612results about How to "Improve accuracy" patented technology

The subject matter of this specification can be embodied in, among other things, a method that includes receiving an audio signal and initiating speech recognition tasks by a plurality of speech recognition systems (SRS's). Each SRS is configured to generate a recognition result specifying possible speech included in the audio signal and a confidence value indicating a confidence in a correctness of the speech result. The method also includes completing a portion of the speech recognition tasks including generating one or more recognition results and one or more confidence values for the one or more recognition results, determining whether the one or more confidence values meets a confidence threshold, aborting a remaining portion of the speech recognition tasks for SRS's that have not completed generating a recognition result, and outputting a final recognition result based on at least one of the generated one or more speech results.

An information input method of a computer system having a virtual keyboard includes detecting motion information of a user's hands and fingers, determining the locations of the user's hands and fingers by interpreting the detected motion information, displaying an input apparatus having a predetermined shape on the virtual keyboard of a screen based on the determined locations of the user's hands and fingers, and applying force to a finger corresponding to the location where information is input, if the input information is input by the displayed input apparatus.

The invention discloses a pre-warning method for a vehicle. According to the method, the most appropriate pre-warning rule can be customized according to specific feature information of the vehicle, the pre-warning accuracy is high, and risks in the vehicle driving process are greatly reduced. The method is executed by a vehicle terminal and a background processing system mounted on the vehicle and comprises steps as follows: (1), indexes requiring attention are abstracted according to factors related to driving safety, and a corresponding weighted value is set for each index; (2), vehicle data are collected and comprise static data, vehicle driving dynamic data, space data and driver information; (3), a risk coefficient of each index is calculated according to the weighted value of each index and the corresponding vehicle data, the risk coefficients of all the indexes are accumulated, and a total risk coefficient is obtained; (4), whether the total risk coefficient is in a specified range is judged, if yes, the step (5) is executed, and otherwise, the step (2) is executed; (5), a pre-warning message is sent to the vehicle terminal.

A computerized method for detecting malware associated with an object. The method includes operations of analyzing an object to obtain a first set of attributes, where the first set of attributes include one or more characteristics associated with the object. Furthermore, the object is processed with a virtual machine to obtain a second set of attributes. The second set of attributes corresponds to one or more monitored behaviors of the virtual machine during processing of the object. Thereafter, a threat index is determined based, at least in part, on a combination of at least one attribute of the first set of attributes and at least one attribute of the second set of attributes. The threat index represents a probability of maliciousness associated with the object.

A method of obtaining information about the position and / or orientation of a magnetic component relatively to a magnetometric detector, the magnetic component and the magnetometric detector being moveable independently from each other relatively to a static secondary magnetic field, the method comprising the steps of: measuring in the presence of the combination of both the magnetic field of the magnetic component and the static secondary magnetic field essentially simultaneously the strength and / or orientation of a magnetic field at at least a first position and a second position spatially associated with the magnetometric detector, the second position being distanced from the first position; and combining the results of the measurements to computationally eliminate the effect of the secondary magnetic field and derive the information about the position and / or orientation of the magnetic component.

The invention is suitable for the technical field of machine vision, and provides an object recognition and positioning method and device and terminal equipment. The method comprises the steps: obtaining a two-dimensional image and point cloud data of a to-be-detected region; detecting the two-dimensional image through a pre-trained deep learning model, and identifying a two-dimensional target area and a geometrical shape type corresponding to a target object in the two-dimensional image; mapping the two-dimensional target area to the point cloud data, and determining a first three-dimensionalarea of the target object according to a mapping result; and according to the geometrical shape type and the first three-dimensional area, determining a second three-dimensional area of the target object and positioning the target object. According to the embodiment of the invention, the 3D object recognition and positioning efficiency and accuracy can be improved.

