34results about How to "Reduce labeling costs" patented technology

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 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 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.

This invention discloses an online detection system for airbag coating defects based on machine vision and deep learning, belonging to the field of airbag coating detection technology. It includes: a multispectral imaging module containing visible and infrared light sources, forming a ring-shaped multispectral light source array for acquiring visible and infrared spectral images of the airbag coating surface; an FPGA-accelerated lightweight deep learning model, including adaptive convolution kernels and channel-parallel computing modules, achieving real-time inference at <50ms / frame; an adaptive illumination control module integrating an ambient light sensor and deep learning model output, adjusting light source parameters in real time through a closed-loop control algorithm; and a multi-angle imaging and image fusion module, including a ring camera array and a compressed sensing encoder, for acquiring multi-angle images of the airbag surface and generating a 3D defect distribution map. This invention achieves high-precision, high-speed, and highly robust online detection of minute defects in airbag coatings.

This application discloses a method for identifying and diagnosing abnormal states in fish by fusing multimodal data. The method is implemented through a multimodal data-fused fish abnormal state identification model, comprising a multimodal encoder, a conditional modulation meta-network, a shared classifier head, and a text decoder. The method includes: constructing a multimodal dataset; jointly training the multimodal encoder and text decoder using the multimodal dataset, and freezing the parameters of the multimodal encoder after training; constructing multiple meta-tasks based on the multimodal dataset; jointly training the conditional modulation meta-network and the shared classifier head, used to generate modulation parameters for the multimodal encoder, based on K samples from the current meta-task, to obtain the trained multimodal data-fused fish abnormal state identification model; and using this trained model to identify and diagnose abnormal states in the fish under monitoring. This method can identify unknown anomalies, reduce annotation costs, and improve the accuracy of fish abnormal state identification and diagnosis.

The application discloses an active domain adaptation semantic segmentation method, system, device and storage medium, selects and labels taking superpixels as units, which is different from image-level and pixel-level labeling methods, and the superpixel-level labeling greatly improves labeling efficiency by only assigning a semantic category to each superpixel; in addition, different from the existing scheme based on the uncertainty selection labeling strategy, the application focuses on difficult example samples in the domain adaptation scene, and proposes a selection strategy based on domain information quantity to label superpixels most valuable for domain adaptation learning; by adopting the superpixel-level labeling method and the selection strategy based on the domain information quantity, the application greatly reduces the labeling cost while improving the labeling quality, and guarantees the performance of the domain adaptation semantic segmentation.

This invention relates to the fields of unmanned aerial vehicles (UAVs) and computer vision technology, and particularly to a UAV-based method for classifying bridge components. The technical problem is that existing point cloud-based automatic classification methods for bridge components suffer from issues such as scarce samples, high annotation costs, insufficient model generalization ability, and low processing efficiency. The technical solution is a UAV-based method for classifying bridge components, comprising a data acquisition step, a data augmentation step, a data preprocessing step, a model training step, and a component classification step. In the model training step, preprocessed 3D point cloud data and its corresponding component category labels are used to train a deep learning-based point cloud semantic segmentation model, enabling the model to learn the feature representations of different bridge components. This invention uses a systematic data augmentation strategy to simulate real-world acquisition interference, combined with an efficient point cloud deep learning model, to achieve accurate, robust, and efficient classification of various bridge components.

This invention discloses a cross-modal data processing system for the safe operation of hydrogen refueling stations, including a raw data acquisition and processing module, a full-variable safety scanning module, a physical relationship coupling diagnosis module, an adaptive operating condition clustering module, a question-answer pair construction module, a hydrogen refueling station time-series command data acquisition module, and a fault type identification module. By constructing a three-layer semantic enhancement logic and model adaptation strategy, this invention effectively solves the technical problems in the prior art, such as the lack of supervision signals in the raw data of hydrogen refueling stations, the difficulty in identifying hidden faults under complex operating conditions, and the difficulty in adapting heterogeneous feature space models.

