36results about How to "Reduce marking costs" patented technology

The invention belongs to the technical field of pedestrian re-identification, and discloses an unsupervised image video pedestrian re-identification method and system based on a migration network, andthe method comprises the steps: carrying out the feature extraction of an image and a video data set in a source domain through employing an improved triple network; training the generative adversarial network by using the source domain data set and the target domain training set; generating a depth feature by using the trained generative adversarial network according to the pedestrian image Itito be identified in the target training set; calculating the Euclidean distance between the depth feature of the image and the depth feature of the video in the target domain; and selecting a video which is closest to the query image, and marking a class mark which is the same as the image. According to the invention, an unsupervised method is used to eliminate a gap between an image and a video,so that the marking cost is greatly saved, and the pedestrian re-identification efficiency is improved.Through carrying out unsupervised deep learning on different modal images and videos, the cross-mode recognition efficiency is effectively improved.

The invention provides a machine learning-based automatic application interface test method and test system. The test method comprises the steps of converting a UI into a picture when it is judged that the obtained UI of a to-be-tested application occurs for the first time; inputting the obtained picture to a preset UI classifier to obtain a type of the UI, wherein the UI classifier is a UI classifier based on a GAN (Generative Adversarial Networks) and a CNN (Convolutional Neural Nework); according to the type of the UI, obtaining a corresponding test outline script from a preset test outlinescript library; matching the test outline script with a layout of the UI to obtain a test script corresponding to the UI; and testing the UI through the test script. The picture tagging cost can be effectively reduced; and the test scene validity and the test efficiency are greatly improved.

The invention provides an intelligent paper inspection system for multiple question types. The system comprises a test paper obtaining module, a test paper preprocessing module, a question judging andscoring module and a result output module, wherein the test paper obtaining module is used for obtaining questions and question answering contents of test paper and standard answers of the questions;the test paper preprocessing module is used for preprocessing the received questions and question answering contents of the test paper: carrying out type classification on each question so as to obtain a question type, and carrying out question answering content segmentation, extraction and format standardization on the question answering content corresponding to each question according to the question type and a preset first rule, so as to obtain preprocessed question answering contents; the question judging and scoring module is used for carrying out question judging and scoring on the preprocessed question answering contents according to the question types and the standard answers and a preset second rule; and the result output module is used for outputting received scoring results. According to the intelligent paper inspection system for multiple question types, automatic paper inspection of multiple question types is realized, so that the working efficiency of paper inspection isimproved and the paper inspection cost is reduced.

The invention provides an electrical chemiluminescence immunoassay method based on gold magnetic particles, which realizes the quantitative detection on the content of antigen/antibody in human serum, blood plasma, cell culture supernatant or other related biological liquid so as to solve the problems of high cost, complicated steps, poor reaction controllability, no generality and the like in the prior art. The method mainly comprises the steps of 1) preparing immunomagnetic beads by gold magnetic particles; 2) labeling the antigen/antibody by Ru(bpy) <2+>3; 3) establishing a standard curve for quantitatively detecting the antigen/antibody; and 4) detecting a sample to be detected. A novel electrical chemiluminescence immunoassay technology is adopted, and the electrical chemiluminescence immunoassay method based on gold magnetic particles has the advantages that the operation is simple and convenient, the valence of antibody can not be affected in the process that the antibody is labeled by Ru(bpy) <2+>3, the labeling cost is low, and the universality is realized.

The embodiment of the invention discloses a multi-scale detection method and device for tumors in gastrointestinal endoscopic images. The method comprises the following steps: carrying out tumor imagemarking on each gastrointestinal endoscopy image sample, and carrying out tumor property grade marking on the gastrointestinal endoscopy image sample marked as a tumor image according to a preset tumor property grade; On the basis of the gastrointestinal endoscopy image sample subjected to tumor property grade marking, obtaining a gastrointestinal endoscopy image detection model by training; Obtaining an endoscope image to be detected; And obtaining a detection result of the to-be-detected endoscopic image according to the gastrointestinal endoscopic image detection model, the detection result comprising one or more of a first classification result, a second classification result and a multi-scale tumor region detection result. By implementing the embodiment of the invention, the markingcost can be reduced, and meanwhile, the tumor region in the gastrointestinal endoscopy image can be accurately positioned.

