124results about How to "Reduce labeling costs" patented technology

The invention discloses a domain-adaptive retrieval enhancement generation method and system, and the method comprises the steps: carrying out the document analysis and semantic partitioning of a document through a semantic understanding model and a document rule, and storing document fragments; on the basis of the large language model and a preset question library, generating a question and disassembling the question; performing correlation evaluation and sorting on the initial retrieval results in combination with a retrieval technology; candidate content most matched with the cue words is screened out through a large language model, and question-retrieval fragments are scored and marked; performing fusion on the initial retrieval result and the scoring and labeling result to construct positive and negative samples, and constructing a supervised data set; and based on a preset loss function, utilizing the supervised data set to carry out supervised training on the knowledge semantic search model, replacing the RAG mixed search module, and carrying out continuous iteration. By means of the technical scheme, new domain knowledge and operation feedback data can be flexibly applied, and domain self-adaptive retrieval enhancement generation is achieved.

The invention provides an automatic driving obstacle recognition method and system based on deep learning, and relates to the technical field of automatic driving, and the method comprises the steps: collecting sample image data and sample point cloud data; performing feature extraction on the marked point cloud and the marked image by using a 3DCNN network, and performing training according to an extraction result to obtain a classifier and a prediction head; enhancing the unmarked image and the unmarked point cloud; performing feature extraction on the enhanced image and the enhanced point cloud to obtain a multi-scale image feature and a multi-scale point cloud feature; dynamically adjusting the weights of the multi-scale image features and the multi-scale point cloud features based on a feature fusion method of an attention mechanism to obtain fusion features; and inputting the fused features into the trained classifier and the prediction head to obtain an obstacle recognition result. According to the method, the accuracy, robustness and generalization ability of obstacle recognition are remarkably improved, the marking cost is reduced, and the high requirements of an automatic driving scene for real-time performance and reliability are met.

The invention discloses an encrypted traffic detection method based on LLM enhanced semi-supervised learning. The method comprises the following steps: firstly, training a traffic detection model for detecting encrypted traffic under an LESS framework; and detecting the encrypted traffic by using the trained traffic detection model. Components of the LESS framework include a Prompt orchestrator, an NSSN, and an LLM optimizer. The components work cooperatively, a detection model is trained by using a labeled training set and a label-free training set, and each piece of data in a data set is formed by arranging length values of data packets in a network flow in sequence; the Prompt composer guides the LLM to generate label-free enhanced data in different scenes; the NSSN automatically sets a pseudo label for the enhanced label-free traffic data, and learns traffic detection knowledge accordingly; under the guidance of the SSL loss, the LLM optimizer adjusts the enhancement strategy of the LLM to align the semi-supervised training requirements of the detection model. The training steps of the detection model include LLM pre-fine tuning; marking the LLM enhanced data; and continuous demand alignment can be realized.

The invention relates to an insect pest semi-supervised target detection method combining active learning and multi-scale view learning. The method comprises the following steps: constructing a semi-supervised target detection data set; performing data enhancement on the semi-supervised target detection data set; constructing a semi-supervised model and generating a pseudo label; performing multi-scale view learning through the student model to obtain a multi-scale view prediction result; performing semi-supervised training; performing active learning, and updating the semi-supervised target detection data set; and inputting a to-be-detected field corn image into the finally trained teacher model to obtain a final pest detection result. According to the method, active learning is combined to select some data which is beneficial to the model and can effectively improve the performance for labeling, and the labeling cost is utilized to the maximum extent; adding a multi-scale view learning structure, and capturing effective information; and finally, under the condition of limited 20% annotation data, the detection precision of more than 80% annotation data can be realized, and the annotation cost is greatly saved.

The invention discloses a hole making process parameter recommendation method based on a knowledge graph, which uses an optimized seven-step method, and provides an ontology evaluation system and an ontology updating method, so that a hole making process ontology can be updated and iterated according to different processing requirements; according to the method, the fine-tuning UIE model is adopted to perform knowledge extraction on the hole-making process text data, good extraction accuracy can be obtained only by performing fine-tuning on the UIE model by using small sample data, the marking cost of knowledge extraction of the hole-making process text data is reduced, and the extraction efficiency is improved; according to the method, an entity alignment algorithm and a processing case filtering algorithm based on text similarity and word list mapping are designed, so that the repetition rate of knowledge in the knowledge graph is reduced, and the retrieval efficiency is improved.

