77results about How to "Reduce the number of false positives" patented technology

The invention discloses a driver posture recognition method based on depth images and virtual data. The method includes driver joint detection: the driver depth image and the joint label are used as the input of a deep learning framework Caffe, a deep convolution neural network model is trained, and the joint position of a driver in the image is detected through a deep learning model; virtual driver head posture data set establishment: a driver model is constructed through the modeling function of the three-dimensional modeling software, the head movement of the driver is set by utilizing theanimation function, the driver images is rendered in batches by utilizing the rendering function, the head segmentation processing is carried out on the driver images, and the virtual driver head posture data set is constructed; driver head posture analysis: a deep migration learning model is trained by means of a migration learning method according to the virtual driver head posture data obtainedby the virtual driver head posture data set construction process, and the driver head posture in the target domain image is estimated and the facial feature point position is detected.

The invention discloses a cross-layer route optimization method and a cross-layer route optimization device for a wireless Mesh network. The method comprises the following steps of: estimating the channel state of a current network according to the transmission efficiency parameter of a wireless link, judging the major cause of current packet loss according to a packet success ratio and a residual load rate when the channel state is poor, and adjusting the maximum retransmission time according to the packet loss cause; and acquiring the transmission efficiency PTEP of an entire route according to the transmission efficiency parameter of the wireless link, acquiring the residual load rate L-P of the route from which a source node and a target node are removed, computing the effective bandwidth etaB(c) of the route, constructing a route decision function according to the transmission efficiency PTEP of the route, the residual load rate L-P of the route and the effective bandwidth etaB(c) of the route, performing route discovery and route maintenance operation by using the route decision function, and selecting an optical packet transmission route between the source node and the target node. According to the invention, the route quality of an MAC (Media Access Control) layer can be sensed more accurately, and the error judgment times of the route are reduced.

The invention provides a flame detection method based on a video frame image, which is characterized by comprising the following steps of: S10) obtaining a video frame image; S20) according to a flame luminance threshold, obtaining binary foreground images, and obtaining an interested target chain by a plurality of binary foreground images; S30) according to the target characteristic of a target block in a target chain, calculating flame frequency, and judging whether flame exists in the video frame image in the target chain according to the flame frequency and the confidence coefficient, wherein the target characteristic comprises the average luminance variation, the area variation and the center position shake of the target block; S40) if the flame exists in the video frame image in the target chain, generating an alarm triggering signal; and otherwise, returning to step S10). According to the flame detection method based on the video frame image, which is disclosed by the invention, the flame can be accurately detected, so that the flame false alarm rate is greatly lowered so as to lighten the pressure of working personnel, and the working efficiency of the working personnel is improved.

The invention provides a moving vehicle detecting method based on aerially photographed video images, and relates to image movement analysis in image data processing. The moving vehicle detecting method based on the aerially photographed video images comprises the following steps: matching of input colored sequence images of a moving vehicle and further obtaining of background compensated images: carrying out image matching based on SURF feature points and estimating global moving parameters of a camera to obtain the background compensated images; roughly detecting the position of the moving vehicle; and accurate detection of the position of the moving vehicle: adaptively detecting roads, extracting CHLBP features of an alternative vehicle region and judging CHLBP features by using a SVM classifier to obtain the position of the moving vehicle. In the method, time and space characteristics are fused, and the shortcomings that a method in the prior art is only suitable for detecting a simple moving vehicle in single scenes, and is difficultly suitable for detecting multiple moving vehicles in different scenes, and detection accuracy is easily affected by changing in scale, complicated environments and movement of cameras are overcome.

