58results about How to "Accurate labeling" patented technology

The invention discloses a road disease detection system and method based on multi-source data fusion. The system comprises a data acquisition module, a data preprocessing module, a disease detection module, a three-dimensional modeling module, a data analysis module and a maintenance module. The data acquisition module is used for acquiring road surface data and sending the acquired data to the data preprocessing module; the road surface data comprises road data, road disease data, road vehicle data, non-motor vehicle data and pedestrian data, compared with the prior art, more comprehensive and more accurate detection of road diseases is realized by fusing multi-source data; by means of multi-dimensional data acquisition of radar data, video data, three-dimensional point cloud data, geographical location information and the like, the limitation of a single data source is overcome; the data preprocessing module can effectively remove noise and abnormal values, accurately classify and mark diseases, and fuse data collected by different sensors, so that the quality and integrity of the data are improved.

The invention relates to the technical field of image processing, and provides a training data set labeling method and device for AI visual identification, and the method comprises the steps: obtaining and processing original image data, extracting feature mapping, inputting a target detection model to obtain a preliminary labeling result, carrying out the analysis, and determining a clustering center and a candidate target set; the candidate target set is classified based on the clustering center, the confidence degree of the classification labeling result is calculated, labeling data with the confidence degree higher than a threshold value serve as high-confidence-degree labeling data, rechecking and consistency checking are conducted on data with the confidence degree lower than the threshold value, and consistency labeling data are generated; and the target detection model is optimized in combination with the high-confidence data, the low-confidence data is re-labeled by using the optimization model, and the high-confidence data and the re-labeled data are combined to generate a complete labeled training set to train the recognition model. Through the above scheme, the consistency and confidence of the data are ensured, and the problems of inconsistent labeling and missed labeling and false labeling in complex scenes and multi-target recognition are effectively improved.

The invention provides a driving behavior marking method and device based on an end-to-end automatic driving large model. The method comprises the following steps: acquiring a multi-view image of a current driving scene; obtaining multiple groups of query vectors corresponding to the multi-view image based on a pre-trained end-to-end automatic driving large model; and obtaining a prompt word text, and obtaining a labeling result of the current driving behavior according to the prompt word text and the plurality of groups of query vectors. According to the method, a plurality of groups of query vectors output by a plurality of functional modules in an end-to-end automatic driving large model are taken as one of the triggering conditions of driving behavior marking instead of only taking an image as the triggering condition of driving behavior marking, so that accurate and effective marking of the automatic driving behavior can be realized; the labeling result and the intermediate output result of the large automatic driving model have high consistency, so that the labeling precision and interpretability of the automatic driving behavior are improved.

The invention discloses a line equipment defect labeling method based on an unmanned aerial vehicle power line image, and relates to the technical field of defect labeling, and the method comprises the following steps: obtaining a historical picture of a power transmission line, and carrying out the preprocessing of the historical picture; performing feature extraction on the preprocessed historical picture by using an image segmentation technology to obtain a line image feature value; manually marking defect pictures in the historical pictures to obtain the defect pictures, and marking the line image feature values of the defect pictures as defect image feature values; and taking the defect image feature value and the line image feature value as a training set, and performing model training by using a convolutional neural network to obtain a power line defect identification model. According to the defect identification algorithm, various defects of the power line equipment can be accurately detected and labeled, and the detection rate and accuracy of equipment faults are improved.

The invention relates to the technical field of industrial intelligent manufacturing, in particular to a flexible production line appearance defect optical detection method. Comprising the steps of performing image acquisition on semi-finished products and finished products on a production line; carrying out preprocessing and spatial transformation on the collected image; performing feature extraction and clustering analysis on the positive sample image of the to-be-detected product by using unsupervised learning to form a feature library; comparing the feature vector of the to-be-detected image with a feature library, and outputting the defect position and type of the sample; an improved RTDETR model with supervised learning is adopted, a sample judged as a defect is used for training, and the improved RTDETR model adopts a ResNet and Visual Transform network structure; and performing real-time detection and result feedback on the deployment model. According to the technical scheme, the appearance defect detection efficiency and the product quality of products manufactured on the flexible production line can be improved.