The invention relates to a fault detection method based on a multi-modal attention mechanism. The method comprises the following steps: performing data fusion on multiple data sources in an IT system; wherein the multiple data sources comprise log data, monitoring data and user behavior data; inputting the fused data into a fault detection model to obtain a fault detection result; wherein the fault detection model is constructed based on a GRU neural network and a multi-head attention mechanism. According to the invention, by introducing the large language model, the GRU neural network, the multi-head attention mechanism and the multi-modal data fusion technology, the accuracy and reliability of system fault detection can be effectively improved.

The invention belongs to the technical field of natural language processing, and discloses a text data labeling method and system based on weak supervised learning. The method comprises the following steps: performing pseudo label generation on a plurality of historical text data by using a weak supervised learning algorithm to obtain a plurality of source domain data provided with real labels and a plurality of target domain data provided with pseudo labels; constructing a text data labeling model by using a deep learning and adversarial training algorithm, and adjusting the text data labeling model by using a continuous learning algorithm; and labeling the real-time text data by using the adjusted text data labeling model to obtain labeled real-time text data, and updating the adjusted text data labeling model by using a continuous learning algorithm. The problems that in the prior art, manual labeling cost is high, labeling quality is uneven, and dependence on label data is high are solved.

The invention provides an end-to-end automatic driving system and method. The system comprises an environment sensing module, a feature extraction module, a data fusion module, a decision generation model module and a decision analysis module. The environment sensing module comprises a plurality of types of sensors, and each sensor collects environment information data of different modes. The feature extraction module extracts features of each modal environment information data. And the data fusion module fuses the features of the modal environment information data to obtain fused features. The decision generation model module determines a driving decision by using a trained decision generation model based on the fusion features. And the decision analysis module converts the driving decision output by the decision generation model module into a specific vehicle control instruction. The method has higher adaptability to complex and dynamic driving scenes, reduces the cost of data annotation during training, and improves the generalization ability of the model in various driving scenes.

The invention discloses a 3D target detection method and system based on binocular vision and open vocabularies, and relates to the technical field of target detection. The method comprises the following steps: calculating pixel features of each image by adopting an open vocabulary 2D detection model; pixel features of the left view and the right view are aggregated to each three-dimensional point; obtaining a parallax estimation map based on the pixel features of the left and right views; introducing a time aggregation guide parallax consistency constraint to obtain a depth parallax estimation map; converting the depth parallax estimation map into a 3D point cloud, and extracting features to obtain 3D features; and carrying out correlation integration on the text features, the 2D fusion feature vectors and the 3D features by utilizing object-level image guide detection loss, object-level cross comparison alignment and object-level self-comparison learning to obtain a three-dimensional target detection model, realizing three-dimensional target detection and distinguishing different target objects. According to the invention, accurate detection and distinguishing of the target can be realized, so that the flexibility and universality of 3D target identification are improved.

The invention provides a phytoplankton chromatography sequence identification method and a phytoplankton chromatography sequence model building method, and belongs to the technical field of image enhancement identification. The method comprises the following steps: firstly, acquiring microscopic chromatography sequence data of phytoplankton, performing view field extraction and serialization recombination, and constructing a three-dimensional data set; then, constructing a three-dimensional recognition model containing physical perception and a sequence aggregation mechanism, extracting single-frame semantic features by the model by adopting a parameter-shared twin network, and introducing a physical definition prior module to calculate a space-frequency domain quality score of a slice; secondly, designing a deep perception sequence aggregation module, and adaptively aggregating key features of a high signal-to-noise ratio by taking definition scores as gating signals and combining spatial context information between slices; and finally, training and optimizing the model based on the image-level weak supervision label to obtain an optimal model. According to the method, the problems of information truncation and out-of-focus noise interference caused by extremely shallow depth of field of high-power microscopic imaging are solved, and full-depth-of-field stereoscopic perception can be realized under the condition that frame-by-frame fine labeling is not needed.

