2511results about How to "Reduce manual intervention" patented technology

The invention provides an image denoising method based on generative adversarial networks, and belongs to the technical field of computer vision. The method comprises the following steps: (1) designing a neural network for estimation for noise intensity of an image containing noises; (2) using image blocks in an image library to add noises of the intensity according to the estimated noise intensity to use the same as samples of training the networks; (3) in network training, designing a new generation network and discrimination network, and adopting a form of fixing the generation network to train the discrimination network and fixing discrimination network parameters to train the generation network to enable the networks to carry out adversarial training; and (4) using the trained generation network as a denoising network, and selecting a network parameter according to a result, which is obtained by the noise recognition network, to denoise the image containing the noises. The methodhas the effects and the advantages that a visual effect of the denoised image is improved without the need for manual intervention for adjusting the parameter, and texture details of the image can bebetter restored.

A fork-lift truck includes a load carrier, a first optical detector, an optical analysing unit, and a second optical detector. The analysing unit receives and analyses information from the second optical detector, wherein the first and second optical detector are movable together with the load carrier, and the first optical detector is positioned in relation to the load carrier, such that a first field of view includes at least a part of the load carrier and an extension of the load carrier. The second optical detector is positioned, such that a second field of view includes a predetermined volume above the load carrier The position and orientation of the second detector is calibrated in relation to a predetermined position on the fork-lift truck, in case one or more objects are positioned in the predetermined volume, at least one three-dimensional relative position can be determined.

The invention discloses an integrated control method of cold-rolling strip steel flatness and lateral thickness difference, comprising: moderately and mainly controlling the shape of the strip steel section on the upstream rack of a cold-rolling mill, and mainly controlling strip steel flatness on the downstream rack; based on a profiler configured on a No.1 rack inlet of the cold-rolling mill, actually measuring and hot rolling incoming material section shape and flatness and rolling technological parameter measured values of each rack, wherein the rolling technological parameter measured values comprise rolling force and a plate shape regulating mechanism measured value; actually measuring the cold-rolling strip steel flatness on the basis of a shape meter configured on the cold-rollingmill outlet; performing feed-forward control to the shape of the strip steel section by taking the upstream rack, in particular the No. 1 rack as a key point; and performing feedback control to the strip steel flatness by taking the downstream rack, in particular an end rack as a key point to realize integrated control to the cold-rolling strip steel flatness and lateral thickness difference. Theinvention performs integrated coordination control by the strip steel flatness and section shape and reduces cold-rolling strip steel lateral thickness difference on the basis of further improving cold-rolling strip steel flatness quality.

The invention discloses a method and module for extracting and interpreting information of a remote-sensing image, and provides a geographical element object based automatic interpreting method based on SRMMHR and random forest. The module extracts and interprets the information of the remote-sensing image by the adoption of the method. The method and module for extracting and interpreting the information of the remote-sensing image can achieve effective segmentation of the remote-sensing image, automatic optimization of characteristics, and automatic establishment of a classification rule set; a means which is little in manual intervention, high in degree of automation and high in interpreting precision is provided for extracting and interpreting the information of the remote-sensing image.

The invention belongs to the warehouse operation intellectualization technical field, and more specifically relates to a warehouse intelligent operation execution system and method. The warehouse operation execution system based on intelligent objects comprises RFID labels attached to cargos, trays and warehouse goods allocations and used for respectively identifying and storing cargo, tray and goods allocation information; RFID reader-writers arranged at inlets and outlets of fork trucks and warehouses and used for reading information stored in the RFID labels; and a fork truck intelligent terminal rear end system used for performing information interaction with the RFID reader-writers, storing RFID label information read by the RFID reader-writers in a database, distributing storage locations for the trays, and distributing tasks for fork trucks. The warehouse operation execution system and method can manage warehouse cargo incoming and outgoing operation processes, reduce manual intervention and increase work efficiency.

The invention provides a sentence similarity calculation method and system. By using a word2vec algorithm, a pre-built corpus is trained to obtain vectors of all words in the corpus; performing intelligent word segmentation on two sentences to be subjected to similarity calculation; finding vectors corresponding to each segmented word in the first sentence and the second sentence from the corpus; sequentially calculating the similarity between each segmented word of the first sentence and each segmented word of the second sentence; obtaining two groups of segmented word sets with the segmented word similarity exceeding a preset threshold; calculating the similarity contribution value of each group of segmented words in the whole sentence according to the deviation quantity of each group of segmented words in the position of the sentence; and adding the contribution values of the segmented words in the two sentences to obtain the similarity between the sentences. The method and the system provided by the invention have the advantages that the semantic similarity of words is calculated by word2vec; and through mass corpus automatic training, convenience is provided for accurate information retrieval, file classification or system answering.