The invention discloses an industry dynamic tag identification method combining big data and ROBERTa, and belongs to the technical field of AIOT, and the method comprises the steps: collecting text data related to the industry from a plurality of big data sources through an automatic means, carrying out the cleaning and preprocessing of the collected data, preparing the data for subsequent processing, and carrying out the recognition of the industry dynamic tag. Through the big data processing technology and the RoBERTa pre-training language model, efficient and accurate label identification of industry data is realized, and powerful support is provided for industry analysis, market prediction and the like. According to the method, the universality of big data and the deep understanding ability of RoBERTa in text processing are combined, and the accuracy and efficiency of tag identification are effectively improved.

The invention relates to the technical field of computer security, in particular to an APT attribution analysis method and system based on large model semantic transformation. The big language model can mine key features of software gene information and spontaneously form association mapping of feature elements in different semantic spaces; extracting a training data set and an evaluation data set for multiple times, training and evaluating the large language model based on the training data set and the evaluation data set extracted each time, updating a hyper-parameter candidate set through a hyper-parameter adaptive selection mechanism, and taking an optimal hyper-parameter obtained after model convergence as an optimal hyper-parameter combination of the large language model; and based on the optimal hyper-parameter combination, full-amount data set task fine tuning is carried out on the large language model by using sample information, so that the large language model learns APT attribution analysis reasoning logic. According to the method, the accuracy and efficiency of APT attribution analysis can be improved, the generalization ability of the model is enhanced, and the method can be applied to multiple application scenes such as network security defense and malicious software analysis.

The invention is suitable for the technical field of building structure recognition, and provides a building structure recognition method and device and electronic equipment, and the method comprises the steps: obtaining the information of a building structure, and the information of the building structure comprises the contour information of a wall body and / or a hole; wall body section contours with the same elevation are determined according to the information of the building structure; a DCEL structure is constructed according to the wall body section contours with the same elevation; determining all surfaces corresponding to the DCEL structure; the surfaces corresponding to the walls and / or the holes are removed from all the surfaces corresponding to the DCEL structure, the surfaces of the space in the DCEL structure are obtained, the surfaces corresponding to the walls are determined according to the contour information of the walls, and the surfaces corresponding to the holes are determined according to the contour information of the holes. Through the method, the marking efficiency and consistency can be improved.

The invention discloses an automatic data labeling method and device and a storage medium, relates to the technical field of data processing, and is used for automatically labeling data and improving labeling efficiency and labeling quality at the same time. The method comprises the following steps: manually pre-labeling data of each task to obtain pre-labeled data; automatically deriving a plurality of different cue word schemes for each task; according to the cue word schemes, all tasks are imported into a multi-mode large model for independent question asking, and first output results corresponding to the cue word schemes are obtained; performing result extraction by adopting an unsupervised voting algorithm to generate a voting statistical result of each task; if the number of the positive correlation marks reaches or exceeds an election threshold value, judging that the task is a positive correlation sample; inputting the pre-annotated data and the positive correlation sample into a single-mode small model for training iteration; and automatically labeling the target data to be labeled in batches by using the single-mode small model, and outputting a labeling result.

The invention relates to the field of intelligent assistance, and particularly discloses a gastrointestinal surgical operation assistance system and method, and the system comprises the steps: obtaining a gastrointestinal CT scanning image of a target patient object, carrying out the image preprocessing of the gastrointestinal CT scanning image through employing an image recognition and processing algorithm based on computer vision and AI, and obtaining a gastrointestinal CT scanning image; and then, extracting a trunk semantic feature and a boundary semantic feature from the preprocessed image, and intelligently generating an image semantic segmentation result containing the marked tumor boundary based on fine-grained enhanced fusion representation between the two features. Therefore, small changes which are difficult to identify by human eyes can be captured, so that tumor boundary information can be identified and labeled more accurately, and intuitive visual assistance is provided for doctors. In addition, the automatic processing flow significantly reduces the manual analysis time of doctors, and improves the diagnosis speed and efficiency.