The invention provides a compression braking device of a four-stroke cycle internal combustion engine. A control valve mechanism (13) and a clearance compensation mechanism (23) with the compensated clearance being the lift range of a braking projection are installed in parallel on one end of an amounting elephant feet of a rocker (3); a clearance compensation mechanism (11) with the compensated clearance being the lift range of a valve clearance is arranged in the midpoint of the connecting positions of an exhaust valve bridge (5) and two exhaust valves (6); the elephant feet (24) are coaxially connected below the lower end of the previous clearance compensation mechanism (23); the upper end of the next clearance compensation mechanism (11) is always in contact with the lower ends of the elephant feet; and the spring force of the clearance compensation mechanism is smaller than braking oil pressure and larger than dead weight of the elephant feet. Compared with the prior art, the compression braking device improves the reliability and the balance of a valve actuating mechanism, simplifies the structure, reduces the processing difficulty of the rocker, improves the rigidity of the rocker, and improves the accuracy and operability of the adjustment of the valve clearance. In addition, when the engine compresses and brakes, the two exhaust valves can be opened at the same time, thereby improving the braking effect of the engine.

The present invention discloses a pipeline water quality risk evaluation method based on a water supply network dynamic model. Establishing a water supply network dynamic waterpower and water quality model and performing accurate checking of the water supply network dynamic waterpower and water quality model through determination of the classification type and the evaluation indexes of the pipeline water quality risk; determining the weight of each evaluation index through adoption of the analytic hierarchy process; and dividing the grading standard of each evaluation index parameters based on the state data samples of the pipe network dynamic model, determining the degree of membership of each evaluation index, performing the fuzzy comprehensive evaluation, and determining the grade of the water quality risk of each pipeline in the pipe network. Though the steps mentioned above, all the pipelines of the whole water supply network are subjected to water quality risk evaluation, and the comparison and the ordering are performed according to the evaluation result to determine risk degree ordering of water quality changing and provide decision support for the management and maintenance of the pipe network; and moreover, the pipeline water quality risk evaluation method based on the water supply network dynamic model overcomes the defects that other evaluation methods are simple in index and do not take incomprehensive consideration, and the evaluation result is more reliable to ensure the accuracy and the life cycle of the evaluation.

The invention provides an Android application safety analysis method based on sensitive behavior identification. The Android application safety analysis method comprises the following steps: 1) obtaining and analyzing a source code; 2) obtaining sensitive behaviors; 3) extracting a UI (User Interface) text; 4) processing sensitive behavior data; 5) extracting feature values; 6) forming a training data set; and 7) analyzing safety. A relationship between the UI text and a sensitive API (Application Program Interface) is taken as the characteristics of machine learning, and accuracy for detecting the malicious applications of an Android platform is effectively improved.

Based on an image captured by an onboard camera, an arrow signal detector sets an arrow signal area on the basis of a signal light distance between a lit red signal light of a traffic light and a vehicle equipped with the arrow signal recognition device, counts the number of color tone effective pixels assumed as being lit within each arrow signal area, further searches for and counts color tone ineffective pixels in the color tone effective pixels on the basis of pixel information on the vicinity of each color tone effective pixel, and calculates an arrow effective pixel number from the difference between the number of color tone effective pixels and the number of color tone ineffective pixels.

This invention is a method and apparatus for monitoring the concentration of carbon dioxide dissolved in water by means of conductivity. It distinguishes between the conductivity resulting from carbon dioxide and the conductivity resulting from other constituents dissolved in water. It can be used to monitor the quality of demineralized water, boiler feedwater, steam, or condensate in electric power generation and other industrial facilities. It is constructed by adding a column containing weak base anion exchange resin and a conductivity instrument to a typical cation conductivity monitor. A sample of the water to be monitored flows first through a typical cation conductivity monitor, then through a weak base anion exchange column, and then through an additional conductivity instrument. Conductivity measured at the outlet of the weak base anion exchange column will be essentially due to whatever concentration of carbon dioxide is dissolved in the sample because other dissolved constituents that affect conductivity have been essentially removed by either the cation exchange resin that is part of a typical cation conductivity monitor, or by the weak base anion exchange resin. By subtracting the value of conductivity due to carbon dioxide (at the outlet of the weak base anion exchange column) from the value of cation conductivity (at the outlet of the cation exchange column), the value of degassed cation conductivity is obtained. In the title of the invention, CWB conductivity is an abbreviation for cation—weak base conductivity.In combination with existing methods for oxidizing organic compounds dissolved in water, this invention is also a method and apparatus for monitoring the concentration of dissolved or total organic carbon in water by means of conductivity. It distinguishes between the conductivity resulting from organic carbon and the conductivity resulting from inorganic constituents dissolved in water including carbon dioxide.