The invention discloses a steel leakage visualized characteristic forecasting method based on an improved neural network and belongs to the technical field of steel metallurgy continuous casting detection. The steel leakage visualized characteristic forecasting method specifically comprises the steps that a thermocouple temperature signal of a crystallizer copper plate is online detected, the temperature change rate of the crystallizer copper plate is visually presented through a thermal imaging technology; on the basis of searching and extracting of the area, temperature change, position, transmission rate and other characteristics of a temperature anomaly area, a back-propagation (BP) neural network steel leakage forecasting model is established; in addition, by virtue of the self-organization and self-adaptability of a genetic algorithm, the power value and threshold value of the model are optimized, so that crystallizer steel leakage visualized online detection and forecasting are achieved. According to the steel leakage visualized characteristic forecasting method, not only are the distribution, anomalous change and development trend of the crystallizer temperature visually presented, but also a crystallizer steel leakage accident can be prevented in real time accurately, so that the false alarm times are reduced, and the accuracy rate of a steel leakage forecasting system is improved.

The utility model particularly relates to a remote management method for a door control, belonging to the environment monitoring field, wherein, name lists and corresponding mobile phone numbers of the entrants are stored, serving as basis to verify whether the entrants having authorities to enter; three aspects of verifications as the examination of identity and rights information in light of the mobile phone numbers, the check of the door control information according to the contents of the message, and the verification whether the entrant is near the door control according to the mobile phone positioning service, are carried out via message sent by the entrants; the door alarm information of the door control is shielded and the entrant is allowed to enter, after the verifications are adopted. The utility model has the advantages of remote and centralized management of the door control, and improved door control monitoring efficiency, and is particularly suitable for the machinery environment monitoring for large companies.

The invention discloses a database monitoring method and a device for solving the problems that the traditional database monitoring method has more false positives and low alarm accuracy rate for instantaneous monitoring objects. The method comprises: a marker bit is arranged on each monitoring object which needs double checks for marking whether collected data exceeds thresholds or not; the data of the monitoring objects is regularly collected and compared with the corresponding thresholds; corresponding marker bit is read when in need of double checks if the data exceeds the threshold, whether the data collected at the last time exceeds the threshold or not is judged according to the marker bit, if yes, an alarm message is emitted; and then the current comparison result is upgraded to the marker bit, wherein, the number of checks is defined according to the frequency of data collection. In the invention, the alarm message is emitted only when the monitoring object continuously exceeds the corresponding threshold for a plurality of times, thereby reducing the number of false positives, improving the accuracy rate of alarm, and providing a strong guarantee for the safe operation of a database.

The invention discloses a crystallizer bleed-out visual forecasting method based on machine vision and belongs to the technical field of steel metallurgical continuous casting detection. A thermocouple temperature signal of a copper plate of a crystallizer is detected on line; the temperature and a change rate of the temperature of the copper plate of the crystallizer are visually displayed by adopting a thermal imaging technology; and according to a machine vision theory, an abnormal temperature region is searched, and important information such as a geometric position, temperature change and temperature transmission of the abnormal region are extracted, so that a bleed-out sign is identified, and bleed-out of the crystallizer can be judged and forecast. The crystallizer bleed-out visual forecasting method comprises the following steps that: visualizing the temperature and the change rate of the temperature of the copper plate of the crystallizer; partitioning and marking a threshold value of the abnormal temperature region; extracting feature information of the abnormal temperature region; and identifying and judging a bleed-out temperature mode. The crystallizer bleed-out visual forecasting method has the advantages that visualization and a machine vision technology are organically combined, so that the temperature distribution and the abnormity change and development tendency of the crystallizer are derectly displayed; bleed-out of the crystallizer can be intuitively displayed and accurately identified by extracting features such as geometric position, temperature change and temperature transmission of the abnormal region; and the forecast accuracy can be effectively improved.

The invention provides a detection method and system for a sensitive video. With the method and system, a problem of inaccuracy of a detection result of a sensitive video can be solved. The detection method comprises: key frames are extracted from a to-be-detected video; frame difference processing is carried out on any two continuous key frames and a human body object region is obtained according to a frame difference result; the human body object region is tracked in multiple continuous key frames and human body object regions of all key frames are determined according to the tracking result; skin color detection is carried out on the human body object regions of all key frames at at least two color spade so as to determine skin color points of all key frames; on the basis of the skin color points of all key frames, skin color point feature information of the to-be-detected video is calculated, wherein the skin color point feature information contains a skin color proportion and a skin color point proportion changing amplitude; and whether the skin color point feature information of the to-be-detected video is larger than a preset feature threshold value; if so, the to-be-detected video is determined to be a sensitive video. With the method and system, precision and accuracy of skin color detection are improved.