The invention discloses a method for generating a particle accumulation body simulation model of natural three-dimensional shape characteristics of a building material. The method comprises the following steps: step 1, manufacturing a natural particle simulation model; the randomly generated three-dimensional coordinate control points are endowed with natural sphericity and roundness to obtain a natural particle simulation model; 2, making a particle accumulation body simulation model; the natural particle simulation model obtained in the first step is filled into a discrete element block mass, gravity deposition is applied to the discrete element block mass through a discrete element method, and a particle accumulation body simulation model is obtained. A solid foundation is provided for exploring the microscopic geometrical characteristics of the contact particle accumulation body and the macromechanical properties of civil engineering materials, and a new way is provided for efficiently generating a particle accumulation body database with accurate contact information.

The invention relates to the technical field of image recognition, in particular to a roadway crack recognition method and system based on image recognition. The method comprises the following steps: acquiring roadway structure data; performing regional camera deployment according to the roadway structure data to obtain roadway camera deployment data; performing roadway image acquisition according to the roadway camera deployment data to generate a roadway time sequence image set; performing roadway form change analysis on the roadway time sequence image set to generate roadway time sequence change data; performing surface crack area identification on the roadway time sequence image set through the roadway time sequence change data to generate a roadway surface crack area image; and performing surface crack trend analysis on the roadway surface crack area image to generate surface crack texture trend data. According to the method, through dynamic change detection, hidden crack identification, crack classification and analysis and crack age deduction, the identification accuracy and comprehensiveness are improved.

The invention provides a graph neural network construction method and device, electronic equipment and a storage medium, and relates to the technical field of biological information. The method comprises the following steps: encoding single cell data in a single cell data set into a low-dimensional submerged space to obtain a first feature vector which retains key features and / or type specificity information of the single cell data; obtaining a constructed gene regulatory network, and encoding the gene regulatory network into low-dimensional-factor nodes; under the condition that the global network structure of the gene regulation and control network is reserved, extracting local topological features of the low-dimensional factor nodes to obtain a second feature vector; the second feature vector is used for reflecting interaction among all genes in the single cell data set; constructing a graph neural network based on the first feature vector and the second feature vector; and the graph neural network is used for automatically labeling the single cell data. The problems that cell labeling is inaccurate, time and labor are wasted, and the cell labeling is limited by the batch effect in the prior art are solved.

The invention relates to the technical field of lumbar vertebra segmentation, and particularly discloses a small sample lumbar vertebra segmentation method based on space prior information and prototype iteration updating, comprising the following steps: obtaining a lumbar vertebra MR image sample set comprising a plurality of sample lumbar vertebra MR images; constructing a small sample lumbar vertebra labeling model, and training the small sample lumbar vertebra labeling model by using the lumbar vertebra MR image sample set to obtain a trained small sample lumbar vertebra labeling model; and inputting a to-be-labeled lumbar vertebra MR image into the trained small sample lumbar vertebra labeling model for labeling so as to obtain a lumbar vertebra labeling prediction result of the to-be-labeled lumbar vertebra MR image. According to the small sample lumbar vertebra segmentation method based on space prior information and prototype iteration updating, rapid, accurate and high-throughput lumbar vertebra labeling can be realized.

The invention relates to the technical field of image processing, and provides an image labeling method, device and equipment, a medium and a computer program product.The method comprises the steps that based on the position relation between a virtual capsule endoscope and a gastral cavity model, space pose information of the virtual capsule endoscope is determined; the gastral cavity model is created based on historical gastral cavity data; establishing an association relationship between the virtual gastral cavity image and the spatial pose information; comparing a real gastral cavity image with the virtual gastral cavity image, and determining spatial multi-dimensional data of the real gastral cavity image based on a comparison result and the association relationship; and marking the real gastral cavity image based on the spatial multi-dimensional data to obtain multi-dimensional image marking information. According to the method, by labeling the multi-dimensional spatial data for the gastral cavity image, the workload of personnel is reduced, the data labeling efficiency of the gastral cavity image is improved, more comprehensive and accurate data are labeled for the gastral cavity image, and accurate construction of gastral cavity analysis related algorithms is facilitated.