The invention brings forward a method for automatically positioning a standard tangent plane from an ultrasonic image. The method comprises the following steps: S103, constructing a radial component module to describe the anatomical structure of a target area, performing area dividing on the target area in the ultrasonic image by use of the radial component module, and obtaining a limitation area; S105, respectively detecting different anatomical structures in the limitation area by use of multiple classifiers trained by a machine learning method, and respectively obtaining detection results; S107, fusing the detection results by use of a fusion classier, and obtaining a fusion result; and S109, according to the fusion result, determining whether the ultrasonic image is the standard tangent plane. The method provided by the invention can effectively improve the detection accuracy.

A work cutting apparatus comprises a bed. The bed has an upper surface provided with a column including a rail slidably mounted with a slider. The slider has a front surface mounted with a supporting portion supporting two end portions of a rotating shaft. The rotating shaft is mounted with a plurality of cutting blade blocks. Each of the cutting blade blocks includes a plurality of cutting blades and a thicker cutting blade at each end of the cutting blade block. A table provided with a recess having a V-shaped section is disposed on the bed right beneath the cutting blade blocks. A plurality of works are disposed in the recess, each fixed by a fixing member. The cutting blades lowered while rotating, thereby cutting the works. During this operation, coolant is discharged from a plurality of supplying ports of a coolant supplying portion as well as from a supplying port of a coolant supplying path.

The invention discloses a ground traffic sign real-time detection method based on intelligent driving, and belongs to the field of traffic information detection in the intelligent traffic industry. The method comprises the steps of obtaining an image Src of a road in front of an intelligent vehicle in a real-time mode, conducting cutting, grey-level transformation, Gaussian filtering and binarization processing on the obtained original image Src to obtain a binary image Src_bw, meanwhile, and reading in a ground traffic sign template image temp_i prepared in advance, wherein each template can find one best matching area corresponding to the template in the Src_bw through template matching; cutting the templates to be stored as new images dst_i, and conducting subtraction on the temp_i and the corresponding dst_i to obtain new images diff_i; conducting statistic on white pixel number diff_i_Sum in each diff_i, wherein the template corresponding to the diff_i with the smallest white pixel number is most similar to a ground sign; when the diff_i_Sum is smaller than the set threshold value, considering that the ground traffic sign in the template exists in the Src_bw. The ground traffic sign real-time detection method is applicable for the intelligent driving in complex city road environment.

An inspection system and an inspection control method based on an electric power inspection robot are disclosed. The inspection control method comprise the steps: when the electric power inspection robot inspects electric power equipment and reaches the location of target electric power equipment, acquiring state information of the target electric power equipment, wherein the state information comprises a temperature signal, a sound frequency signal and a visible light image; judging whether the target power equipment is abnormal or not according to the temperature signal, the sound frequencysignal and the visible light image; and if so, obtaining monitoring information of an online monitoring system of the power equipment, and when the monitoring information shows that the target power equipment has a fault, determining that the target power equipment has the fault. According to the method, comprehensive decision making is carried out on the equipment state through multiple kinds ofperception information, the equipment state is judged in combination with the monitoring result of the online monitoring system, and the phenomena of misjudgment and missed judgment on the equipment state are greatly reduced.

