 12;When there is still wrinkle area after flattening operation, determine permanent wrinkle and output defect air bag bag signal.The present application has the effect of timely maintenance to tool and accurate traceability abnormal point in production process.

The invention provides a flower quality grading quality inspection method and system based on AI and image processing, and belongs to the technical field of computer vision and pattern recognizing.The method comprises the steps that an original image of the surface of a flower is collected through high-resolution imaging equipment, and high-frequency sub-band data representing tiny physical characteristics are extracted through multistage discrete wavelet transform; meanwhile, a sliding window is adopted to traverse the image, the color information entropy of a local area is calculated, and a color entropy graph is constructed. And after vectoring and splicing the two types of features, inputting the two types of features into a deep belief network model based on a multilayer restricted Boltzmann machine, and extracting deep abnormal feature codes. And finally, performing linear discriminant analysis on the code by utilizing a classification projection vector based on inter-class and intra-class distance optimization to generate an insect attack infection index, and realizing automatic grading quality inspection of the flower quality according to the insect attack infection index. According to the method, precise recognition and quantitative grading of tiny insect pests and recessive lesions on the surfaces of the fresh flowers are realized.

An optical time-of-flight sensor includes a sensor array and a channel sampling circuit. The channel sampling circuit is configured to receive a photon pulse signal generated from a sensor cell of the sensor array, receive a first channel select signal to generate a first freeze signal, generate a first channel event signal in response to the first freeze signal and the photon pulse signal to a first counting circuit, receive the first channel event signal, and generate and count a number of pulses of the photon pulse signal received by the first channel. The first freeze signal maintains a previous state when a rising edge of the photon pulse signal occurs until a falling edge of the photon pulse signal occurs. The provided time-of-flight sensor can significantly improve accuracy of a distance estimation operation by avoiding channel misjudgments without repeatedly calculating a number of arrived photons.

The invention relates to a walnut pairing method and device based on category feature self-adaption, equipment and a medium, and relates to the technical field of amusement walnut pairing, and the method comprises the steps: carrying out the rigid alignment processing of first contour data corresponding to a first walnut and second contour data corresponding to a second walnut; based on the first contour data and the second contour data after alignment processing, obtaining total contour loss between the first walnut and the second walnut, and based on the total contour loss, obtaining a target contour score; based on the type information of the first walnut or the second walnut, obtaining a target bottom line score and a target front line score between the first walnut and the second walnut; and based on the target contour score, the target bottom line score and the front line score, obtaining a comprehensive pairing score between the first walnut and the second walnut. According to the method, automatic pairing frameworks of different analysis strategies can be adaptively selected according to walnut category features, and the accuracy and efficiency of pairing are improved.

This invention relates to the technical field of deep integration of radio frequency identification (RFID) technology, indoor positioning technology, and intelligent algorithms. It discloses an RFID point-to-point positioning method and system that integrates inspection behavior characteristics. The method includes: within a single scanning cycle, using a mobile reading terminal to read multiple RFID tags and collect the identification information and multi-dimensional signal characteristics of each tag, including signal strength, TOA timestamp, and reading frequency; based on signal strength and reading frequency, eliminating invalid tags to obtain a candidate tag set; when the candidate tag set contains multiple tags, automatically determining a unique target tag from the multiple tags using a progressive judgment logic, where the inspection route information includes the order of each inspection node and its corresponding path weight; and outputting the physical location corresponding to the target tag as the current location. This invention fundamentally solves the technical problem of not being able to automatically distinguish valid locations when multiple tag signals conflict in traditional RFID tag-based positioning.

The utility model relates to the technical field of battery protection, and discloses a lead-acid battery protection circuit and a lead-acid battery, and the protection circuit comprises a main control unit which is connected with a charging switch, a pre-discharging branch circuit and a main discharging switch in parallel; the first end of the charging switch is connected with the anode of an external power supply, the first end of the main control unit is connected with the cathode of the external power supply, and the second end of the charging switch is connected with the second end of the main control unit; the first end of the pre-discharge branch and the first end of the main discharge switch are both connected with the positive electrode of an external power supply, and the second end of the pre-discharge branch and the second end of the main discharge switch are both connected with the positive electrode of a load; the third end of the master control unit is connected with the cathode of the load. According to the utility model, the problem of misjudgment of over-current or short-circuit protection of the lead-acid battery protection board on the load is solved.

This invention relates to the field of packer testing technology, specifically to a device for testing the airtightness of packers. The device includes a mounting base with a transparent water tank on its upper surface. A vertical lifting assembly is mounted on the mounting base, with its upper end extending outside the mounting base and fixedly mounted with a connecting block. One end of the connecting block is fixedly connected to a mounting plate, on which an air compressor is fixedly mounted. The other end of the connecting block is fixedly connected to a support block. A rotating hole is formed inside the support block, and a storage tube is rotatably connected to the rotating hole via a bearing. The storage tube is filled with an environmentally friendly water-based marking solution. The packer is quickly fixed via a threaded joint. A push plate and positioning rod control the entry of the marking solution. The sealing performance is visually assessed using the conspicuous, water-insoluble marking solution, air compressor pressurization, and the transparent water tank. A filter membrane separates the marking solution from the water; both are recyclable, residue-free, and environmentally friendly.

