17 results about "Class prediction" patented technology

This invention relates to the field of cardiovascular disease auxiliary diagnosis technology, specifically to a method and system for auxiliary diagnosis of high-altitude hypertension based on electrocardiogram (ECG). It uses 12-lead ECG images as input, constructs a multi-branch adaptive classification model that integrates regional priors, sequentially extracts shallow general features through a shared feature extraction module, combines independent feature branches from multiple leads with a paired spatial attention module to capture relative potential differences between leads, and a multi-path adaptive fusion module dynamically fuses differential features between high-altitude and plain areas using regional labels as gating signals. Finally, a hybrid classifier composed of capsule networks and DenseNet cascades the features to complete a four-class prediction: high-altitude hypertension, high-altitude normal, plain hypertension, and plain normal. This invention effectively overcomes the problem of cross-regional data domain shift, improves the ability to capture weak pathological features in ECG images, and has strong clinical transferability and robustness, making it suitable for auxiliary screening of hypertension across high-altitude and plain areas.

The application discloses a photovoltaic panel damage detection method and system based on an open vocabulary, and a medium, solves the problems of high model training cost and long training time in the prior art, has the beneficial effects of reducing training cost and improving recognition accuracy, and the specific scheme is as follows: a photovoltaic panel damage detection method based on an open vocabulary, comprising constructing a training data set; performing knowledge distillation pre-training, using the pre-trained open vocabulary target detection model as the teacher model, using the detection model to be trained as the student model, distilling the intermediate layer features and class prediction distribution of the teacher model respectively, and the student model learns basic semantic understanding ability through the first pre-training data set; freezing the knowledge distillation module of the student model, collecting images of photovoltaic panels to be detected and preprocessing, and inputting the fused multi-scale feature map and the updated text embedding vector into the detection output module.

This invention discloses an adaptive model reuse method based on adapter fine-tuning. Taking object detection in computer vision as an example, firstly, a model pre-trained on a large-scale image dataset is obtained as the original model, and candidate adapters are set for it according to the structure of the pre-trained model. Then, the Taylor expansion method is used to calculate the global gain of adding adapters at different positions in the network for images with localized bounding boxes and class labels in the target domain, thereby achieving adaptive adapter configuration. In the fine-tuning stage, the global gain of adapters at different positions for labeled image samples is recalculated. The original model parameters are frozen, and the adapter parameters are updated using gradient backpropagation with a learning rate weighted by the global gain softmax. Finally, the fine-tuned model is reused in downstream tasks to locate objects in images and provide class predictions. This invention achieves adaptive adapter configuration based on global gain, realizing more accurate object localization and class prediction in object detection tasks.

This invention relates to the fields of network security and radio transmission signal processing technology, specifically to a method, system, device, and storage medium for identifying individual radio emission sources in open set communication scenarios. The method includes acquiring a discrete baseband complex signal transmitted by a radio signal source; obtaining backbone intermediate features from the discrete baseband complex signal; obtaining low-frequency structural components and high-frequency detail components from the backbone intermediate features; obtaining low-frequency enhancement features from the low-frequency structural components; obtaining high-frequency enhancement features from the high-frequency detail components; obtaining fusion features from the backbone intermediate features, low-frequency enhancement features, and high-frequency enhancement features; generating and outputting discriminant embedding and known class prediction scores based on the fusion features; calculating unknown scores based on the discriminant embedding; and combining the unknown scores and known class prediction scores to identify the category of the radio signal source. This solution can improve the known recognition capability and unknown rejection robustness of radio signal sources in open set scenarios.

This application discloses a training-free differential-guided weighted interpolation method, medium, and device for class prediction, relating to the field of machine learning. The method includes generating branch adaptive inputs for a pre-trained model and a fine-tuned model based on the sample to be predicted; calculating prediction information for the pre-trained and fine-tuned models based on the two branch adaptive inputs, including initial interpolation confidence and cross-layer representation consistency; calculating sample-level interpolation coefficients based on the cross-layer representation consistency and initial interpolation confidence of the pre-trained and fine-tuned models; performing weighted interpolation on the parameters of the pre-trained and fine-tuned models based on the sample-level interpolation coefficients to generate temporary interpolation parameters; introducing the temporary interpolation parameters into the target network architecture to generate a temporary interpolation model; and using the temporary interpolation model to predict the class of the sample to be predicted. This method solves the problem that sample-level interpolation in current class prediction methods can compromise out-of-distribution robustness.