The application discloses a mass spectrum combined characterization qualitative and quantitative method and system of unified hidden state, relates to the cross field of analytical chemistry, bioinformatics and artificial intelligence data processing, the method obtains original data and instrument / method metadata containing LC-MS / MS, encodes to form a condition vector and is used for conditional modulation, constructs wave encoder and particle encoder, respectively Tokenizes continuous chromatographic signal and discrete fragment peak set into wave encoder and particle encoder to obtain wave representation and particle representation, realizes cross-modal interaction through wave-particle fusion module, forms unified hidden state or object-level unified hidden state set, adopts two-stage training of self-supervised pre-training and task fine-tuning, and applies conditional quantitative operator and qualitative operator to unified hidden state or object-level unified hidden state set in the inference stage, and parallelly outputs quantitative results and identification score / probability, and is suitable for end-to-end qualitative and quantitative analysis in DDA / DIA scene.

This invention provides a method and system for detecting internal defects in continuously cast billets based on a backdistillation network, comprising the following steps: inputting an image of the continuously cast billet to be tested into a backdistillation network to obtain a multi-scale feature set; constructing a pixel-level anomaly score map based on the obtained multi-scale feature set; and detecting internal defects in the image of the continuously cast billet to be tested based on the pixel-level anomaly score map. The backdistillation network includes a teacher branch network and a student branch network, with an enhanced cross-mapping feature fusion module between the teacher and student branch networks; the student branch network includes an asymmetric guided jump connection module. This invention not only effectively solves the problem of dependence on a large amount of labeled data in traditional defect detection methods but also improves the model's detection capability under conditions of few samples.

The present application relates to the technical field of computer vision and image processing, and discloses a remote sensing weakly supervised fine-grained target detection and recognition method and device, based on a general text prompt corresponding to a small amount of labeled remote sensing data samples and a large amount of unlabeled remote sensing data samples, by extracting the geometric features of the samples and the cross-modal features representing the visual text differences, combining the fine-grained class labels of the labeled remote sensing data samples, prior knowledge prototypes of various fine-grained classes are constructed; then, the prior knowledge prototypes of various fine-grained classes are used to construct fine-grained soft labels of the unlabeled remote sensing data samples, which are used as supervision signals to realize the training of the model, solve the cost consumption problem of fine-grained labeling, improve the label labeling efficiency of the unlabeled remote sensing data samples, and further improve the model training efficiency and target recognition efficiency.

The application discloses an infrared small target detection method based on space-time context perception and compact geometric representation, comprising the following steps: acquiring three adjacent images in a sequence of infrared images to be detected; inputting a pre-trained small target detection model to output a center heat map, a center offset and an effective radius prediction result, and completing positioning and scale estimation of the infrared small target; the small target detection model is used to extract multi-time space features through a backbone network and a feature pyramid sharing weights, obtain space-time representation features suitable for infrared small target detection by introducing space-time context perception information and constructing a time domain difference enhancement and global gating adjustment mechanism, and realize direct prediction of the center position and the effective radius of the small target in combination with a decoupled geometric parameter prediction network. The infrared small target detection task is modeled as target center position and effective radius prediction, so that effective detection is realized while reducing model calculation complexity and labeling cost.

The application discloses a kind of multi-level man-machine collaborative data labeling method and system based on task ambiguity evaluation, first constructs and trains ambiguity prediction model, utilizes lightweight ambiguity prediction model, predicts the cognitive ambiguity score and logic ambiguity score of each data instance;Four-stage labeling resource pool containing low-performance large language model, high-performance large language model, ordinary crowd sourcing labeler and expert labeler is constructed;According to the predicted two-dimensional ambiguity score, the data instance is automatically routed to the optimal labeling resource by dynamic scheduling strategy, for the task with high cognitive ambiguity, the upgrading mechanism combining crowd sourcing voting and expert arbitration is used, and the final labeled data set is output.The application can intelligently match labeling cost and ability according to the intrinsic properties of the task, significantly reduce the labeling cost and improve the labeling efficiency under the premise of ensuring the data labeling quality, solve the problem of high cost of traditional crowd sourcing and unstable quality of single model labeling.

The invention discloses a text prompt type heart function intelligent evaluation method based on a visual language large model. The method comprises the following steps that 1, an echocardiography mark data set is constructed; 2, dividing four chambers of the heart; step 3, hierarchical feature learning; 4, performing multi-scale feature fusion; and 5, predicting the cardiac function (ejection fraction). According to the method, the heart can be modeled into a multi-scale semantic system, and features from macroscopic chamber dynamics to microstructure motion are captured through a hierarchical feature learning mechanism. Deep integration of imaging data and clinical knowledge is realized by aligning hierarchical visual features with professional medical descriptions in a shared semantic space. The method has excellent performance in the aspect of ejection fraction prediction.