An augmented reality-based multi-person collaborative guidance and docking method for aircraft wings includes steps such as inter-frame transformation matrix calculation and 3D reconstruction, creation of a high-precision pose estimation dataset, construction of a deep learning framework, wing axis instance segmentation based on the FCN model, wing axis pose estimation method based on the DenseFusion network, multi-terminal world scene construction, wing axis synchronization method and access control, and visualization method for multi-person collaborative guidance and docking. This method applies visual and image processing methods to component pose estimation and proposes a multi-functional approach integrating wing axis pose estimation, 3D registration, multi-person collaboration, and visual guidance. Utilizing deep learning and image processing methods, the wing axis pose can be estimated, improving both efficiency and accuracy. The method is scalable and portable, easily adaptable to different component types. It also offers high visualization capabilities, improving assembly line efficiency and enhancing the overall intelligence of the production line.

The invention discloses an off-line conference recording box and a voice recognition system thereof, belongs to the field of voice recognition, and aims to solve the problems of accurately and efficiently capturing and recognizing voices of a plurality of speakers and reducing the influence of environmental noise on recognition accuracy. The method comprises the following steps: capturing an audio signal from a microphone array to form audio data, receiving a voice audio signal through voice activity detection and dynamic volume adjustment, and providing early warning feedback when an audio quality problem is detected; the method comprises the steps of capturing audio data, performing data processing on the captured audio data, performing de-noising processing on original voice audio signals in the audio data, extracting audio features, and performing voiceprint recognition through clustering analysis of the audio features so as to obtain sound source separation signals with voiceprint information marks; and performing independent voice recognition according to the sound source separation signal, and organizing and managing voice data.

The invention provides a method and system for constructing a high-quality data set for a large model in the urban rail field, and the method comprises the following steps: collecting and summarizing data meeting business requirements, and enabling the collected data to serve as an input data source; data is cleaned through data preprocessing, so that the quality of the data is improved; labels are added to the preprocessed data, so that key features of the data can be better understood and learned; and performing quality auditing and dynamic management on the annotated data through data maintenance. The invention provides a method and a system for constructing a high-quality data set for a large model in the urban rail field. The method and the system cover multiple aspects of data acquisition, preprocessing, labeling, maintenance, safety and the like. The systematization and standardization degree of the whole data processing process is improved, and the data quality and the model training effect can be improved.

The present invention provides a system and method for marking engineering drawings, which relates to the field of engineering design technology. The system includes: a marking classification module, which is used to collect marking information, classify the marking information, and obtain the type of the marking information; a marking management module, which is used to obtain the classified marking information and the corresponding type sent by the marking classification module, and update the marking information; wherein the update includes an update based on a user update request and an update based on an error update request; a marking module, which is used to obtain the marking information sent by the marking management module, and integrate the marking information into a marking tool. The present invention can update the marking tool function, meet the requirements of designers and enterprises for the diversity, accuracy and design cycle of marking design, and improve the efficiency and quality of enterprise digital design.

The present application discloses a method, device, system and target object detection method for generating training samples. The training sample generation method includes: obtaining a point-annotated image and a frame-annotated image, each point-annotated image includes an object to be annotated and a target pixel corresponding to each object to be annotated, each frame-annotated image includes an annotated object and a target annotated frame corresponding to each annotated object, and the number of point-annotated images is greater than the number of frame-annotated images; determining a target prediction frame corresponding to the object to be annotated based on the target pixel; annotating the object to be annotated based on the target prediction frame to obtain an annotated point-annotated image; generating a target training sample based on the annotated point-annotated image and the frame-annotated image, and the target training sample is used to train a target detection model to detect the object in the image to be detected. The present application solves the technical problem of high annotation cost in the prior art when annotating training samples.