The application provides a drive system gear fault diagnosis method based on incremental learning, which comprises the following steps: first, using a sensor to collect drive system monitoring data and constructing an incremental data set, which is divided into fault diagnosis tasks of different stages; constructing an initial model and training the model based on initial fault data; defining a generalized entropy index to determine whether the fault is a new fault in the test stage; when new fault category data appears, a small amount of samples are randomly selected from old class data and fused with new class data to construct a training set; in order to avoid the forgetting of the model to the old class fault data, knowledge distillation is performed on the attention weight and the old class prediction output during training; when the generalized entropy index continues to judge that an unknown fault appears, the network training step is repeated to update the model, and finally the incremental learning of the fault mode is realized. The application can realize the continuous updating of the fault diagnosis model for the streaming data, improve the state monitoring capability of the drive system, and provide protection for the service safety of the mechanical system.

This invention proposes a multi-class unified 3D anomaly detection method based on hash-coded neural fields, belonging to the fields of computer vision and industrial quality inspection technology. It solves the problem that existing 3D anomaly detection methods struggle to simultaneously handle multi-class unified modeling, pose change adaptation, and anomaly repair. The method includes: 1. Acquiring multi-class normal point cloud samples and performing pose normalization on the samples; 2. Extracting multi-scale geometric features from the normalized point clouds and training a class prediction model; 3. Constructing a class-conditionalized multi-resolution hash-coded neural field and training a signed distance field model using normal samples; 4. Performing pose normalization, class prediction, and prototype registration on the point cloud to be tested; 5. Querying the trained neural field to complete point-by-point anomaly localization and target-level anomaly determination; 6. Reconstructing the surface of the anomaly region based on the zero isosurface of the signed distance field to obtain the repaired point cloud result. This invention can achieve multi-class point cloud anomaly detection, localization, and repair under a single model, and has advantages such as good pose robustness, strong fine-grained geometric expression capability, and high detection efficiency.

The application discloses an image segmentation method and system based on local-global fusion and channel modulation, and relates to the technical field of image segmentation. The specific steps are as follows: a local-global fusion attention module and a depth estimation module are combined to encode an input remote sensing image, and a multi-scale fusion feature map is output; the multi-scale fusion feature map is subjected to feature compression and segmentation, cross fusion, channel modulation and adaptive weighting, and a modulation feature map is output; the spatial resolution of the modulation feature map is restored through step-by-step upsampling and convolution, and a pixel-level class prediction map is generated through segmentation. The application optimizes the spatial distribution consistency of features, helps to improve the class consistency of the overall segmentation result, reduces the class confusion phenomenon, and is especially suitable for complex remote sensing image segmentation tasks containing multiple classes and multiple scales.

The application discloses a multi-scale adaptive gating MambaPlus network construction method and device, belongs to the technical field of artificial intelligence and machine learning, and comprises the following steps: initializing network configuration and constructing a basic structure; pre-processing input data to generate a standard data set; mapping to a hidden space through an input mapping layer, extracting backbone features from a Mamba main body; constructing at least two parallel scale branches in the hidden space to obtain a multi-scale feature set; inputting the backbone features and the multi-scale features into an adaptive gating module, dynamically distributing weights and adaptively fusing through a hierarchical gating mechanism to generate fused features; then, cross-scale attention enhancement and feedforward enhancement are performed, and residual superposition is performed to generate final discriminative features; and finally, class prediction and model training evaluation are completed. While retaining the advantages of Mamba long-range dependence modeling, the application solves the problems of insufficient multi-scale information utilization, poor feature fusion adaptability and low robustness in complex scenes.

The application provides a kind of power transmission line bird identification method based on deep learning, it is related to the technical field of bird damage detection of power transmission line.The application converts monitoring video frame into pixel color value array and inputs target detection network, extracts multi-scale feature map using backbone feature, and realizes attention enhancement and multi-scale feature fusion through improved feature fusion network containing scale sequence attention module SSA and down-sampling feature fusion module ADFF, generates fusion feature map corresponding to P2 to P5 scale.Again, by perception detection head module, boundary box regression and class prediction are executed, and bird class and position results are output.Training phase is based on bird image with real label to construct training sample set, and PSIoU-based PSLoss loss function is used to iteratively update network parameters.The application can realize high-precision and strong-robustness identification of multi-scale bird targets in complex line scene.