The invention relates to a laser sintering sand, a preparation method thereof, a sand core and the preparation method thereof. The laser sintering sand consists of a first component and a second component which are evenly mixed, the first component consists of a first type of raw sand with the coefficient of angularity of less than 1.3, thermosetting resin and additives, wherein the thermosetting resin and the additives are coated on the surface of the first type of the raw sand; and the second component consists of a second type of raw sand with the coefficient of angularity of less than 1.1, a thermoplastic material and the additives, wherein the thermoplastic material and the additives are coated on the surface of the second type of the raw sand. The adoption of the laser sintering sand does not need to add an auxiliary support structure during the laser sintering process, thereby improving the molding efficiency and reducing the manual intervention; and the any complex sand core can be manufactured without using a mold, and the sand core simultaneously has high sintering initial strength and can meet the requirements on the thermal strength and the gas-generating capacity of the sand core during the metal casting process.

The invention relates to the technical field of biomedicine, and relates to a biomedicine event trigger word identification method based on characteristic automatic learning. The biomedicine event trigger word identification method comprises the following steps of 1, data pre-processing; 2, construction of an event trigger word dictionary; 3, construction of candidate trigger word examples; 4, characteristic learning by means of a convolutional neural network model; 5, training by means of a neural network model; and 6, classification of event trigger words. The biomedicine event trigger word identification method is advantaged in that 1, complex preprocessing to data is simplified, and tedious steps for carrying out a characteristic design by people are saved; 2, domain knowledge is introduced, and a lot of external resources such as unlabeled linguistic data are effectively utilized; 3, characteristic automatic learning is carried out by means of a convolutional neural network, manual intervention is reduced, sentence level characteristics in a deeper level can be excavated and explored, through the fusion of local characteristics, implicit global characteristics are discovered, and the category of trigger words can be identified; and 4, a better experiment result is obtained in MLEE linguistic data, and the whole performance on event trigger word detection is improved.

The invention relates to a method for optimizing an orbital transfer strategy of a geostationary orbit satellite, which comprises the following steps of: 1, determining orbital transfer times, orbital transfer circle times and the controlled variable of each-time orbital transfer; and 2, determining time and a thrust direction in each-time orbital transfer. The process of launching the geostationary orbit satellite at present generally comprises the following steps of: launching the satellite into a highly elliptic transfer orbit with an inclination angle by using a carrier rocket; performing apogee/perigee orbital transfer for several times by using a self-contained liquid engine of the satellite, and transferring to a geosynchronous orbit; and correcting and rounding the inclination angle of the orbit to realize a geostationary orbit. For the satellite, operation for changing the transfer orbit into the geostationary orbit by performing apogee/perigee orbital transfer for several times is complex, so too many orbital transfer times is not suitable, and orbital transfer complexity and risk are prevented from being increased; in addition, factors such as the capacity of the liquid engine of the satellite, arc segment loss in an orbital transfer period, and the like are considered, so too few orbital transfer times is not suitable.

The invention discloses a cleaning robot and a control method thereof and belongs to the field of smart home. The method comprises: acquiring a marked position of a charging pile in a map, wherein the map is a map drawn by the cleaning robot; controlling the cleaning robot to move to a position in front of the marked position and identifying the charging pile; if the charging pile is not identified, generating a searching path based on the marked position; controlling the cleaning robot to move forward according to the searching path, and identifying the charging pile in a movement process. By adopting the cleaning robot and the control method thereof, the problems that an actual position of the charging pile is different from the marked position in the map and the cleaning robot cannot find the charging pile and cannot perform pile returning action are solved and the effects of improving the intelligent degree of the cleaning robot and reducing manual intervention are realized.

The invention discloses a mobile phone advertising method based on a user consumption feature vector, and belongs to the field of mobile phone data mining and analyzing. The method includes the steps of establishing a user keyword to represent a vector according to browsing contents of a user mobile internet; establishing a user consumption feature vector of a corresponding user according to the browsing contents of the mobile internet; extracting a keyword from an advertisement experience text database and setting a keyword weight, and constructing a characteristic matrix for each user feature vector; computing each component value of the user consumption feature vector of each user; setting the mapping relations between a user consumption tag and each component of the consumption feature vector of the user within different threshold ranges; labeling a corresponding consumption tag for each user and then conducting advertising according to each component of the current consumption feature vector of the user and the mapping relation. The mobile phone advertising method can improve precision of advertising, reduces manual interference, and effectively lowers platform running cost.

A light field microscopical method based on FPM includes the following steps of building a light field microscopical platform based on FPM, collecting high-definition wide-vision images through the light field microscopical platform based on FPM, carrying out view separation on the obtained high-definition wide-vision images, and obtaining wanted results through refocusing by means of results of view separation. Light field information is collected by means of a light field microscope based on an FPM algorithm, definition of each microlens image in the light field microscope is improved, angle definition is improved, the collected light field information is enriched, and a better three-dimensional structure can be reconstructed for an object.

