217results about How to "Accurate labeling" patented technology

The invention discloses a sample data processing method and device and a computer readable storage medium. The method comprises the steps that a classification model is acquired; automatic annotationof model retraining sample data is performed through the classification model; the model retraining sample data and an automatic annotation result are displayed, the selection of correcting the displayed automatic annotation result is received, and an annotation expansion result of the model retraining sample data is obtained; and the model retraining sample data and the annotation expansion result are fed back to the model for training till the classification model no longer obtains classification performance improvement and iterative optimization of the annotation result corresponding to themodel retraining sample data is performed, and the error correction and leakage finding process of the corresponding annotation result is completed. In this way, annotation accuracy is improved, thesample data is continuously expanded, the corresponding automatic annotation result is obtained after automatic annotation through the classification model, the more precise annotation expansion result is obtained through correction, and the sample data in a large scale and large-quantity wide-dimensional annotation performed on the sample data are obtained.

A medical syringe apparatus includes a syringe unit, a hypodermic needle attached to the syringe unit and at least one interfacing member for mating a labeled medication container to the syringe unit. A filling needle is attached to the syringe unit for drawing fluid from a container. A movable needle sheath protects the hypodermic needle when not in use and may be moved out of the way when administering an injection. The medication label is visible. The container remains mated to the syringe when the syringe is in use. Risk of mislabeling the syringe may therefore be substantially reduced or eliminated and patient safety may accordingly be improved.

A semantic-based approach that classifies Android malware via dependency graphs. To battle transformation attacks, a weighted contextual API dependency graph is extracted as program semantics to construct feature sets. To fight against malware variants and zero-day malware, graph similarity metrics are used to uncover homogeneous application behaviors while tolerating minor implementation differences.

The present invention is a combined image viewing, drawing, and reporting system, which provides: an easy way to access digital images while creating a digital drawing; and an automated method of creating medical reports by translating the drawing data into standard reports. The present invention system provides automated medical drawings and creates medical reports as a result of translating drawing results. The system provides invaluable tools for drawing, diagnosing and treating ophthalmic and/or other types of diseases by generating an automated translation report based on the physician drawing. Customized reports (e.g., a medical drawing report, a referral note, a medical letter, billing based information, a proofsheet for the patient chart, etc.) can be added by the user. A physician can design a format for a report via system utility tools. The present invention may be a stand-alone system, and/or connected to a network.

The invention relates to live-action information system and method thereof based on a GPS positioning and direction identification technology, comprising a live-action information device, a communication network and an interest database. The live-action information device comprises a camera, a GPS positioning module, a horizontal and vertical direction identification module, a display module, a data storage module and a processing control module, wherein the processing control module is connected with the other six modules. The method comprises the following steps of: picking up an external live-action image by the camera, acquiring current GPS coordinate information by the GPS positioning module, acquiring information by adopting the horizontal and vertical direction identification module, analyzing the information to establish searching criteria, identifying, acquiring points of interest and describing data of the points of interest, transmitting the data to the interest database, and overlying and fusing the information to show the describing information of the points of interest. The invention has the advantages of perfect dynamic fusion of live action and data, lively observation and inquiry to surrounding scenes, precise, convenient and visualized information about directions and distances of the points of interest, and rich content of the surrounding points of interest.

The invention relates to an electronic map, in particular to an electronic map provided with roadside panoramic view picture strips, a method for manufacturing the same, and a method for marking interested places. The invention is dependent on road information of the prior electronic map and the correlation between the road information and the panoramic view picture strips for the overlay of the panoramic view picture strips showing both roadsides on roads displayed on the electronic map. Each panoramic view picture strip is a narrow picture which records all the scenes on a section of road, and users can comprehensively, continuously and intuitively know the conditions on both sides of a street through the panoramic view picture strip and determine the actual position of a place in the panoramic view picture strip on the electronic map.

The present invention relates to an image automatic marking method based on Monte Carlo data balance. The method comprises the steps of carrying out the region segmentation on the training sample images in a public image library, enabling the segmented regions possessing different characteristic description to correspond to one marking word, then carrying out the Monte Carlo data balance on the different types of image sets, extracting the multiscale characteristics of the balanced images, and finally inputting the extracted characteristic vectors in a robustness least squares increment limit learning machine to carry out the classification training to obtain a classification model in the image automatic marking; for the to-be-marked images, carrying out the region segmentation on the to-be-marked images, adopting the same multiscale characteristic fusion extraction method and inputting the extracted characteristic vectors in the least squares increment limit learning machine to obtain a final image marking result. Compared with a conventional image automatic marking method, the method of the present invention enables the images to be marked more effectively, is strong in timeliness, can be used for the automatic marking of the large-scale images, and possesses the actual application meaning.