The present application relates to a kind of visual statistics methods of material in and out of warehouse, comprising: target material image is collected, the target material image is input into RTMDet detection model, and target material detection result is obtained;RTMDet detection model utilizes training set to obtain by multi-task learning optimization improved RTMDet network model's class prediction and scale regression;Training set includes: original target material detection result and corresponding volume characteristics;Improved RTMDet network model is obtained by improving the basic building unit of RTMDet network model;Target material detection result is fused with fixed stack area boundary information in time sequence, and final detection result is obtained;Fixed stack area boundary information is calculated using the target point cloud data extracted by depth map to obtain.

The application relates to the technical field of image processing, and particularly discloses a coal mine conveying belt coal gangue recognition system and method based on an improved YOLOv11, which comprises an image acquisition unit for acquiring real-time images of a coal mine conveying belt; an edge computing processing unit loads and runs an improved YOLOv11 algorithm model, and performs real-time coal gangue detection on the conveying belt images acquired by the image acquisition unit; the improved YOLOv11 algorithm model comprises a backbone network, a neck network and a detection head connected in sequence; the backbone network comprises at least one adaptive down-sampling module for down-sampling input feature maps; the neck network is used for fusing multi-scale features output by the backbone network to generate a fused feature map; the detection head is a P2-P5 lightweight detection head, which is used for receiving four layers of feature maps output by the neck network respectively, and after the feature maps are fused through an adaptive fusion layer, the detection head is input into a bounding box regression branch, a class prediction branch and a confidence prediction branch respectively for decoupled output.

The application discloses a power distribution line ground wire broken strand defect identification method, device, equipment and medium, and relates to the technical field of target detection. The method comprises the following steps: acquiring a power distribution line image collected by an unmanned aerial vehicle, and extracting a multi-scale feature map from the power distribution line image; performing two-class prediction on the multi-scale feature map to obtain a feature map two-class result; performing self-adaptive enhancement according to the two-class result to obtain a reset feature map; performing down-sampling on the reset feature map, and performing feature fusion according to the down-sampled reset feature map and the multi-scale feature map to obtain a target fusion feature map; and performing identification on the target fusion feature map to obtain a target detection and identification result of a power distribution line ground wire broken strand. In this way, the two-class identification, self-adaptive feature enhancement and dynamic fusion mechanism are utilized, the ground wire broken strand defect identification effect is effectively improved, the false positives are greatly reduced, and the overall identification effect of the model network is improved.

The application discloses a cross-patient seizure detection method based on a graph-enhanced pulse state space model, first, the brain electrical signal data of a patient is acquired and an electrode space topology graph is constructed; then, a pulse state space model is used to extract the time sequence features of physiological channels, and a pulse graph convolution network is used to perform message aggregation between adjacent nodes in a discrete pulse domain across time steps; finally, feature aggregation is performed through a global average pooling operation at a graph level, and a two-class prediction probability is output through a fully connected classifier. The application follows the design logic of "first time and then space", the pulse state space model models the potential brain state dynamics of the time sequence change, and realizes stable long context feature processing through the cyclic updating mode of the adaptive hardware; the pulse graph neural network completes the information transmission and fusion between channels based on the electrode correlation graph, can capture the spatial dependence relationship which is crucial for the representation of seizures, reduces the calculation energy consumption, and improves the overall performance.

The application provides a text classification method based on residual semantic enhancement and a multi-scale convolutional network, and the method comprises the following steps: S1, obtaining a feature after semantic enhancement; S2, generating a local enhanced feature and obtaining a wide-area context feature; S3, splicing and fusing the local enhanced feature and the wide-area context feature to obtain a spliced feature vector, and calculating a class probability distribution of the text by using a Softmax function to perform class prediction. The application can enrich the early semantics of the bottom layer word vector representation without destroying the original word sequence, and provide an extremely robust high-dimensional semantic base for subsequent multi-scale mapping; the method can capture rich local feature combinations by using multi-size convolution kernels, and can multiply the receptive field of the model without increasing the parameter amount, so that the network has a powerful long-span chapter-level context dependence capturing capability.