The invention discloses a multi-modal data enhancement method and system based on an AI training platform, and belongs to the technical field of multi-modal data processing.The method comprises the steps that firstly, a multi-modal data fusion pipeline is developed on the AI training platform, and data of different modals are collected through a data collection module for preprocessing; in the preprocessing stage, a multi-modal data fusion pipeline is used for carrying out feature extraction on data from different modals, and joint modeling is carried out at a specific level; then, automatically generating high-quality synthetic data by adopting a data enhancement pipeline; and finally, developing a multi-modal deep learning pipeline to fully learn and understand deep features of different modal data. According to the method, the problem of model learning bias caused by single-modal data in the prior art is solved, the generalization ability and the data processing efficiency of the model are improved, meanwhile, the processing difficulty of a multi-modal data task is simplified for an AI training platform user, and the method has important application value.

The invention provides a point cloud semantic segmentation method and system based on a dynamic uncertainty active learning strategy, and belongs to the technical field of pavement monitoring, and the method comprises the steps: obtaining a plurality of labeled first super-point samples and a plurality of unlabeled second super-point samples based on a three-dimensional point cloud data sample; training the initial model based on all the first super-point samples to obtain a first training model; predicting each second super-point sample by using the first training model to obtain the actual uncertainty of each second super-point sample; screening all the second super-point samples based on the actual uncertainty to obtain a plurality of candidate super-point samples; constructing a training data set based on all candidate super-point samples; and performing updating training on the first training model based on the training data set to obtain a point cloud semantic segmentation model. Through intelligent screening and optimization of annotation sample distribution, the category diversity is improved, and the annotation cost is reduced, so that the segmentation precision and generalization ability of the model are remarkably improved.

The invention relates to the technical field of image processing and computer vision, in particular to a text-guided multi-view Mars topographic feature image segmentation method based on SAM, and the method comprises the steps: constructing and training an MTSNet model, inputting a to-be-segmented Mars image and a topographic prompt text into the trained MTSNet model, and obtaining a segmentation result; the MTSNet model comprises an MTG-SAM branch, an LTE (Long Term Evolution Network) branch, a feature fusion module, a text prompt encoder and a Mask Decoder; the MTG-SAM branch is formed by cascading a plurality of Transform Blocks, and a TCAM (Ternary Content Addressable Memory) module is introduced into each Transform Block; the text prompt encoder comprises a CLIP text encoder, a text affine layer and a prompt encoder; according to the method, the accuracy and convenience of terrain segmentation of the Mars image are effectively improved.

The invention relates to the technical field related to data processing, in particular to an automatic labeling method, device and equipment based on a large model and a clustering algorithm. The method comprises the following steps: performing preliminary analysis on an original data set by using a natural language processing technology, and generating an initial label for each piece of data in the original data set; performing clustering analysis on the generated initial labels by applying a clustering algorithm, and grouping the initial labels based on similarity or relevance among the labels to obtain a plurality of initial label groups; selecting the most representative label from each initial label group to replace all labels in the initial label group so as to simplify a label system and reduce redundancy; judging the similarity between any two pieces of data in the original data set; and adjusting the label system based on the similarity.

The invention relates to a Fourier contour syntax guided remote sensing target detection method, equipment and a medium, belongs to the technical field of remote sensing image target detection, and can effectively detect obviously symmetrical targets such as airplanes or ships in a remote sensing scene lacking contour / target mask labels. According to the invention, in a small sample scene, detection of a target category and positioning of a target can be realized, and a closed contour of the target can also be output. Fourier low-frequency spectrum is used to realize contour coding, and only 128-dimensional descriptors are used to replace traditional discrete contour point prediction. Through a triple Fourier guidance mechanism formed by cyclic alignment loss, low-frequency amplitude regularization and symmetric constraint, the noise interference of a weak supervision false contour is effectively relieved, the model is forced to preferentially learn the global shape and symmetric structure of a target, and the adaptability to scale change, any rotation angle and complex background interference is remarkably improved. And meanwhile, the detection accuracy in a small sample scene is improved, and practical value and theoretical value are both considered.