The invention relates to the technical field of AI multi-modal large models, and discloses an automatic line labeling method based on an AI multi-modal large model, and the method comprises the steps: data set and data processing: firstly, searching two-dimensional and three-dimensional drawing data sets in a targeted manner; converting the three-dimensional structure diagram of the drawing into a two-dimensional three-view diagram; carrying out linear, radius, diameter and angle automatic labeling and manual checking on a two-dimensional three-view graph and a two-dimensional drawing which are based on the three-dimensional structure graph characteristics; performing structured text processing on the two-dimensional drawing, verifying data legality, and removing and repairing data with poor quality; a two-dimensional drawing scheme is segmented, the whole scheme is converted into a plurality of N blocks with fixed sizes, and then structured texts of all the blocks are compressed. The automatic line labeling method based on the AI multi-modal large model has universality and can be widely applied to the field of engineering design, engineers are helped to rapidly and accurately label drawings, the design efficiency and quality are improved, and the problems that the requirement for developers is high, and the customization cost is high are solved.

The invention provides equipment for synthesizing data, and relates to the technical field of data synthesis, and the equipment comprises a picture screening module which is used for dividing to-be-labeled data into pictures difficult to label and pictures easy to label through the module, enabling the pictures difficult to label to enter a manual adjustment module, then enabling the pictures difficult to label to enter a model updating module, expanding a training set, and retraining a network, the pictures easy to label enter an automatic labeling module; according to the equipment for synthesizing the data, automatic annotation of pictures can be realized, universality can be realized to a certain extent by updating a model function, and the efficiency of image annotation is greatly improved; the pictures can be screened more comprehensively through a proposed query strategy, so that the required model performance can be achieved with a smaller number of annotation samples; by providing an automatic labeling method based on instance segmentation, the problem that overlapped pictures are difficult to label is solved.

The present invention provides a system and method for the segmentation and annotation of photographed document images combined with unsupervised learning, wherein the method comprises the following steps: extracting image segmentation and annotation data and then performing grayscale conversion of traditional image processing, and then obtaining matrix position coordinate data through unsupervised target segmentation model processing, which is used for unsupervised image annotation to obtain the result contour data of photographed document segmentation. Next, the traditional image processing data and the result contour data are received, and the pixel intersection and union ratio algorithm is used to compare and traverse the pixel points to determine whether they belong to the segmented area. The inaccurately annotated data is transmitted to the data annotation module for re-annotation. Finally, the accurately annotated data is received and stored in the data server. The whole process realizes the functions of image segmentation, annotation, inspection and storage. Through the scheme of the present application, the training data of photographed document image segmentation can be enriched, and the problem of poor effect of deep learning image segmentation method in complex situations can be solved.

The invention discloses a method for detecting and locating a high-speed motion filter rod burst bead. First, a high-speed motion filter rod burst bead image is collected, the image is preprocessed, an edge detection method is used to obtain a preliminary pixel point area of ​​a target, a clustering algorithm is used to obtain a foreground class and a background class area, and an intersection operation is performed with the previously obtained target area to determine the accurate pixel point area of ​​the target, and then the image is annotated, a sample training set, a verification set and a test set are reasonably generated, a deep convolutional neural network including an SPP structure and an SE module is built, the fusion of feature maps of different sizes is completed, and an adaptive weight is given to each channel to achieve accurate detection and positioning of the target. The method of the invention combines edge detection and image clustering to accurately locate the target area, remove interference and facilitate annotation, and adjusts the method for obtaining the target area and the network structure according to different sample types. After the sample is trained, a high detection rate can be guaranteed to achieve accurate positioning of the target.

The invention discloses an image annotation method and device, equipment and a storage medium. The method comprises the steps of determining a to-be-processed image; in response to detection of a first labeling operation and a second labeling operation on a to-be-labeled object in the to-be-processed image, determining a first labeling point and a second labeling point respectively corresponding to the first labeling operation and the second labeling operation; detecting the movement position of the cursor at each moment; and under the condition that a third labeling operation on the to-be-labeled object is detected, displaying a labeling box for labeling the to-be-labeled object based on the first labeling point, the second labeling point, the moving position at each moment and the third labeling operation. The labeling frame determined by the labeling method is the rotating frame, and the size, the position and the rotating angle of the rotating frame are determined according to the to-be-labeled object, so that the labeling frame can better fit the to-be-labeled object, and the labeling frame does not need to be adjusted after labeling is completed; therefore, image annotation can be accurately and quickly performed, and the annotation efficiency and the annotation accuracy of the image are improved.