This invention discloses a big data-based operational analysis method, system, and medium, relating to the field of big data analysis technology. The method involves a central server first performing hierarchical or channel pruning on an initial model based on client terminal computing resource information, generating multiple compression ratio versions and adapting and distributing them. Then, the client terminal uses local historical behavioral time-series data to train the received model locally, allocating differentiated noise injection amounts to model parameters based on the impact of behavioral characteristics. After noise addition and homomorphic encryption, the parameters are uploaded. Next, the central server securely decrypts and aggregates the encrypted parameters, updates the model, and recompresses and adapts it before distribution. Then, the client terminal uses the new model to predict the churn probability of the latest behavioral time-series data, uploading early warning information when thresholds are exceeded. Finally, the central server generates an operational analysis report based on the early warning information. This achieves a privacy-preserving federated learning and real-time prediction closed loop, balancing data privacy and prediction accuracy.

The application discloses a kind of hair follicle position and growth direction combined detection method and system of hair, belong to hair detection technical field, comprising: obtaining a plurality of hair mirror images and all hair follicle points are single point marked to construct data set;Data set is divided into training set and test set, and the image in training set is carried out random data enhancement;The training set after random data enhancement is used to train hair follicle point prediction model and test set is used to test, obtain the coordinate of all hair follicle points in the hair mirror image to be measured that is predicted by trained hair follicle point prediction model;According to the coordinate of each hair follicle point in the hair mirror image to be measured, based on local skeleton constraint is iteratively tracked to obtain the final hair growth direction of corresponding hair follicle point;The coordinate of each hair follicle point in the hair mirror image to be measured and final hair growth direction are used as the combined detection result of the hair mirror image to be measured. Training data labeling cost can be reduced, avoid growth direction semantic ambiguity, and detection accuracy is high.

This invention discloses a hierarchical multi-label attribution method and system that integrates atomic rule-driven credible features and knowledge distillation, belonging to the field of natural language processing technology. First, this invention constructs an atomic rule base for weakly supervised text annotation. Then, it uses a large language model as a teacher model to correct and supplement the weak annotation results, extracting the probability distribution of soft labels and intermediate layer feature representations on each level of labels. Next, it evaluates the credibility of the teacher model's output, selecting a subset of credible soft labels and credible feature dimensions. Then, it constructs a student model with a hierarchical output structure, designs a joint loss function, and distills the student model for training. Finally, it deploys only the student model for inference, outputting hierarchical multi-label attribution results and key evidence fragments. This invention, through the combination of atomic rules and credible knowledge distillation, significantly reduces inference costs while improving the accuracy, stability, and interpretability of hierarchical multi-label attribution.

This disclosure relates to a model training method, apparatus, vehicle, storage medium, and computer program product, belonging to the field of image processing technology. The method includes iteratively executing the following steps to obtain a target network model: inputting a collected image of a storage location into an initial network model to predict the predicted three-dimensional coordinates of a target point in the storage location; converting the predicted three-dimensional coordinates into predicted two-dimensional coordinates; updating the network parameters in the initial network model based on the loss value between the predicted two-dimensional coordinates and the actual two-dimensional coordinates; the actual two-dimensional coordinates are obtained by labeling the target point in the storage location image; if the loss value meets the convergence condition, the initial network model that meets the convergence condition is used as the target network model; the target network model is used to obtain the three-dimensional coordinates of the target point from the storage location image. Using the model training method proposed in this disclosure, the cost of manual labeling can be reduced while obtaining highly accurate predicted three-dimensional coordinates.

This invention discloses an automatic detection and classification method and system for micro-defects in OLED displays, relating to the field of display panel quality inspection technology. The method includes acquiring OLED display images and preprocessing them; modeling pixel responses based on the preprocessed images and generating an ideal background image; performing differential processing and multi-scale residual feature enhancement to obtain an enhanced residual image; segmenting the enhanced residual image to extract a set of candidate defect regions and extracting global features; based on the set of candidate defect regions, performing frequency domain enhancement on the global features using a local frequency domain prior-guided complex filter to obtain enhanced global context features; based on the enhanced global context features, using a meta-learning mechanism to obtain new screen type adaptation parameters for defect category probability prediction; and performing uncertainty estimation based on the defect category probability prediction results. This invention enables more accurate, flexible, and reliable OLED display defect detection and classification.