The invention provides an electrical equipment defect data labeling method and system based on cross-modal active learning, and belongs to the field of electrical equipment defect data labeling. S1, electrical equipment defect data is converted into multi-modal data, the multi-modal data is mapped to a unified feature space, and a multi-modal classification model is constructed; s2, part of samples are selected for labeling, and an initial training set is constructed; s3, training the multi-modal classification model, inputting an unlabeled sample in the multi-modal data into the multi-modal classification model, and outputting a prediction label; s4, according to an active learning strategy, selecting samples with annotation values from the multi-modal data for annotation to obtain annotation samples, adding the annotation samples into the initial training set, and repeating the steps S3 to S4 to meet annotation requirements or stop conditions; classifying the defect data of the power equipment by using the trained multi-modal classification model to obtain a final labeling result; the data labeling cost and workload are reduced, and the labeling accuracy is improved.

A kind of flow video active interaction control method, equipment and medium for the scene of helping blind, construct the hierarchical task system for the scene of helping blind, train the semantic understanding and behavior alignment of multimodal large model to specific task of helping blind;Introduce the active interaction control mechanism for flow video, under the premise of not changing the bottom model parameter, realize the trigger output of environmental perception by the dynamic suffix switching of determination mode and generation mode;Finally, the incremental context updating strategy is used to optimize the inference process, while keeping the timing consistency and reducing the calculation redundancy.The invention solves the technical problems of lack of special data, passive interaction logic and calculation redundancy of multimodal large model in flow environment through the synergistic effect of data-driven task adaptation and active interaction mechanism.The general framework proposed in the invention provides an important theoretical basis and technical paradigm for the paradigm evolution of flow video understanding from "passive analysis" to "active interaction".

This invention belongs to the field of industrial manufacturing technology, specifically providing a method and system for visual detection of industrial defects by decoupling defect features from imaging conditions. The method includes: collecting defect samples of industrial products under different operating conditions; wherein the defect samples include defect images and corresponding defect annotations; based on the defect samples, constructing a region-guided causal decoupling network model by combining a feature decoupling mechanism guided by defect regions and a causal invariance learning mechanism based on imaging simulation; performing multi-supervised loss joint optimization training on the causal decoupling network model to obtain a converged causal decoupling network model; and detecting defects in industrial products under target operating conditions based on the converged causal decoupling network model to obtain defect detection results. This invention can significantly improve the generalization and robustness of defect detection algorithms in complex industrial environments, enabling rapid and low-cost model transfer between different operating conditions.

The invention relates to medical image data processing, in particular to a medical image report splitting method for large model training, and the method comprises the steps: reading a to-be-split medical image report, and carrying out the sentence segmentation of the medical image report, and obtaining a combination of segmented sentences seen for examination and a combination of segmented sentences of examination impression; extracting word vectors and mask vectors corresponding to each check-seen clause in the check-seen clause combination and each check-impression clause in the check-impression clause combination; inputting the word vector and the mask vector corresponding to each check-seen clause in the check-seen clause combination into a pre-trained check-seen classification model to obtain a check-seen classification result corresponding to each check-seen clause; inputting the word vector and the mask vector corresponding to each check impression clause in the check impression clause combination into a pre-trained check impression classification model to obtain a check impression classification result; according to the technical scheme provided by the invention, the defect that the examination impression in the medical image report is difficult to accurately split according to categories can be overcome.

This invention discloses a target detection method, electronic device, and medium based on knowledge distillation, comprising: inputting a target detection image to be detected into a pre-trained target detection model to obtain target detection results; wherein, the training process of the target detection model includes: acquiring target detection data and constructing it into a labeled dataset and an unlabeled dataset; training a teacher model using the balanced labeled dataset; inputting the unlabeled dataset into the teacher model to obtain confidence-labeled data; combining the unlabeled dataset and the confidence-labeled data, and after filtering, obtaining a confidence dataset; setting a distillation loss function, and training a student model based on the distillation loss function using the confidence dataset to obtain a target student model, i.e., a target detection model; wherein the distillation loss function is the sum of bounding box loss, target loss, and classification loss; wherein the target loss and classification loss are both constructed based on the confidence output by the target teacher model.