The invention discloses an NSST (NonsubsampledShearlet Transform) domain MRF (Markov Random Field) and adaptive threshold fused remote sensing image change detection method, which solves the problem that edge information of a change region cannot be kept while a miscellaneous point is removed in the conventional change detection method. An implementation process for the method comprises the following steps of: inputting two remote sensing images of different time phases and constructing a difference image by using a difference value method; performing nonsubsampledShearlet decomposition on the difference image; combining a directional sub-band of each layer into a high-frequency sub-band; performing adaptive threshold classification on a high-frequency sub-band and a low-frequency sub-band of each layer to obtain a high-frequency adaptive threshold classification chart and a low-frequency adaptive threshold classification chart at each layer; performing MRF classification on the low-frequency sub-band of each layer respectively to obtain an MRF classification chart for each layer; and fusing the classification results to obtain a change detection result. The method has the advantages of high anti-noise property, high edge information retention capacity, less false drop of a detection result and high accuracy. The method is used for the fields such as urban area change monitoring, forestry and vegetation change monitoring and military target monitoring.

The invention discloses an improved bilateral filtering and clustered SAR based image change detection method mainly for solving the problem of high speckle noise and low accuracy of the existing image change detection method. The image change detection method is implemented by the following steps: 1, inputting two to-be-detected images of the same size; 2, performing denoising pretreatment on the two images to configure an initial difference image; 3, performing median filtering on the initial difference image to obtain a final difference image; 4, performing clustering on the final difference image to obtain an unchanged type fuzzy membership matrix uu and a changed type fuzzy membership matrix uc; and 5, performing assignment and classification on the elements in the changed type fuzzy membership matrix uc to obtain a final change detection result image. According to the improved bilateral filtering and clustered SAR based image change detection method, the number of false detection and the speckle noise are reduced, more image information is kept, and the accuracy rate and the precision of the change detection are improved, so that the image change detection method can be used for evaluation of disaster situations, urban construction and forest change monitoring.

The invention belongs to the technical field of steel metallurgy continuous casting inspection, and provides a crystallizer steel leakage forecasting method based on feature vectors and hierarchical clustering. The crystallizer steel leakage forecasting method based on the feature vectors and the hierarchical clustering comprises the following steps of 1, extracting characteristic vectors of sticking steel leakage, normal working condition historical data and temperature of online measured data to establish a feature vector sample set; 2, carrying out normalization processing to the feature vector sample set, and carrying out hierarchical clustering; and 3, checking and judging whether the feature vectors extracted online belong to steel leakage clusters or not, and further identifying andforecasting steel leakage of the crystallizer. According to the crystallizer steel leakage forecasting method based on the feature vectors and the hierarchical clustering, tedious debugging and modification steps involving alarm thresholds and the like are avoided, dependence of people on a previous steel leakage forecasting method is overcome, and good robustness and mobility are achieved; through extraction of temperature characteristic, a temperature mode of the sticking steel leakage can be accurately recognized, missing reports are avoided, false alarm frequency is remarkably reduced, data calculation amount and calculation time can be greatly reduced, and real-time performance of online forecasting is ensured.

The invention discloses a device and a method for detecting an electric bicycle attempting to enter an elevator. The device for detecting the electric bicycle attempting to enter the elevator comprises a module A which detects opening and closing status of an elevator door, a module B which detects the electric bicycle an a module C which outputs a detection result, wherein the module A is used for checking the opening and closing status of the elevator door, the module B is used for checking if the electric bicycle exists or not, and if the elevator door is detected to be open and simultaneously the electric bicycle is detected to exist, the module C sends out a warning message. The device and the method for detecting the electric bicycle attempting to enter the elevator reduce detection time scope and remove time intervals which do not need detection at all.