The application provides a product detection method, device and equipment and a storage medium. The product detection method comprises: inputting a product image of a target product into a pre-trained image classification model, performing classification prediction on a detection result category to which the product image belongs by the image classification model, and outputting a coarse-grained classification prediction result for the target product; obtaining an intermediate feature map generated by the image classification model in an intermediate stage of classifying and predicting the product image; calculating a fine-grained classification prediction result of the target product based on the intermediate feature map and a standard image corresponding to each detection result category; and jointly determining a final product detection result of the target product based on the coarse-grained classification prediction result and the fine-grained classification prediction result. In this way, the application can realize fine-grained detection of the target product without introducing an image segmentation model, thereby effectively reducing the labeling cost of image labeling in the early model training stage.

This invention relates to the field of spatial transcriptomics technology and provides a few-sample spatial transcriptomics cell annotation system and method based on graph cue learning, comprising: a preprocessing module; a graph construction module, which constructs spatial adjacency relationships based on cell spatial coordinates and constrains or weights graph connectivity relationships by combining inter-cell transcriptomics expression similarity; a graph representation learning module, which learns discriminative cell embedding representations from the cell graph structure under unsupervised or weakly supervised conditions; a cue fine-tuning module, which injects the category information contained in a small number of labeled samples into a pre-trained graph representation model in the downstream cell type annotation task; and a cell type prediction module, which completes cell type annotation based on a small number of labeled samples. This invention can fully integrate gene expression information and multi-source spatial relationships under conditions of very few labeled samples to construct highly discriminative cell representations, achieving high-precision automatic annotation of large-scale spatial transcriptomics data.

The application discloses a point supervision semantic segmentation model construction and segmentation method based on SAM and CNN feature fusion, comprising the following steps: step one, acquiring remote sensing image data as a training set, and performing point labeling on target ground objects in the remote sensing image data to obtain a point label set, wherein the point label set is the category and position of the target ground objects; a point supervision semantic segmentation model is constructed; object structure information generated by using a convolutional neural network branch and a SAM driven fusion branch is used to realize collaborative modeling of image semantic information and structure information in combination with a cross-level fusion module CFM; meanwhile, under the condition of point supervision, the model can enhance the integrity control of the target region, improve the boundary positioning accuracy, and reduce the category confusion phenomenon, thereby significantly improving the effect of remote sensing image semantic segmentation. The technical problem that the existing point supervision semantic segmentation method is difficult to fully utilize image structure information and semantic information under the condition of sparse supervision information is solved.

The application discloses a kind of generation methods, devices and equipment of vehicle expression dataset and storage medium, by obtaining the scene parameter library containing multiple vehicle environment parameters and the anonymization standard expression library without identity as basic material, expression data is decomposed into multiple action unit combinations using action unit analysis module, action features and parameter features are extracted and fused, initial expression data is obtained by inputting generative adversarial network generator, and then privacy enhancement and quality check form standardized dataset.The application avoids the privacy disclosure risk collected by real drivers, reduces the collection and labeling cost, by fusing environmental parameters and action unit features, the dataset covers multiple vehicle conditions, improves scene adaptability, and privacy enhancement and quality check guarantee data security and quality, which can be directly used for vehicle expression recognition algorithm training, meet the demand of multi-scene, high-quality, compliance data.The technical scheme of the application can be widely applied to the field of vehicle technology.

This application discloses a unified latent state-based mass spectrometry joint characterization qualitative and quantitative method and system, involving the interdisciplinary fields of analytical chemistry, bioinformatics, and artificial intelligence data processing. The method acquires raw data from LC-MS / MS and instrument / method metadata, encodes it into conditional vectors, and uses them for conditional modulation. A wave encoder and a particle encoder are constructed, and continuous chromatographic signals and discrete fragment peak sets are tokenized and input into the wave encoder and particle encoder respectively to obtain wave and particle characterizations. Cross-modal interaction is achieved through a wave-particle fusion module to form a unified latent state or an object-level unified latent state set. A two-stage training process of self-supervised pre-training and task fine-tuning is employed. During the inference stage, conditional quantitative and qualitative operators are applied to the unified latent state or object-level unified latent state set, outputting quantitative results and identification scores / probabilities in parallel. This method is suitable for end-to-end qualitative and quantitative analysis in DDA / DIA scenarios.