The application discloses a kind of man-machine collaborative multi-modal data intelligent labeling and quality checking method, it is related to man-machine collaborative technical field.The method first obtains at least containing two modalities of original data of image, text, audio, generates multi-modal joint data sample after pre-processing and modal alignment;Input pre-trained multi-modal AI pre-labeling model generates pre-labeling result with confidence;Through cross-modal consistency check calculation consistency score, sample is divided into high, medium, low confidence three categories;High confidence sample is directly adopted and enters the sampling pool, and medium and low confidence sample is based on labeling personnel ability portrait and AI model dynamic capability evaluation and carries out intelligent task distribution.The application effectively solves the problem that efficiency and quality are difficult to balance in multi-modal data labeling, significantly improves labeling accuracy and automation rate, reduces labeling cost, and can meet the construction demand of large-scale high-quality multi-modal data set.

The application discloses a material mechanical property inference method based on multi-modal pre-training representation. In view of the problems that the existing method needs to rely on expensive scanning electron microscope images in the inference stage, and the prediction accuracy is low in the small sample scene, the application constructs a multi-modal network containing an image encoder, a table encoder, a shared projection branch and a modal specific projection branch, and carries out joint pre-training through cross-modal alignment loss, intra-modal decoupling loss based on contrast logarithmic ratio upper bound mutual information estimation and table mask recovery loss. Then, the pre-trained table encoder and its auxiliary projection head are migrated to the downstream multi-task regression network, and the mechanical properties can be predicted only by inputting process parameters. The method can also maintain stable performance in the small sample scene of iterative sample collection, significantly reduces the material representation cost, and improves the prediction accuracy and generalization ability.

This application discloses a method for detecting apple disease spores using microscopic data based on multimodal and semi-supervised learning. The method includes: acquiring raw microscopic data of apple disease fungal spores, performing edge detection and texture enhancement to generate texture-enhanced data; extracting features from the raw microscopic data and texture-enhanced data using a dual-branch encoder, and fusing them using a cross-attention mechanism to obtain enhanced visual features; inputting the textual description information of the disease fungal spores into a text encoder for encoding to obtain global text features; aligning the enhanced visual features and global text features across modalities based on a multimodal object detection network, outputting multimodal fused features, and inputting them into a semi-supervised learning framework to train a student-teacher model using labeled and unlabeled data; inputting the microscopic data to be detected into the trained model and outputting the detection results. This method improves the detection accuracy and robustness of microscopic data with extremely low annotation costs.

The invention relates to the technical field of image recognition, and discloses a wind power fan blade defect recognition method and system based on image recognition, and the method comprises the steps: carrying out the blocking cutting, illumination normalization and image enhancement processing of an original image, constructing a defect-free sample and a defect sample, and dividing the samples into a training set and a test set; the method comprises the following steps: taking a GANopen network as a basic framework, fusing the lightweight design of Mamba-YOLO, constructing a joint loss function by adversarial loss, reconstruction loss and coding loss based on a training set, carrying out unsupervised training, optimizing network parameters until convergence, and obtaining a defect identification model; inputting the preprocessed to-be-detected fan blade image into the trained defect recognition model, performing defect recognition and positioning, and outputting the defect type and position; based on an output result of the defect identification model, dynamically adjusting an early warning level and response measures through a self-adaptive early warning mechanism; according to the invention, the efficiency and accuracy of wind power fan blade defect identification are improved.

The application discloses a benthic animal rare species image recognition method and system based on small sample learning, relates to the technical field of computer vision and artificial intelligence, and comprises the following steps: S100, data preparation and preprocessing; S200, small sample training based on meta learning; S300, domain self-adaptive data enhancement; S400, small sample rapid adaptation based on meta learning; S500, attention-guided difficult example mining; S600, online incremental learning mechanism; through the collaborative design of meta learning and a multi-level feature fusion network, the common species data is used for meta training, and cross-species general visual knowledge is learned; the new rare species can be rapidly adapted through a small amount of labeled samples and several gradient updates, the labeling cost is reduced while the recognition accuracy is improved, and the double bottlenecks of rare species recognition difficulty and labeling cost are effectively broken through.

