48results about How to "Rich scene" patented technology

The invention discloses a text detection method of a document image in a natural scene. Commonly used Chinese characters are selected to make Chinese character pictures, a dataset 1 is formed, randomrevolving and cropping operations are carried out on labeled document images, then a manner of Poisson cloning is used to fuse different background images, and a dataset 2 is formed; the dataset 1 isadopted to carry out training of a text classification model on a VGG16 network, and after the model converges, obtained parameters are used to initialize a fully convolutional neural network model, and the dataset 2 is used to train the model; the trained fully-convolutional neural network model is used to process the image, a classification situation of each pixel point is obtained according toa maximum probability method, and a text-non-text binary image is formed; a method of connected regions is used to obtain text regions, the original image is binarized, and only text information in the text regions in the text-non-text region binary image is extracted to obtain a text binary image; the image is corrected through a maximum variance method; and projection is carried out again on thecorrected image, and the text-non-text region binary image is refined.

The embodiment of the invention discloses a data mining method of a terminal user social network, a correlation method, a device and a system, which are characterized in that the user identification of a terminal is obtained; then, on one hand, communication records of a user are obtained by utilizing the user identification, and first social information is obtained according to the user identification and the communication records; on the other hand, a data packet of social network service visited by the user is obtained according to the user identification, and second social information is obtained through the user identification and the data packet; on the third hand, information distributed by the user on an internet can also be obtained according to the user identification, and then, the third social information is obtained according to the user identification and the distributed information; and finally, a user social database is established or updated by utilizing the first social information, the second social information and the third social information to form the social network with rich scenes, detailed contents and complete information, thereby conveniently improving the efficiency of various services required to be operated based on the social network.

The invention discloses a method and device of path automatic generation of 2D / 3D RPC game scenes in mobile platform. The method includes following steps: initialize 3D scene model, plan paths, generate main paths and generate branch paths etc. The device includes following modules: path planning module, main path module, branch path module, loop path module, obstacle module and pre-processing module. The invention simplifies 3D scenes to be 2D map of terrain. Then, the map can be represented as 2D paths.

The invention discloses a thermal runaway monitoring device and method for a power lithium ion battery pack. The thermal runaway monitoring device comprises a first unit, a second unit and a third unit which are sequentially connected, wherein the first unit is used for acquiring a voltage signal, a temperature signal and a current signal of a battery pack and transmitting the acquired voltage signal, temperature signal and current signal of the battery pack to the second unit; the second unit is used for monitoring the thermal runaway condition or risk of the battery pack in real time througha set algorithm according to the temperature signal, the voltage signal and the current signal of the battery pack acquired by the first unit; and the second unit is also used for periodically sending a voltage signal, a temperature signal and a current signal of the battery pack to the third unit through the communication bus. In combination with other structures or methods, the defect that a power lithium battery management system in the prior art lacks an advanced prediction and early warning method and function for battery thermal runaway is effectively avoided.

The invention discloses a data processing method, apparatus and device, and a storage medium. The method comprises the steps of sending an obtaining request of a uniform resource locator; interceptingthe obtaining request of the uniform resource locator, and sending the obtaining request to an agent tool; enabling the agent tool to obtain the uniform resource locator according to a plug-in interface provided by a package capture tool; sending a query request of analog data matched with the uniform resource locator; and when the analog data matched with the uniform resource locator is queriedthrough the agent tool, outputting the analog data. The coupling is reduced in front and back end development to realize parallel development, so that the overall front end development efficiency is improved; and the analog data for a back end is accurate and stable, and scenes are rich, so that the development efficiency is further improved.

The invention discloses a virtual upper limb control system based on myoelectricity and motion capture and a method of the system. The system comprises a myoelectric device, a motion capture device, avirtual game development environment and a virtual / augmented reality device, wherein the virtual game development environment comprises a virtual upper limb model and a virtual game scene; the myoelectric device is used for detecting myoelectric signals and gesture information of different gestures and sending the myoelectric signals and the gesture information to the virtual upper limb model; the motion capture device is used for acquiring joint coordinates and spatial posture information of upper limbs and sending the joint coordinates and the spatial posture information to the virtual upper limb model. The virtual game scene receives data sent by the virtual upper limb model and sends the data to the virtual / augmented reality device, the virtual / augmented reality device feeds back visual auditory stimulation to the user, and virtual objects are placed in a corresponding real environment for operation. The system has the advantages of being low in cost, high in precision, comfort degree, timeliness and interactivity and diversified in scene, has an upper limb integrated game pattern and can increase the game enthusiasm of users.