The invention relates to an automatic fluid preparation system and an automatic fluid preparation machine used for preparing intravenous fluids. The automatic fluid preparation system comprises an ampoule bottle conveying device, a penicillin bottle conveying device, an ampoule bottle breaking device, an injector arrangement device, a shaking up device, an infusion bottle conveying device and a multifunctional manipulator. Through the organic combination of the above devices, the fluid preparation is automatically completed in class 100 clean sterile environment without manual operations, so that safety and convenience are increased. In addition, after medical staff inputs an instruction sheet for fluid preparation, the automatic fluid preparation system can automatically check the types of configured medicine bottles, if differences occur, a dispensing action is stopped and an alarm is generated. The automatic fluid preparation system carries out a dispensing program after the medicine staff arranges the medicine which needs to be prepared, the whole process is carried out in a clean environment, so that manual intervention is reduced and the purpose of clean and safe production is achieved.

The invention discloses an SQL (Structured Query Language)-based database management method and device. The method comprises the steps of: compiling a corresponding SQL script by aiming at each database management task item to form an SQL script set; establishing a logic mapping file of each database management task item with a corresponding SQL script; and deploying the SQL script set and the logic mapping file at a position available for access of a database management client side. According to the invention, by configuring the mapping relation between the database management task items andthe corresponding SQL scripts, the automatic batch execution and the result analysis and storage of the database management task items can be realized, so that manual interventions in a database management process can be reduced, the execution efficiency can be increased and the maintenance cost can be reduced.

The invention discloses a power transmission line sag on-line monitoring method. The method includes: firstly installing an on-line monitoring device on the lead between two adjacent towers and close to one tower, using a laser range finder to measure the vertical distance from the tops of the two adjacent towers to the ground and the horizontal distance between the two adjacent towers, and recording data; energizing the power transmission line, and adopting a lead mutual inductance energy acquiring way for the on-line monitoring device; configuring the measured data into the on-line monitoring device by means of GPRS or a handset; measuring the inclination angle between one lead end and the horizontal plane by an inclination angle sensor; employing an accurate sag calculation method to calculate the height from the lowest point of the lead to the ground by a microcontroller in the on-line monitoring device; composing the position, sag, equation and inclination angle information into a data frame by the on-line monitoring device, and sending the data frame to a remote server or internet through GPRS. The method and device provided by the invention performs mathematical modeling on the power transmission line, accurately calculates the maximum sag value, and is conducive to on-line monitoring of the sag.

The invention discloses a parking lot entrance guide system of a driverless vehicle and a guide method using the system. The guide system comprises a mobile internet client terminal, a vehicle guide cloud platform, a vehicle-mounted unit arranged on a guided vehicle, a barrier gate and an image information collection device, wherein the barrier gate and the image information collection device are arranged at a parking lot entrance. The guide method comprises the following steps: triggering the guide system; determining a vehicle initial state of the guided vehicle via the vehicle guide cloud platform; calculating vehicle control parameters; carrying out vehicle control and simultaneously carrying out on-line simulation of vehicle driving via the vehicle guide cloud platform; carrying out an interrupt operation when an interrupt condition is triggered, or performing the next step; and exiting guide when an ending condition is achieved.

The invention discloses a logistic processing system and a logistics processing method in direct connection with an e-commerce platform. According to the logistic processing system and the logistics processing method provided by the invention, a client side is directly connected with the e-commerce platform through a port to acquire confirmed electronic order data information in the e-commerce platform and generate destination codes and next-level transfer codes accordingly; and the client side prints express lists accordingly and transmits duplicate information to a central management unit and a station management unit. The station management unit is used for storing and managing information data of all stations of a logistics company; and the central management unit, which is used for storing and managing data information of the whole logistics processing system, is respectively connected with the client side and data of the station management unit. According to the logistic processing system and the logistics processing method provided by the invention, 40-50% of labour cost in a current logistics express industry can be reduced, unit operation time can be shortened by 2-3hours, work pressure from bottom to up in the current logistics express industry can be relieved, and the whole operating efficiency can be increased by more than one third.

The invention provides a direction field distribution-based fingerprint correction method, which comprises the following steps of: making statistics on local modes and space distribution in the ridge directions of fingerprints in a cluster learning method, searching for an area similar to a fingerprint mode in a target image in a local mode comparison mode, eliminating the influence of the local modes at irrational positions by combining the spatial position of each local mode, estimating the positions and the directions of the fingerprints, and cutting and correcting the fingerprints according to the information of the positions and the directions. Fingerprints at nonstandard postures are corrected into fingerprints at standard postures to improve fingerprint comparison and query efficiency; all fingerprints can be automatically detected and cut from a background-complex field fingerprint image with a plurality of fingerprints, and the postures of the fingerprints are corrected; and the method has the advantages of high calculation efficiency and high accuracy.

The invention discloses an automatic matching method between a point of interest and a road network. The road attribute correlation in points of interest and the road network are made full use of to correct system deviation between points of interest from different sources and the road network; through building a road network skeleton map and a point of interest skeleton map, a geometric pattern correspondence relation between the point of interest and the road network is dug, a point-line homonymous control point recognition problem between the point of interest and the road network is converted into a skeleton map matching problem, and automatic position deviation correction between the points of interest from different sources and the road network data is carried out based on a skeleton map crossing point matching relation. The relative position precision of each space data can be greatly improved, manual intervention is reduced, and automatic and precise matching between the point of interest and the road network is realized.