The invention provides a method for establishing a gait data set. The method comprises steps as follows: 1) a test person walks along a straight line in a test area, and wearable sensors fixed on the left foot and the right foot of the test person acquire sensing data of a current walking process sample to serve as gait data of the sample; the wearable sensors comprise an inertial sensor and a sound sensor; 2) gait parameters of the current walking process sample are obtained by the aid of a camera preinstalled in the test area and/or footprints left in the test area; 3) gait data of the corresponding sample are marked by means of the gait parameters obtained in the step 2), and the marked gait data set is obtained. The invention further provides a corresponding gait analysis method based on the labeled data set. The accuracy of precise gait analysis is easily improved on the basis of the wearable sensors; multi-modal data are integrated, and the obtained data set is large in data volume, clear in classification and convenient to analyze and research.

The invention discloses an intelligent optical fiber connecting and distributing system and an adapter port state information identification method. The intelligent optical fiber connecting and distributing system comprises an optical fiber connecting and distributing device, a mobile communication terminal and an information processing center; optical fiber adapters, a microprocessor assembly for forming an electronic tag and a wireless communication module are arranged in the optical fiber connecting and distributing device; the electronic tag is used for storing port state information of each optical fiber adapters and is communicated with the mobile communication terminal by the wireless communication module; and the mobile communication terminal is used for searching the port state information of the optical fiber adapters by accessing the electronic tag and transmitting update data to the electronic tag and the information processing center when the port state information needs to be updated so as to complete the dynamic updating on the port state information of the corresponding optical fiber adapters. The invention solves the difficult problem that the port state information of a conventional optical fiber connecting and distributing device cannot be electronically acquired, so that an optical network cannot implement the intelligent management, and improves the management efficiency of the optical network.

Disclosed are systems and methods for generating data sets for training deep learning networks for key point annotations and measurements extraction from photos taken using a mobile device camera. The method includes the steps of receiving a 3D scan model of a 3D object or subject captured from a 3D scanner and a 2D photograph of the same 3D object or subject at a virtual workspace. The 3D scan model is rigged with one or more key points. A superimposed image of a pose-adjusted and aligned 3D scan model superimposed over the 2D photograph is captured by a virtual camera in the virtual workspace. Training data for a key point annotation DLN is generated by repeating the steps for a plurality of objects belonging to a plurality of object categories. The key point annotation DLN learns from the training data to produce key point annotations of objects from 2D photographs captured using any mobile device camera.

The invention discloses an event element identification method and device, and relates to the technical field of information extraction. According to the specific implementation scheme, the event element identification method comprises: obtaining candidate sentences from a target corpus according to a preset trigger word list, and the candidate sentences comprising trigger words in the preset trigger word list; determining event statements and event types of the event statements from the candidate statements through statement characteristics of the candidate statements by utilizing a rapid text classification model; and according to the event type of the event statement, identifying event elements in the event statement. According to the method, the statement characteristics of the candidate statements are obtained by utilizing the rapid text classification model, and the event statements can be rapidly and accurately screened out, so that event elements in the identified event statements can be more accurate.

The invention discloses a pathological image automatic labeling method based on reinforcement learning and a deep neural network, and the method comprises the following steps: carrying out the multi-scale superpixel division of a dyed standardized original pathological image, and marking the epithelial probability threshold of superpixels; constructing a super-pixel classification training set; training the marked superpixels based on a machine learning method to obtain a learning model, classifying the superpixels in the test image by the learning model, awarding and punishing a classification result by a pathologist, feeding back awarding and punishing results to the learning model, and adjusting the learning model again through an awarding and punishing mechanism; constructing a training set of an end-to-end labeling model by adopting a pre-established method; adopting end-to-end learning based on a deep neural network to automatically mark and segment an organization region; constructing a labeling model; and testing the constructed annotation model by using pre-configured real data. The method has the beneficial effect that the learning models of different tissue regions can be quickly, accurately and automatically marked.

An automatic positioning method of network key resource pages belongs to the Internet information handling sector. The invention is characterized in that: the invention firstly screens out inquiry theme words with timeliness and typicality from a number of search engine user inquiry click information records, extracts the corresponding user click result page and 'user click rate' on the result page, and obtains the inquiry theme collection, key resource page candidate collection, and 'user click distribution'; and then, the invention integrates the 'user click distribution' obtained from user inquiry click information records of different search engines, and obtains the integrated 'user click distribution' through 'inquiry liability' information; at last, the invention judges the pages in the key resource page candidate collection according to the integrated 'user click distribution' so as to obtain the subject-related key resource page. The method has the advantages of being automatic, accurate and objective, and being quickly positioned with computers.

The invention discloses an image annotation method and system based on a weak matching probability canonical correlation model, relating to the technical field of processing of network cross-media information. The image annotation method comprises the following steps: obtaining an annotated image and a non-annotated image in an image database, respectively extracting image features and textual features of the annotated image and the non-annotated image, and generating a matched sample set and an unmatched sample set, wherein the matched sample set contains an annotated image feature set and an annotated textual feature set; and the unmatched sample set contains a non-annotated image feature set and a non-annotated textual feature set; training the weak matching probability canonical correlation model according to the matched sample set and the unmatched sample set; and annotating an image to be annotated through the weak matching probability canonical correlation model. According to the invention, correlation between a visual modality and a textual modality is learned by using the annotated image, keywords of the annotated image and the non-annotated image simultaneously; and an unknown image can be accurately annotated.