The invention provides a passive active field self-adaptive polyp semi-supervised segmentation and classification system and a passive active field self-adaptive polyp semi-supervised segmentation and classification method. The segmentation and classification system comprises a data collection module, a data preprocessing module, a source domain sample screening module, a target domain sample screening module, a data labeling module, a polyp segmentation and classification model construction module and a polyp segmentation and classification model semi-supervised training module. According to the method, a certain number of representative samples are screened out through a source domain sample screening module, and a certain number of representative samples, inter-domain similar samples with relatively high similarity with source domain features and inter-domain difference samples with relatively low similarity with the source domain features are screened out through a target domain sample screening module; and marking the screened samples, and subsequently training a polyp segmentation and classification model through a cross pseudo-supervision consistency strategy to obtain final model parameters. According to the method, the dependence of the polyp segmentation and classification model on the labeled data volume can be greatly improved, and the method has good cross-medical center robustness.

The invention discloses a remote sensing image enhancement processing method and device, a storage medium and electronic equipment. The method comprises the following steps: acquiring layout condition information of a remote sensing image corresponding to a remote sensing task type; performing feature extraction on the layout condition information by using a layout encoder to obtain a corresponding first layout condition feature vector; analyzing the first layout condition feature vector by using a layout condition generator to obtain a plurality of second layout condition feature vectors; and analyzing the first layout condition feature vector and the plurality of second layout condition feature vectors by using an image generator to obtain a first enhanced remote sensing image and a plurality of second enhanced remote sensing images corresponding to the remote sensing image. The technical problem that a remote sensing task model obtained based on remote sensing image training is high in cost and poor in performance due to the fact that enhancement processing cannot be efficiently carried out on the remote sensing image at low cost in related technologies is solved.

The present disclosure provides a data labeling, data generation method, device, equipment and storage medium, data fragmentation processing is carried out for the first type of data, N data fragments corresponding to the first type of data are obtained, and the fragmentation characteristics of each data fragment in the N data fragments are extracted; from the data feature library corresponding to the second type of data, a data feature set whose similarity with the fragmentation characteristics of each data fragment in the N data fragments satisfies a preset condition is obtained, N data feature sets corresponding to the N data fragments are obtained; the data features in each data feature set in the N data feature sets are respectively counted, the occurrence times of the data features in the N data feature sets are counted, and the first type of data is labeled based on the second type of data corresponding to one or more data features with the most occurrence times.

The application is a medical image segmentation method based on multi-modal self-supervision. First, the multi-modal medical image of the lesion tissue is obtained, including A-mode image and B-mode image, and the image is preprocessed. Then, a cycle-consistent modal contrast domain translation network is constructed, including two generators and two discriminators. The generator is used to convert the image of one mode into the image of another mode, including an encoder, an intermediate shared module and a decoder. The discriminator is used to judge the source of the input. Then, the cycle-consistent modal contrast domain translation network is pre-trained, the training loss is calculated, and the loss function includes multi-modal semantic consistency loss, adversarial loss, cross-domain translation loss and cycle consistency loss. Finally, the A-mode segmentation network and the B-mode segmentation network are constructed, the pre-trained weights are migrated to the two segmentation networks, and the trained two segmentation networks are respectively used for medical image segmentation of the corresponding mode. The contrast cross-domain translation is used as a multi-modal self-supervised pre-training task to learn more comprehensive modal features, promote the network to better learn modal characteristics and common knowledge, and improve the segmentation ability.

The present application relates to a kind of heart ultrasound video semantic segmentation pseudo-label generation method based on optical flow guide, belong to computer vision and medical image processing field.The method includes: selecting two key frames in heart ultrasound video frame sequence, and obtaining the segmentation mask of two key frames by artificial labeling;Optical flow model is fine-tuned using two key frames, the frame between two key frames, the frame of first key frame left side preset quantity and the frame of second key frame right side preset quantity;Based on the fine-tuned optical flow model, the forward optical flow sequence from first key frame to second key frame and the reverse optical flow sequence from second key frame to first key frame are predicted;Based on forward optical flow sequence and reverse optical flow sequence, generate forward propagation mask sequence and reverse propagation mask sequence, and carry out position weighted fusion, obtain the pseudo-label of unlabelled frame.The technical problem that the present application aims to solve is that the label of heart ultrasound video semantic segmentation is difficult to obtain and the quality of pseudo-label obtained is poor.