This application relates to a data processing method and apparatus for incomplete multi-source information. The method includes: performing quality assessment on multi-source data channels to generate topological fingerprints characterizing the availability status of each data channel; dynamically routing data in a pre-constructed semantic topology graph to determine the target node corresponding to the current input data and the statistical parameter library of the target node; calculating the class-conditional inconsistency score of the current sample in the feature space; calibrating the information entropy of the inconsistency score distribution of the sample; adaptively weighting the sample for risk; and determining a dynamic safety threshold based on the weighted quantiles; comparing the inconsistency score of the current sample in each candidate category with the dynamic safety threshold to generate a prediction set containing categories that meet the threshold conditions as output. This method can ensure the stability and maintenance efficiency of the system during multi-source information fusion decision-making without the need for manual annotation.

This invention discloses a fake image detection method based on artifact decomposition. The method includes acquiring the image to be detected and inputting it into a trained three-branch artifact-aware encoder (scene consistency branch, imaging realism branch, and signal naturalness branch) for feature extraction, resulting in three artifact reflection maps. A trained cross-dimensional gated collaborative fusion module is used to fuse the features of the three artifact reflection maps, generating a unified artifact embedding representation. This unified artifact embedding representation is then input into a trained classification head, and the label classification head of the classification head outputs the true / false prediction probability of the image to be detected. This fake image detection method solves the problems of existing technologies, such as inability to uniformly detect multi-source heterogeneous fake images, poor generalization ability, and susceptibility to overfitting.

This invention discloses a method and system for metallographic structure identification of aluminum alloy piston materials based on neural networks, comprising: acquiring a preprocessed image, performing pixel-level segmentation, and generating a set of candidate defect regions; extracting morphological feature vectors, boundary feature vectors, and context feature vectors for each candidate region; fusing the morphological feature vectors, boundary feature vectors, and context feature vectors to construct a comprehensive discrimination feature vector, and classifying and discriminating each candidate region to output the identification results of pores or inclusions; mapping the classification results back to the corresponding region in the pixel-level segmentation probability map to generate a spatial distribution map of pores and inclusions, and performing structural statistical analysis based on the distribution map to output porosity and inclusion distribution information; this invention solves the problems of inaccurate detection of small or complex-shaped defects, imprecise statistical analysis of the spatial distribution of pores and inclusions, and high cost of high-precision annotation.

The application relates to a weakly supervised target detection method based on an adversarial collaborative learning, which comprises the following steps: acquiring image data to be detected, inputting the image data into a pre-trained adversarial collaborative network, and outputting a target detection result, wherein the adversarial collaborative network comprises two peer network models with the same structure, each peer network model is a standard WSOD model, the WSOD model comprises a candidate region feature extractor and a task head, and the task head comprises a multi-instance detection module, an online instance classification module and a detection head. Compared with the prior art, the application has the advantages of reducing the dependence on a large amount of accurate labeled data, improving the perception ability of complete objects and the like.

A method for extracting triples from knowledge graphs in the mining industry based on ontology constraints and thought chain principles is proposed. The method constructs a highly complete mining domain ontology, acquires and preprocesses unstructured text data sources from the mining domain, and then constructs structured prompts containing ontology constraint information and thought chain reasoning instructions, transforming the triple extraction task into a constrained generative reasoning process. The structured prompts are input into a generative large language model. After receiving the input, the model, relying on its self-attention mechanism and pre-trained knowledge, strictly follows the pre-defined thought chain paths in the structured prompts to conduct reasoning. The output of the generative large language model is parsed using a parsing algorithm, separating the thought chain part and the final result part of the model output. Triple data structures are extracted from the final result part to form a candidate triple list, followed by ontology-based post-processing verification and optimization. This method enables efficient, accurate, and standardized extraction of knowledge in the mining domain.