The invention discloses a laser preparing method of glass black markers. The method includes the following steps of firstly, fixing a graphite sheet to a working table, fixing a glass substrate to the upper portion of the graphite piece so that the glass substrate can make close contact with the graphite sheet; secondly, making laser beams irradiate the interface of the glass substrate and the graphite sheet, making laser beams and the working table for installing workpieces combined and linearly move in the Z direction, and conducting scanning through a galvanometer in the X/Y direction; thirdly, cleaning the glass substrate through an ultrasonic cleaning machine after marking is completed, and removing graphite residues on the surface of glass. The method has the advantages that the black markers of glass are obtained through the graphite back etching method, and the markers of high contrast can be prepared; the marking method is small in grooving heat effect and good in marking quality and has a great effect on increasing the recognition degree of the glass material markers; in addition, through the laser preparing method, the machined markers are resistant to corrosion, high temperature, water and friction and basically cause no feelings when touched by hand, and stability and durability of the markers are effectively achieved; finally, the marking method adopts the graphite sheet as the consumption material, one graphite sheet can be repeatedly used, cost is low and no pollution is caused.

The invention discloses a semi-supervised collaborative training analog circuit early fault diagnosis method, which can classify and identify analog circuit faults without using a large amount of labeled data so as to realize early fault diagnosis. The method comprises the following steps: firstly, carrying out Fourier transform on a known analog circuit normal signal and a fault signal; carryingout spectrum partitioning, empirical wavelet transform on the frequency spectrum; obtaining AM-FM components, carrying out hilbert transformation on the data; obtaining a plurality of EWF components;superposing the energy of the plurality of EWF components; finally, through a semi-supervised collaborative training algorithm, utilizing a small amount of labeled data fully for learning and utilizing a large amount of unlabeled data fully, thus the accuracy of marking actions is guaranteed to a certain extent, a large amount of marking cost is saved, and therefore early fault diagnosis of theanalog circuit based on semi-supervised collaborative training is achieved.

The invention relates to an otolith daily increment labeling method for game fish, and relates to an otolith daily increment labeling method for fish. The invention solves the problems that: the prior labeling method takes time and labor, and has expensive labeling cost, harm to fish bodies and complex operation; and the label is easily lost and the labeling quantity is limited. The method comprises the following steps: controlling the water body temperature in the game fish incubation period so as to change the otolith daily increment appearance feature of the game fish and form an recognizable game fish otolith daily increment label; and setting the water body temperature in the incubation period 2 to 4 DEG C higher or lower than the normal water temperature in the game fish incubation period. The method has the advantages of time and labor reduction, safety, simple operation, lifelong reading, no harm to the fish bodies, and a large quantity of labels. According to the quantity and category of the labeled fish, the process and the method with different labeling spectrums are designed. The application of the method can reach 100 percent of colony labeling.

The invention discloses a hyperspectral image classification method based on correction prototype learning, and the method comprises the steps: selecting a hyperspectral scene image, and randomly extracting a part of samples as a training set; randomly selecting a support set and a query set; building a deep network model, and calculating an initial class prototype in the learning measurement space; building a convolutional neural network with a residual block; training the deep network model; selecting a test data set, randomly selecting supervision samples from the test data set, and performing correction standardization processing on distribution of the supervision samples; and calculating the Euclidean distance between a test set sample and the test class prototype. According to the invention, the hyperspectral image is classified by using the hyperspectral classification method based on correction prototype learning, and the support set and the query set are selected, so that the use of a large number of labeled samples is avoided, and the sample labeling cost is saved. Compared with a traditional prototype network, the method obtains better classification precision, and has important application value in the aspects of hyperspectral image earth surface fine classification and the like.

The invention relates to a method for marking larimichthys crocea by using alizarin red. The method comprises the steps that 1, an alizarin red is prepared for mother solution preparation, a mother solution is diluted to reach application concentration, then the salinity is regulated with refined salt and the like to reach 1.5% or above, and an artificial seawater dye liquor is obtained; 2, larimichthys crocea fry to be dyed is soaked in the dye liquor for dyeing, the fry is transferred to natural seawater for rinsing after dyeing is completed, and then larimichthys crocea marked with the alizarin red is obtained. By the adoption of the method, the influence of calcium ions on alizarin red dissolution and precipitation are avoided, the effectively concentration and a dip dyeing effect of the alizarin red are improved, and the problem of massive dead due to the fact that suspending precipitates easily block the gills of the larimichthys crocea is solved.