The invention discloses an attention training system based on virtual reality. The attention training system comprises a brain wave music playing module, an eye tracker calibration module, an evaluation module, an eye movement data acquisition module, an electroencephalogram data acquisition module, a data processing module, an automatic training module and a report module. VR is adopted, countless spaces and various styles of scenes can be virtualized, and children can focus attention more easily. Due to the immersion and imagination of VR, children can be better immersed in the training game. In a VR environment, a child focuses on an object through an eye tracker, watches the object and focuses on the object within a specified time, and the object is moved to a specified position; children can directly focus on attention and focus on attention by watching an object with eyes. According to the invention, the efficiency of restoring training scenes, clearing scenes and configuring scenes by doctors can be greatly reduced, and the working intensity of doctors is reduced.

The invention relates to an interface automatic test method and system based on an interface automatic test system. The interface automatic test system is connected with the development terminal, thetest terminal and the tested terminal in a communicatable manner. The method comprises the following steps of: 1, receiving a test case execution request from a test terminal; wherein the test case execution request comprises a test case body and the information for requesting the tested terminal to execute a test case; step 2, analyzing the required information of the test case from the test caseexecution request and sending to the tested terminal; step 3, after the test case is executed at the tested terminal, receiving the execution result from the tested terminal; step 4, recording the test case and the execution result thereof for calling by the test terminal and / or the development terminal.

The embodiment of the invention discloses a simulation management method, device and equipment for railway vehicle service receiving and dispatching service. The simulation management method comprises: constructing a railway vehicle service receiving and dispatching service simulation library according to given station yard graph data, interlocking table data, running chart data and receiving anddispatching operation standard data; obtaining identity information of a target operator; selecting a target simulation operation test question corresponding to the identity information of the targetoperator from the railway train service receiving and dispatching service simulation library; providing the target simulation operation test question for the target operator, and obtaining operation information data of the target operator for the target simulation operation test question; and obtaining evaluation information of the target operator according to the operation information data and standard operation data of the target simulation operation test question. According to the invention, analog simulation can be carried out on railway vehicle service receiving and dispatching services,the application range is wide, the scenes are multiple, the exercise content is rich, and the human intervention is little.

The invention relates to an H5-based electronic coupon issue and verification-cancelling system and an issue method thereof. The electronic coupon issue and verification-cancelling system comprises an electronic coupon management device, a mobile terminal, multiple issue servers, a retailing device and an order management device, wherein the electronic coupon management device is used for generating an electronic coupon; the mobile terminal is used for displaying the electronic coupon; all the issue servers are arranged between the mobile terminal and the electronic coupon management device and used for forwarding the electronic coupon to the mobile terminal; the retailing device is connected with the electronic coupon management device and used for scanning the electronic coupon displayed on the mobile terminal to acquire discount information; and the order management device is connected with the electronic coupon management device and the retailing device and used for recording and saving sale logs based on the electronic coupon. Compared with the prior art, all issue channels can access the electronic coupon through URLs according to respective recognition data, therefore, it is not needed to develop a separate interface for each issue server, and a third-party channel can be quickly connected into the issue capability of the electronic coupon management device.

The invention discloses an SUMO traffic simulation scene accurate construction method based on data conversion. The method is based on existing actual measurement road network data and a lane level road network model. Data conversion, self-organization and reprocessing are carried out on basic object element information forming SUMO road network sub-files, a relevant conversion tool is combined, effective and available SUMO road network files are created and generated and used for SUMO simulation, and therefore accurate construction and control of a simulation scene are achieved. According tothe method, the problems of road network topological structure errors, low execution efficiency and the like caused by complex data formats and complex workloads after conversion processing are solved.

The embodiment of the invention relates to an image processing method and device, a video stream scene switching method and device, a virtual three-dimensional scene fusion method and device, equipment and a medium. In at least one embodiment of the invention, a virtual three-dimensional scene is determined through a pre-shot image and depth information corresponding to the image, so that the virtual three-dimensional scene is distinguished from a scene of a target image; and then a main body in the target image is fused with the virtual three-dimensional scene to obtain a fused image. According to the invention, the scene of the target image is switched into the virtual three-dimensional scene under the condition that the main body is not changed, so that the scene is richer; and the two-dimensional scene of the target image is switched into the virtual three-dimensional scene, so that the scene is more vivid.