Embodiments of the invention disclose a subway electric carriage train examination train bottom and side fault detection system and method. The subway electric carriage train examination train bottomand side fault detection system comprises a rail side data acquisition apparatus, a number identification apparatus, a fault identification apparatus and a fault prompting apparatus. The method comprises the following steps: acquiring the number identification information of a subway electric traincarriage and an image of a traincarriage bottom and a traincarriage side; acquiring a number of the subway electric traincarriage according to the number identification information; identifying the image of the traincarriage bottom and train carriage side according to a preset fault identification model corresponding to the number of the subway electric traincarriage, and acquiring the fault information of the subway electric traincarriage; and when the fault information of the subway electric traincarriage is received, sending a prompting signal. By adopting the subway electric carriage train examination train bottom and side fault detection system and method, the fault can be automaticallyidentified, the operation efficiency of an operation management platform can be increased, the workload of the operation personnel can be alleviated, and the daily train vehicle examination efficiencycan be increased.

The invention relates to a breakout prediction method for a slab continuous casting mold based on withdrawal resistance, belonging to the field of continuous casting link control during the metallurgical process and mainly being used for solving the problem of time lag in the existing breakout prediction method based on temperature. The breakout prediction method comprises the following steps: (1) collecting production data at continuous casting site, and calculating the withdrawal resistance; (2) de-noising a resistance signal, respectively extracting resistance features in the cases of normal and abnormal (breakout and casting powder) continuous casting processes, and then constructing feature vectors; (3) training a resistance signal recognition model by means of the feature vectors and a support vector machine (SVM); and (4) transferring a real-time withdrawal resistance signal into the recognition model so as to judge the production condition of the slab continuous casting mold. The breakout prediction method has the beneficial effects that the breakout situation is judged and predicted based on the withdrawal resistance so as to obviously advance the prediction time, thus being beneficial to preventing breakout accidents, improving the casting blank quality and optimizing continuous casting technological parameters.

The invention discloses a hydrological data anomaly detection method based on spatio-temporal information. The method comprises the following steps: dividing associated sites; dividing a water level time sequence; obtaining a model output result by using the trained convolutional neural network (CNN) model, carrying out residual prediction on the model output result by using a Markov chain (MC), and judging an abnormal station according to the model output result and the predicted residual; obtaining abnormal conditions of the to-be-detected station and all associated stations; and performingresult fusion by adopting a dynamic distribution DS evidence theory (DADS) algorithm to obtain a hydrological data exception prediction result. According to the method, the influence of rainstorm seasons on hydrological data is fully considered, the detection precision is improved, a shuffled frog leaping algorithm (SFLA) is introduced to improve convolutional network parameters, an MC algorithm is added to carry out residual prediction, and the accuracy of prediction data is improved; and finally, through a dynamic distribution D-S evidence theory, fully considering spatial factors, and fusing multi-associated site prediction results, so the false alarm frequency is effectively reduced.

The invention provides a power monitoring image anomaly detection method and device, electronic equipment and a storage medium. The power monitoring image anomaly detection method comprises the stepsof classifying RGB features of a power monitoring image by using a pre-trained spatial CNN model; performing fusion classification on the optical flow features and the scene features of the power monitoring image by using a pre-trained fusion classification model; fusing the RGB classification result and the optical flow scene fusion classification result to obtain an anomaly detection result, wherein the optical flow features and the scene features are subjected to fusion classification by using the fusion classification model, so that the scene recognition effect is improved, the misjudgmentfrequency of behavior recognition is reduced, the learning ability of the CNN to the optical flow information is effectively improved, human behaviors can be recognized, and abnormal problem alarmingcan be timely and effectively carried out.

The present invention relates to systems and methods for detecting anomalies in computer network traffic with fewer false positives and without the need for time-consuming and unreliable historical baselines. Upon detection, traffic anomalies can be processed to determine valuable network insights, including health of interfaces, devices and network services, as well as to provide timely alerts in the event of attack.