The invention discloses an automatic marking device for cylindrical neodymium iron boron magnetic steel. The automatic marking device comprises a supporting table, a vibration material feeding disc, a material feeding hose, a first optical fiber sensor, a guide channel, a rotating disc, a supporting disc, a workbench, a driver, a code spraying machine, a material discharging pipeline, an air blowing pipe and a material receiving hose used for packaging the cylindrical neodymium iron boron magnetic steel. The automatic marking device for the cylindrical neodymium iron boron magnetic steel has the advantages that the cylindrical neodymium iron boron magnetic steel can be conveyed after being erected, so that polarity marking can be achieved through an existing assembly line type automatic code spraying marking technology, the marking efficiency is high, the marking cost is low, the automatic marking device is suitable for batch production, made marks are uniform, the cylindrical neodymium iron boron magnetic steel is guaranteed to have high consistency, the labor intensity of workers is low, and the cylindrical neodymium iron boron magnetic steel can be automatically packaged after being marked, the position of the cylindrical neodymium iron boron magnetic steel will not be placed reversely, the actual polarity of the cylindrical neodymium iron boron magnetic steel after follow-up magnetization is guaranteed to be the same as the marked polarity, and the production qualification rate of the cylindrical neodymium iron boron magnetic steel is increased.

The invention relates to a mark generation method and device, a model training method and device, a marking method and device, a storage medium and equipment. The mark generation method comprises the following steps: acquiring a to-be-marked graph of a target object; obtaining a conversion graph corresponding to the to-be-marked graph; acquiring a reference map of the target object; associating the conversion graph with the reference graph to obtain a conversion relation or a transformation relation; and mapping the fiducial mark based on the conversion relation or the transformation relation to obtain a mark. According to the technical scheme, automation of image marking is achieved.

The invention discloses a sample classification method based on active learning and a neural network, and belongs to the field of machine learning in intelligent science and technology. According to the method, the uncertainty of a neural network model to sample points is taken as a reference, three traditional uncertainty indexes of Least content, Margin and Entry are calculated respectively, the three indexes are used for voting samples, the sample with the highest vote number is the finally screened sample points, the sample points are the most uncertain samples of the model, and then training of the neural network model is facilitated. The number of sample points needing to be marked can be effectively reduced, the marking cost is reduced, and the classification precision of the model is improved.

The invention provides a rice plant carbon isotope field in-situ calibration system, which comprises the following components: a carbon marker release device, which comprises a motor, a screw, a sliding member and an injector, wherein the motor is in transmission connection with the screw and is used for driving the screw to rotate, the sliding part is connected with the screw rod and the injectorand is used for sliding under the driving of the screw rod so as to extrude and release the carbon marker in the injector; a marking box, wherein the marking box is hollow, an opening is formed in the bottom of the marking box, and the marking box is communicated with the injector through a carbon marker release tube; and a motor controller connected with the motor through a first signal line andused for sending a signal for driving the screw to rotate to the motor. The invention further provides a rice plant carbon isotope field in-situ calibration method. According to the rice plant carbonisotope field in-situ calibration method, 13C in-situ calibration is conducted on rice plants under the field in-situ condition, the process is controllable, measurement is accurate, human interference is reduced, and the 13C marking cost is reduced.

A method of manufacturing a light-emitting diode module comprises the steps of: coupling an electronic device with a circuit board such that a distinguishing resistor is coupled adjacent to a light-emitting diode; baking the circuit board by a reflow oven; and testing the baked circuit board, whereby the related information of the light-emitting diode including the manufacturer, the product cycle, the product code, and the grade can be obtained via the distinguishing resistor. Accordingly, the marking accuracy will not be affected by the subsequent processes and the marking cost can also be reduced.

The invention discloses a breast cancer data classification method based on a single category. According to the breast cancer data classification method, breast cancer data classification is effectively carried out only by utilizing single-class sample data and unlabeled sample data; a probability graph model is adopted to describe the relation among the data characteristics, the observation tagsand the real labels, and a corresponding joint probability distribution model is given, and then an EM algorithm is used for solving the model, and then test data are classified. The breast cancer data set used in the breast cancer data classification method is only composed of a single-class sample and an unlabeled sample, so that the sample labeling cost is reduced, and the problem that negative-class sample data are lacked in breast cancer diagnosis is solved.

The embodiment of the invention discloses a model training method, a model training device, a storage medium and electronic equipment. According to the embodiment of the invention, the method comprises steps of after a to-be-trained model of a source domain, a first image set and a second image set of a target domain are obtained, training the to-be-trained model by taking each first image as an input and taking an object recognition frame corresponding to each first image and the category of the object recognition frame as training targets, meanwhile, taking each second image as an input, obtaining outputs of the second images in different layers of the to-be-trained model so as to train the attention discrimination model and the feature discrimination model, thereby responding to convergence of a detection loss function of the to-be-trained model, an attention loss function of the attention discrimination model and a feature loss function of the feature discrimination model, and obtaining a training result;determining the trained to-be-trained model as a target model. According to the embodiment of the invention, the to-be-trained model can be trained without marking the target domain image, the object detection capability of the model for the target domain is improved, and the marking cost is reduced.