The invention relates to the technical field of nondestructive testing, in particular to a copper bar coating lower substrate defect nondestructive testing device based on terahertz waves, which comprises a terahertz scanning acquisition module used for controlling a transmitting and receiving unit to perform two-dimensional scanning so as to acquire an original waveform; the flight time dynamic correction module is used for resolving the local coating thickness and carrying out dynamic time axis alignment; the substrate roughness deconvolution module is used for calling the reference model to carry out deconvolution filtering so as to strip background scattering noise; the interface dielectric feature analysis module is used for extracting a dielectric feature vector reflecting an interface dielectric constant mutation state; the defect risk assessment imaging module is used for constructing a three-dimensional chromatography map and outputting a detection result; according to the method, the physical characteristic decoupling of the signal source is realized, the coating thickness fluctuation and the real defect are effectively distinguished, the misjudgment rate is obviously reduced, and the evaluation accuracy is ensured.

The application discloses a data intelligent analysis system and method based on network IP tracing, relates to the technical field of network tracing, and comprises the following steps: collecting network traffic data every unit period, identifying abnormal events in each unit period; identifying abnormal behavior patterns in different unit periods; extracting key information under the same behavior pattern; obtaining IP jump conditions of any abnormal event and generating an IP tracing path; judging the abnormality of each tracing path; analyzing the confidence of each key information and obtaining the comprehensive confidence of the extracted tracing path, setting a tracing confidence threshold of the tracing path; extracting key information from the real-time captured abnormal event and generating an expected tracing path; identifying the effectiveness of the expected tracing result; solving the misjudgment problem caused by static evaluation and greatly improving the reliability of the tracing result.

The invention relates to the technical field of microphone welding quality detection. The microphone welding direction detection method comprises the steps that acoustic feedback information of a microphone to be detected is obtained, the acoustic feedback information is processed, high-frequency interference and background noise are removed, filtered information is obtained, and the microphone welding direction is detected. Extracting frequency spectrum feature information in a preset frequency range from the filtered information, generating target frequency band information, performing phase calculation on the target frequency band information to obtain actual phase value information, inputting the actual phase value information into a welding direction judgment model, generating a feature difference value, and judging the welding direction based on the feature difference value. And determining the microphone with the wrong welding direction. The welding direction detection device has the effect of improving the reliability of welding direction detection.

This invention relates to the field of intelligent manufacturing and industrial data analysis technology, specifically to a method, system, and storage medium for identifying production bottlenecks based on multi-dimensional data fusion. The method includes: acquiring production data and constructing a multi-dimensional data vector; updating the weights of a linear prediction model based on the multi-dimensional data vector; performing multi-dimensional data fusion using the updated weights to obtain a comprehensive index value; obtaining a production bottleneck severity score based on the comprehensive index value; determining the production bottleneck status based on the production bottleneck severity score and the comprehensive index value; and identifying factors influencing the production bottleneck based on weight changes. This invention dynamically adjusts the weights using a gradient descent method to score the bottleneck, and dynamically updates the judgment threshold using an exponentially weighted moving average. Finally, it performs multi-dimensional cause analysis by comprehensively considering the weight contribution, change trend, and data anomaly, achieving adaptive and accurate location and cause identification of production bottlenecks.

The invention relates to a bean lack detection method and device for beverage equipment, the beverage equipment and a storage medium. In response to a beverage making instruction, a correlation signal is detected, the state of a bean grinder is detected at the same time, in any bean grinding period, if the correlation signal is not detected and it is determined that the bean grinder does not have an empty grinding state, a powder pressing link after grinding is executed, and the thickness of pressed powder is obtained; under the condition that the thickness of the pressed powder is smaller than the thickness threshold value, a first bean lack prompt signal is sent out. Whether beans exist in a bean storage bin or not is judged by detecting correlation signals at the beginning of bean grinding or in the bean grinding process, so that the situation that no beans exist in the bean storage bin can be found in time, the number of rotation turns of the bean grinder is dynamically monitored in the bean grinding stage, and bean lack detection is achieved in combination with a dual verification mechanism for monitoring the thickness of pressed powder in the powder pressing stage; the problem of misjudgment caused by detection only through correlation signals can be avoided, and the accuracy and robustness of bean lack detection are remarkably improved.

The present invention discloses a three-dimensional head posture estimation algorithm based on facial feature information, comprising the following steps: obtaining camera calibration parameters and three-dimensional coordinates of multiple model points selected in the facial area of ​​a head model; collecting a facial picture of a human face whose posture is to be estimated by a camera, extracting feature points corresponding to the model points, and obtaining two-dimensional pixel coordinates; calculating head posture estimation parameters according to the camera calibration parameters, the three-dimensional coordinates, and the two-dimensional pixel coordinates; reprojection verification of the head posture estimation parameters, if the reprojection error is greater than a set value, then correcting the head posture estimation parameters by an extended Kalman filter, so that the reprojection error of the corrected head posture parameters is less than or equal to the set value. In this way, by means of the extended Kalman filter correction, it is ensured that the reprojection error of the head posture estimation parameters is small, and it is avoided that the algorithm based on the two-dimensional image will misjudge the head posture estimation.

This disclosure provides a security detection method, apparatus, storage medium, and electronic device based on a large model, relating to the field of network security technology. The method includes: acquiring training sample emails to train a Bayesian classifier, determining the trained Bayesian classifier; using the trained Bayesian classifier to detect emails to be detected, determining a first detection result; if the first detection result is in a first detection interval, determining structured prompt words based on the first detection result; inputting the structured prompt words into a locally deployed large language model, determining the output result of the large language model; and performing either allow or block actions on the emails to be detected based on the first detection result and the output result of the large language model. This achieves accurate email identification.