The invention provides a sensitive information detection method and device and an electronic device. The method comprises the steps of obtaining a suspicious file set matched with sensitive information keywords; extracting a login account and a login password from the file content of each suspicious file for account login, and determining the suspicious file as a first target file containing sensitive information when login of at least one account in the suspicious file is successful; and based on a pre-established sensitive information identification model, identifying suspicious files exceptthe first target file in the suspicious file set to obtain a second target file containing sensitive information. According to the method provided by the invention, the suspicious file set is furtherscreened, so that the false alarm quantity caused by keyword fuzzy query is effectively reduced, the subsequent audit data volume is further reduced, and the screening efficiency is improved.

The invention discloses a touch screen point reporting method and related equipment. The method is applied to the electronic equipment comprising a system processing module, a touch control module anda near field communication (NFC) module, the NFC module is in an on state, and the method comprises the following steps: when the touch control module detects a touch control signal, the touch control module judges whether the NFC module is currently processing a data transmission event or not; wherein the touch module comprises a first report point threshold value and a second report point threshold value, and the first report point threshold value is greater than the second report point threshold value; wherein the touch signal comprises a first report point value and a first coordinate point; if the NFC module is currently processing the data transmission event, the touch module judges whether the first report point value is smaller than a first report point threshold value; and if thefirst report point value is greater than or equal to the first report point threshold, the touch module reports the first coordinate point to the system processing module. By adopting the method of the invention, when the data transmission task is processed through the NFC module, the frequency of touch screen report point errors or false reports can be reduced.

The invention discloses a multispectral image change detection method based on semi-supervised dimensionality reduction and a saliency map, for solving the defect that the spectral characteristic relationship of a multispectral image cannot be accurately reflected in the prior art. The multispectral image change detection method comprises the following realization steps of: (1) inputting two time-phase multispectral images; (2) preprocessing; (3) generating a corresponding waveband difference map; (4) carrying out semi-supervised dimensionality reduction; (5) generating the saliency map; (6) carrying out K mean value clustering; (7) generating a change detection result map; (8) outputting the change detection result map. The method is capable of either keeping the edge information of a change area well or giving consideration to missing detection information and false-alarm information in a change detection result well, good in real-time performance, and high in change detection result accuracy. The method disclosed by the invention can be applied to the fields of urban area extended monitoring, forest and vegetation disaster monitoring, crop growth state dynamic monitoring, military reconnaissance and the like.

The invention discloses an event auditing method and device, terminal equipment and a storage medium, and the method comprises the steps: obtaining preset sample data which comprises a training set and a test set formed by various types of event logs; constructing a fusion classification model through a TF-IDF model created in advance based on a training set and a test set in the sample data and in combination with multiple machine learning algorithms; and performing risk level evaluation on the event logs in the test set according to the fusion classification model. According to the scheme ofthe invention, a TF-IDF model is combined with a plurality of machine learning algorithms to carry out risk level evaluation on the event logs, high-risk events in massive security event logs are intelligently identified, the identification accuracy of the event logs is greatly improved, the false alarm number of the high-risk events is reduced, the workload of auditors is reduced, the auditing efficiency is improved, and meanwhile, the auditing accuracy is also improved.

The embodiment of the invention discloses an injection vulnerability detection method and device. The method comprises the following steps of acquiring a first parameter set, wherein the first parameter set includes different page parameters between a first page and a second page; sending N third requests and M fourth requests to a server, and receiving N third pages returned by the server aimingat the N third requests and M fourth pages returned aiming at the M fourth requests, and then acquiring a second parameter set, wherein the second parameter set includes different page parameters between each two pages in the N third pages and the first page, and different page parameters between each two pages in the M fourth pages and the second page; When at least one page parameter in the first parameter set is not in the second parameter set, determining that an injection vulnerability exists in the server. In the invention, the times of false alarm can be reduced and the accuracy of injection vulnerability detection is increased.