Embodiments of the present specification provide a sample processing method and device, wherein the sample processing method comprises: obtaining a to-be-processed sentence; constructing a word group corresponding to the to-be-processed sentence, and determining a feature label corresponding to each word unit in the word group; constructing at least one negative sample and a target positive sample according to the feature label and the word group; calculating a sample similarity between the target positive sample and each negative sample, and determining a target negative sample in the at least one negative sample according to the sample similarity.

The invention discloses a method for generating a magazine cover or inside page layout based on a large language model. The method is characterized by comprising the following steps of: a) generating text descriptions of elements and position coordinates on a magazine page in batches by utilizing the large language model; b) generating elements one by one by using Hyper-SD according to the description content; c) carrying out weak supervision labeling on the generated graph of the single object by using PP-LiteSeg; d) combining the single elements into a complete magazine page according to the generated element position description, and combining the weak supervision labels of the single elements; e) verifying the generated annotation by using the segform corrected by the self-annotation data set; and f) evaluating the labeled generation data based on crowdsourcing. Compared with the prior art, the method has the advantages that the generation speed is high, the generated image has accurate annotation, the generated image meets the aesthetic appreciation of human beings, and the like, and has a good application prospect.

The invention provides a respiratory tract tumor cell identification method and device based on an artificial intelligence algorithm. The method comprises the following steps: acquiring a respiratory tract tumor cell fluorescence staining picture, slitting the respiratory tract tumor cell fluorescence staining picture, performing artificial intelligence identification on sub-pictures and splicing the sub-pictures. According to the respiratory tract tumor cell identification method based on the artificial intelligence algorithm provided by the invention, through the respiratory tract tumor cell identification neural network model, rapid identification of the respiratory tract tumor cells on the respiratory tract tumor cell fluorescent staining picture is realized, the identification efficiency of the respiratory tract tumor cells is greatly improved, and the identification accuracy is improved. The diagnosis efficiency of the respiratory tumor cells is improved; meanwhile, the method realizes the marking of the respiratory tumor cells based on the artificial intelligence network model, thereby facilitating the realization of automatic operation, and effectively reducing the manpower diagnosis cost of the respiratory tumor cells.

The utility model discloses a labeling machine and a labeling drilling cutting material production line, the labeling machine comprises a rack, a cross beam, a first driving module, a second driving module, a material bearing table top, a printing module, a labeling module and a blowing module, the first driving module can drive the cross beam to move back and forth on the rack in the first direction, and the second driving module can drive the cross beam to move back and forth on the rack in the second direction; the printing module comprises a printer body and a label receiving plate installed below a label outlet of the printer body, the printer body discharges printed label paper from the label outlet to the label receiving plate, the label suction device sucks the label paper from the label receiving plate from bottom to top, and the label receiving plate is provided with air holes. The air blowing module can blow the label paper upwards through the air holes. Due to the fact that the airflow provided by the air blowing module counteracts part of gravity of the label paper, the attaching degree between the label paper and the label receiving plate is reduced, and the phenomenon that the label paper and the label receiving plate are attached too tightly due to the negative pressure phenomenon is greatly reduced.

This application provides a labeling and handling robot, relating to the field of sheet labeling technology. In this labeling and handling robot, a sheet fixing unit is fixed to a sheet handling frame. The sheet fixing unit is used to contact the sheet to be labeled and to fix the sheet to be labeled below the sheet handling frame. A sheet handling mechanism is used to move the sheet to be labeled to a predetermined position after the sheet fixing unit fixes the sheet to be labeled. A labeling mechanism is fixed to the side of the sheet handling frame facing the sheet to be labeled. The labeling mechanism is used to affix a label to the sheet to be labeled when it moves to the predetermined position. This application embodiment can automatically realize the handling and labeling of sheets, effectively reducing labor costs and labor intensity, improving labeling efficiency and achieving accurate labeling. Furthermore, it can integrate sheet handling and labeling, significantly improving sheet processing efficiency and equipment costs.