The invention belongs to the technical field of rail transit, and relates to an urban rail transit full-scene passenger flow prediction method and system. The current passenger flow analysis technology cannot provide all-around support for urban rail transit operation organization. The invention provides the urban rail transit full-scene passenger flow prediction method, which comprises the following steps: analyzing the travel characteristic rule of a single passenger in an urban rail transit system, and analyzing the influence of external conditions on the individual travel behavior of the passenger; monitoring whole-network passenger flow distribution in real time, starting from a time dimension, realizing a full-scene urban rail transit passenger flow prediction technology, giving an alarm prompt under an abnormal condition, and providing passenger flow data support for passenger flow organization adjustment or driving organization adjustment for emergency disposal; generating the corresponding statistical indexes by taking landmarks compiled by metro and the recently developed national standard of Urban Railway Traffic Operation Index System as references, and supporting user-defined report query and data export. Reference is provided for urban rail transit passenger flow control and train operation scheduling.

The embodiment of the invention discloses a music sharing method, system and device, electronic equipment and a storage medium, and the method comprises the following steps:entering a lyric video template display interface related to a target song when an instruction related to the display of a lyric video template is triggered; generating a lyric video according to a video editing operation of a user on the lyric video template display interface; and in response to a video publishing instruction of a user, publishing the lyric video to a target position. According to the technical scheme disclosed by the embodiment of the invention, the problems that in the prior art, the music sharing mode and scene are relatively single, and the requirement that the user shares the music-related, active and rich emotion expression cannot be met are solved, the music expression requirement of the user can be amplified, the user is helped to more actively and richer express the music-related emotion, and a streaming media product can be helped to complete more social media transmission, so that more high-quality new users can be obtained.

The invention discloses an omni-channel chain operation background system for quick response, which comprises an intelligent middle platform, a cashier terminal, a database, a server, an updating module and an intelligent middle platform coupon cloud platform, and is characterized in that the intelligent middle platform is a unified platform oriented to commercial ecology and is used for releasingthe association relationship between each ERP and CRM of the background and the front end; the cashier terminal is used for customer consumption and terminal data management; the server side can be in butt joint with a plurality of databases such as Oracle, DB2, Sqlserver and Mysql to realize data transmission; the system has the beneficial effects that a powerful promotion model and an operationengine support most promotion modes in the retail industry, the business matching degree is high, the adaptability is high, and the operation performance is good; a perfect coupon model and operationand pre-calculation engine supports most coupon marketing scenes on the market, and is high in adaptability, rich in scene and good in operation performance; and different types of interface butt joint modes are provided, so that different systems can be integrated efficiently.

The invention provides a database self-learning optimization method based on a flow mirror image, and the method comprises the steps: obtaining a latest database configuration parameter of a to-be-optimized database through a self-adaptive learning method, and sending the latest database configuration parameter to a training database; establishing a database mirror image model consistent with thedatabase model of the database to be optimized, and sending the database mirror image model to the training database; establishing a flow mirror image model consistent with the flow model received bythe to-be-optimized database, and sending the flow mirror image model to the training database; obtaining a training result fed back to the self-training database; updating database configuration parameters of the to-be-optimized database according to a training result of the training database; and repeating the above steps for R times, obtaining the latest database configuration parameter when the TPS throughput of the training database in the R training results is the maximum value, and configuring the latest database configuration parameter as the final database configuration parameter to the to-be-optimized database. For a real traffic scene, the performance of the database can be effectively improved.

The invention discloses an intelligent cloud platform for pedestrian analysis. The intelligent cloud platform comprises a face analysis engine which is used for carrying out the analysis of the identity characteristics and attribute characteristics of a face, a pedestrian analysis engine which is used for analyzing the attribute characteristics of each part of the pedestrian, a hybrid analysis engine which is used for performing unified analysis modeling and model storage on the face and pedestrian data under the characteristic space-time scene, a training engine which is used for continuouslyupdating the model performance of the cloud platform, and the training input data is in an automatic and manual mode, and a cloud computing architecture which is used for an SaaS architecture and anarchitecture of a software engineering part of data caching, storage and request agency. For the intelligent cloud platform for pedestrian analysis, the face recognition technology and the pedestrianre-recognition technology are combined; far and near pedestrians in a scene are detected, recognized and structurally analyzed; and the integrated identity recognition based on the face and pedestriancharacteristics can improve the accuracy of pedestrian recognition and can better adapt to richer scenes and user requirements.