The invention provides a geographic provincial condition monitoring database management system and a method, and a database. The method comprises the steps of performing inspection on geographic provincial condition monitoring data of different years; processing the geographical province condition monitoring data of different years after the warehousing inspection to enable the unchanged data of the new version to inherit the characteristics of the background data; centrally storing, browsing, inquiring, retrieving and extracting the preprocessed geographical provincial condition monitoring data of different years, and integrally storing and managing the multi-year and multi-temporal monitoring data in a time axis manner; comparing the historical data in the monitoring area stored in the database integrated management module with the real data, performing change analysis of province monitoring data of different versions and different years, monitoring geographical province data elementlevel and land class level changes, and obtaining a change statistical analysis result; according to the method, management, display and comparative analysis of geographical provincial condition monitoring data with different year data standards are realized.

The invention provides a system used for maintaining safe distance. The system comprises a steering monitoring unit used for monitoring whether a vehicle steers or not, a distance measuring meter mounted on the vehicle and used for measuring distance between the vehicle and an ahead vehicle, a tachymeter used for acquiring traveling speed of the vehicle, an alarm and a controller used for receiving monitoring results of the steering monitoring unit. When the vehicle does not steer, the controller compares the distance with a preset safe distance according to the traveling speed of the vehicle, and when the distance is smaller than or equal to the safe distance, the controller controls the alarm to give to alarming signals; when the vehicle steers or the distance is larger than the safe distance according to the traveling speed of the vehicle, the controller controls the alarm not to give out the alarming signals. The invention further provides a method used for maintaining safe distance.

The invention provides an intelligent terminal and a sleep monitoring method, and the method can obtain sleep parameters such as respiration and heartbeat after a user starts a sleep monitoring function, preliminarily judges a sleep abnormal condition according to the sleep parameters, and starts a sound sensor to obtain user audio data when the sleep parameters do not meet a preset health condition. Therefore, the abnormal sleep condition is checked according to the audio data, the abnormal level is calculated, and different abnormal prompts are played according to different abnormal levels. According to the method, the abnormal sleep condition can be accurately detected in a preliminary judgment and secondary proofreading mode, the number of false alarms is reduced, and different abnormal prompts are played according to different abnormal levels so as to remind a user to take correct treatment measures and improve the user experience.

The invention discloses a device and a method for detecting an electric bicycle attempting to enter an elevator. The device comprises an electric bicycle detection algorithm module A, a module B for detecting and judging whether to start the electric bicycle detection algorithm module A, a module C for outputting a primary alarm message, a module D for checking primary alarm message information and performing analysis to make secondary judgment and outputting a secondary alarm message, and a module E for receiving secondary alarm results and processing; the module B analyzes the current stateof the elevator and judges whether the electric bicycle detection operation can be started, if yes, the module A is started, the module A detects whether the electric bicycle exists in the field of view, if the electric bicycle is detected, the module C is started, the module C outputs a primary alarm message and / or a program running state information, and the module D analyzes the output information and judges whether the alarm information and the association information of the secondary judgment result are sent out to the outside or not.

The invention provides an OPC verification method for a gate region. The method comprises the following steps of defining a feature graph of the gate region; calculating an inspection-free region of the feature graph; performing OPC verification of the gate region according to the inspection-free region. According to the method provided by the invention, the inspection-free area for the gate region is specifically set in the gate region of a polysilicon layer, specifically, the inspection-free region of the gate region is set by setting the inspection-free region of the feature graph of the gate region, and the method has the advantage that the number of report errors of the gate region caused by the fillet phenomenon can be greatly reduced, thereby improving the OPC verification efficiency of the gate region.

The invention provides a method for preparing a mask plate. According to the method, a first layer of graphics of a second layer of graphics of covered parts in initial target graphics are classifiedinto a first type of graphics and a second type of graphics according to a design rule of graphics before correcting the second layer of graphics in the initial target graphics with an existing optical proximity correction model; inspection specifications are respectively set for the first type of graphics and the second type of graphics to reduce the number of false positives when simulation software is utilized to inspect correction results, thereby reducing the time required for inspection and improving the final publishing efficiency of wafers.