The invention discloses a method for implementing electronic toll collection, which comprises the following steps: a distance between a trigger ground sensor and a snapping ground sensor is determined; the trigger ground sensor triggers a lane reader-writer when detecting that a vehicle travels into an antenna communication area; a vehicle-mounted electronic label caries out power detection on a received signal sent by the lane reader-writer and returns a response signal added with a detection result to the lane reader-writer; and the lane reader-writer sends a determined MAC (Medium Access Control) address corresponding to the vehicle-mounted electronic label to be stored in the stack of a background computer in order to realize collection transaction. The method disclosed by the invention ensures that the lane reader-writer can determine the corresponding relation of the vehicle-mounted electronic label in temporal order, which not only solves the problem of vehicle following interference, but also enhances the accuracy of vehicle following collection.

The invention belongs to the technical field of tunnel engineering, and discloses a dynamic risk acquisition method for tunnel large deformation disasters, which comprises the following steps: S1, identifying risk disaster-causing factors according to historical data of the tunnel large deformation disasters; S2, selecting a prediction index according to the risk disaster-causing factor, and establishing a probability prediction model according to the prediction index; S3, dynamically updating surrounding rock parameters by using a Bayesian method and a Markov random process method according to the prediction index and the exposed surrounding rock information of tunnel face excavation; S4, according to the surrounding rock parameters, the prediction indexes and the probability prediction model, carrying out posterior risk updating and acquiring a dynamic risk prediction result of the tunnel large deformation disaster. The method solves the problems that in the prior art, a set of quantitative risk evaluation model capable of reflecting a large deformation action mechanism cannot be established, effective feedback of surrounding rock and support information in the construction period cannot be effectively utilized, and a dynamic and informationized risk evaluation theory cannot be formed.

The invention discloses a networked automobile speed prediction method based on space-time sequence information, and belongs to the technical field of automobile intelligent networking. The inventionaims at establishing a future short-time vehicle speed prediction model based on an LSTM neural network after quantitatively analyzing the correlation degree between related characteristics in drivingdata and vehicle speed by utilizing the driving data obtained by an intelligent network connection technology. Finally, the networked vehicle speed prediction method based on the space-time sequenceinformation achieves high-precision prediction of the vehicle speed under all road conditions. The method comprises the following steps: obtaining and processing the correlation degree between an intelligent networked vehicle driving data set and the input and output characteristics of a vehicle speed prediction model, establishing an LSTM neural network vehicle speed prediction model, and training the LSTM neural network model. Accurate vehicle speed preview information is provided for a vehicle control system, and a basis is provided for improving the performance of energy efficiency, safety, comfort and the like of a vehicle.

The invention provides a viewpoint adjustment-based graph convolution loop network skeleton action recognition method and system, relates to the technical field of action recognition, and solves the problem of recognition accuracy reduction caused by different observation visual angles. Utilizing the trained graph convolution recurrent neural network, and taking the preprocessed data as input to obtain spatiotemporal information of the bone data; a Softmax function is adopted, the obtained space-time information serves as input, and a skeletal movement classification result is obtained; the method integrates the advantages of the graph convolution network and the cyclic network, achieves the modeling of the time and space information of the skeleton data, can further improve the accuracy of movement recognition on the basis of an LSTM network movement recognition method, is universal in behavior recognition based on a skeleton data set, and is wide in application prospect.

The invention provides an anti-theft method based on a compass short message function. The anti-theft method based on the compass short message function comprises the following steps: arranging an alarm information receiving address in advance; collecting gesture data of a target device in real time; judging whether a carrier is in a static state according to acceleration data, when the carrier is not in the static state, judging whether the acceleration data and replacement exceed preset values, when the acceleration data and the replacement exceed the preset values, judging that the target device is stolen, and retransmitting alarm information to the alarm information receiving address through a compass communication satellite or a mobile network. The invention further provides an anti-theft system based on the compass short message function. The anti-theft system based on the compass short message function comprises a compass receiving and transmission terminal, a gesture sensor and a controller. The controller is used for storing the preset alarm information receiving address, and judging whether the target device is stolen. The compass receiving and transmission terminal is used for retransmitting the alarm information through the compass communication satellite. The gesture sensor contains compass navigation location, and is used for collecting the gesture data of the target device. The anti-theft method based on the compass short message function and the anti-theft system based on the compass short message function are capable of timely carrying out effective alarms.