The invention relates to the technical field of image processing, and in particular to a leaf shielding detection method based on deep learning and a traditional algorithm. The method comprises: inputting the training set into a convolutional neural network model, training the convolutional neural network model until a loss function of the convolutional neural network model meets a set condition,and terminating the training of the convolutional neural network model, the trained convolutional neural network model being an optimal classification model; and judging whether the test set image isan image without leaf occlusion or not by adopting a traditional algorithm, and if the traditional algorithm detects that the image is an image with leaf occlusion, performing further detection on theimage by adopting a deep learning algorithm. Whether the monitoring camera is shielded by leaves or not can be detected, a worker can conveniently know the condition that the monitoring camera is shielded by the leaves in time so as to adjust the position of the monitoring camera, missing of a monitoring scene is prevented, and the influence on public safety is reduced.

The invention discloses a flood-proof household safety monitoring system with multiple emergency measures. The flood-proof household safety monitoring system comprises a household safety monitoring system server, a central node, a routing node, a water sensor node, a set top box, a user terminal and a broadcasting device, wherein the water sensor node indirectly communicates with the central node by the routing node in a wireless way or the water sensor node directly communicates with the central node; the central node receives the information from the water sensor node and then uploads the information to the set top box; the set top box transmits the information to the household safety monitoring system server by an internet; after the household safety monitoring system server receives the information, a short message is sent to a user and informs the user that flood occurs in the house and the user needs to check in time; an alarming signal is given by the broadcasting device, people in the house can find the flood as soon as possible; an instruction is sent to close a water valve. The user can find the flood in time, the economic and personnel loss is avoided, the water resource waste caused by the flood is avoided, the scene in the house can be more perfectly monitored by the household safety monitoring, and the requirements of the modern people on the household environment monitoring can be more conformed to.

The invention provides an unmanned training data classification method and device and electronic equipment. The method comprises the steps of acquiring N data sets, wherein the N data sets comprise data in collected data samples, each data sample comprises data of N types of attribute information, the attribute information of the data belonging to the same data set is the same, and N is an integerlarger than 1; clustering the N data sets one by one, obtaining M data categories of the finally clustered data set in the N data sets, wherein data samples corresponding to the data in different data categories are used for training different prediction models, the clustering of the (K + 1) th data set is carried out based on the clustering result of the Kth data set, K is a positive integer smaller than N, and M is an integer larger than 1. Thus, the efficiency of training the prediction model can be improved, and the accuracy of the prediction model obtained through training can be improved.

The invention discloses a multi-screen remote display system and method. The system comprises a system control module, an encoding and decoding module, a video controller module, a screen module, a content control module and a content display module; the encoding and decoding module and the screen module are connected with the video controller module; and the content display module is connected with the content control module. By adoption of the multi-screen remote display system and method disclosed by the invention, the comprehensive system cost is reduced effectively, and meanwhile the maintenance amount and the maintenance cost can also be reduced.

The invention relates to a video connection method and device, and the method comprises the steps: obtaining a video connection request set at a preset matching moment, carrying out the pairwise matching of video connection requests in the video connection request set, and obtaining at least one group of matched video connection requests, creating a video connection room corresponding to each group of mutually matched video connection requests, generating identification information of the video connection rooms, and respectively sending the identification information of the video connection rooms to clients corresponding to the mutually matched video connection requests corresponding to the video connection rooms, and sending the identification information of the video connection room to the corresponding client, so that the corresponding client enters the video connection room for video connection according to the identification information of the video connection room and acquires public information of the opposite side during video connection. By a mode of matching the video connection request, the client performs video connection in a one-to-one video connection scene, so thatthe privacy and the adaptability of video social contact can be improved.

The invention discloses a test method and a test system based on depth estimation performance of a depth camera. The test method comprises the following steps of constructing a scene, and erecting thedepth camera and a laser total station; storing the color image and the depth image; the laser total station scanning the field to obtain a projected color image and a depth map; evaluating an overlap degree projection error between the projected color image and a color image acquired by a camera; if the projection error does not meet the requirement, carrying out calibration again; and if a projection error meets the requirement, comparing and evaluating the depth map acquired by the camera and the depth map projected by the total station pixel by pixel, and calculating the error condition of depth estimation of the depth camera. The method is advantaged in that the true value of the depth map is obtained through field scanning of the laser total station, joint calibration of the cameraand the total station and point cloud back projection, more comprehensive evaluation is conducted by directly comparing the depth map, meanwhile, rich scenes are provided, physical equipment is not limited, and higher precision is achieved.

The invention relates to an underwater image real-time restoration model based on a self-attention mechanism and a GAN, and belongs to the technical field of deep learning. A generative adversarial network is used as a basic framework in the invention, the generative network adopts a coding-decoding structure, a synthetic image is generated at an original resolution through nine residual blocks and deconvolution operation, a self-attention module is introduced, richer advanced features can be captured to improve model performance, and in order to keep image content and remove underwater noise at the same time, the discrimination network adopts a structure of a multi-branch discriminator comprising an adversarial branch and a criticality branch. The self-attention module is added, the model structure is improved, the problem of low time efficiency of underwater image processing is solved ideally, and the method based on GAN-RS is trained to adapt to various underwater conditions, and has excellent real-time processing performance.

The invention discloses a VC-OTN high-quality government and enterprise private line opening method and device, and the method comprises the steps: setting a VC-OTN opening module in a super controller SC to meet a high-quality government and enterprise private line opening application scene, and enabling the VC-OTN opening module, a PTN opening module and an SPN opening module to become a core module in the super controller SC together, so that mainstream application scenes including standard government and enterprise special line bearing, 4G service bearing, 5G service bearing and high-quality government and enterprise special line bearing are comprehensively supported.

The invention discloses an information processing method and an electronic device. The method comprises the following steps: obtaining voice data collection by at least a part of voice collection units in at least one voice collection unit; analyzing the collected voice data to obtain an analysis result, and performing processing based on the analysis result to obtain response information of the voice; and outputting the response information through a voice output unit in the same space as the at least a part of voice collection units.

The invention provides a training data generation method, a training data generation device, a computer readable medium and electronic equipment, and relates to the technical field of image processing. The method comprises the following steps: acquiring a video set, wherein the video set comprises at least one 2D video; performing frame screening on each 2D video based on a preset tracking condition to obtain a tracking frame sequence corresponding to each 2D video; and performing three-dimensional reconstruction based on each tracking frame sequence to obtain a reconstructed image and depth data corresponding to the reconstructed image, and taking the reconstructed image and the depth data corresponding to the reconstructed image as first training data. According to the method and the device, videos which exist on the Internet, are shot by different devices and are rich in scenes can be converted into deep truth values, and training data with a large quantity and rich scenes can be generated, so that machine learning or deep learning networks related to the deep data can be trained based on the data.

The invention relates to the technical field of image processing, in particular to a method for detecting leaf occlusion based on deep learning and traditional algorithms. Input the training set into the convolutional neural network model, train the convolutional neural network model until the loss function of the convolutional neural network model meets the set conditions, terminate the training of the convolutional neural network model, and the convolutional neural network model after training The product neural network model is the best classification model. The traditional algorithm is used to judge whether the test set image is an image without leaf occlusion. If the traditional algorithm detects that the image is occluded by leaves, a deep learning algorithm is used to perform a one-step detection on the image. It can detect whether the surveillance camera is blocked by leaves, which is convenient for staff to know in time that the surveillance camera is blocked by leaves, so as to adjust the position of the surveillance camera, prevent the lack of surveillance scenes, and reduce the impact on public safety.

The invention discloses a wire twisting device for elevator traction steel wire rope production, and relates to the technical field of elevator production and manufacturing. The wire twisting device for elevator traction steel wire rope production comprises a workbench, wherein a vibration aging mechanism, a lubricating mechanism, a wire twisting mechanism, a translation mechanism and a transmission mechanism are arranged on the workbench, a threaded rod is rotationally installed on the outer surface of the top of the workbench, a limiting disc is fixedly installed on the outer wall of the threaded rod, three sets of steel wire rollers and a set of fixing discs are installed on the outer wall of the threaded rod in a threaded mode, the three sets of steel wire rollers are attached to one another, and the fixing disc is located above the three sets of steel wire rollers. According to the wire twisting device for elevator traction steel wire rope production, stress left on a steel wire is reduced in a vibration mode before wire twisting operation, so that the steel wire is not prone to rebound and deformation in subsequent wire twisting operation, the wire twisting quality of the device is guaranteed, compared with a method for reducing stress by heating a workpiece, the device has the beneficial effects of being efficient, capable of saving energy and safe, and convenient popularization and application is facilitated.