247 results about "Image prediction" patented technology

The invention relates to a textile product defect identification method based on improved YOLOv11. The method comprises the following steps: acquiring a textile product defect image data set; performing pretreatment; dividing into a training set and a verification set; the method comprises the following steps: introducing MConv into a YOLOv11 backbone network, adding a CCIAP module behind a C2PSA module, and applying BiFPN in a path aggregation network; performing prediction through YOLO Head to obtain N prediction feature maps; the overall loss of the network is calculated, and network parameters are optimized through back propagation; predicting the verification set image through a network to output AP values of various categories; repeating the above steps to obtain a trained YOLOv11 network; and detecting the test image or video by using the trained detector to obtain a detection result. According to the method, the MConv is introduced into the YOLOv11 network to enlarge the receptive field, the CCIAP module is added behind the C2PSA to improve the feature extraction capability, and the BiFPN is applied to the Neck layer to enhance the feature fusion capability, so that the target detection precision is improved and the real-time detection of textile product flaws is realized under the condition that the reasoning speed is not influenced.

The invention discloses a multispectral satellite cloud picture prediction method based on motion stripe decoupling. The method comprises the following steps: carrying out normalization preprocessing on multi-channel satellite observation data; utilizing a motion branch model to extract motion features based on a displacement field, iteratively updating a prediction frame in an autoregressive distortion-correction pipeline, and keeping physical consistency in combination with atmospheric dynamics and smoothness constraint; a texture branch model is utilized to sequentially pass through a high-fidelity encoder, long memory state space modeling and a high-fidelity decoder, time sequence texture features are extracted, and cloud picture details are kept; the motion features output by the motion branches and the texture features output by the texture branches are input into a gating fusion module, adaptive weighting of the features is achieved through convolution and a gating mechanism, and fusion features are output; and carrying out reverse normalization processing on the fusion features to obtain satellite cloud picture prediction results at a plurality of moments in the future. According to the method, the spatial texture fidelity of cloud picture prediction can be improved while the physical interpretability is ensured, and high-precision satellite cloud picture prediction is realized.

A method performed by an electronic apparatus, includes: obtaining position information of an object in an input image; predicting trailing blur state information of the input image, based on the position information and the input image; and obtaining an output image by performing processing on the input image, based on the predicted trailing blur state information.

The invention discloses a wavefront detection-free adaptive optical correction method and system, and the method comprises the steps: obtaining a real in-focus image and a real out-of-focus image of a to-be-corrected light beam, and outputting an updated deformable mirror control voltage through a physical information neural network; generating a predicted wavefront phase according to the deformable mirror control voltage signal and a deformable mirror voltage-phase mapping model; generating a predicted in-focus image and a predicted out-of-focus image through a derivable Fourier optical propagation model; constructing a label-free composite loss function according to the real in-focus image, the real out-of-focus image, the predicted in-focus image and the predicted out-of-focus image; performing unsupervised training updating on the physical information neural network according to the composite loss function, and performing repeated iteration to obtain an optimized deformable mirror control voltage; and driving the deformable mirror to carry out self-adaptive optical correction in real time according to the optimized deformable mirror control voltage to obtain a correction result. According to the invention, the correction efficiency and the system stability are obviously improved.

The invention relates to a method and a device for generating a report under the assistance of a label, and the method comprises the following steps: 1) extracting a structured label set from a text report of a sample based on a large language model, wherein the structured label set comprises a multi-classification group consisting of dichotomous labels and mutual exclusion options; 2) aggregating the labels in batches, after a threshold value is reached, merging and de-weighting, performing specification and mutual exclusion group merging on synonymous, near-synonymous and redundant labels, and converging into a unified label library; 3) based on the text report and the tag library, outputting a tag subset of each sample through a large language model; 4) multi-modal multi-label classification model training: extracting each visual modal feature, performing weighted aggregation and splicing, and outputting each label group logits through a classification head to perform weighted group loss optimization; (5) carrying out joint training on the multi-modal large language model by using samples of'only images-reports' and'images + labels-reports', and (6) carrying out label prediction and screening on the images by using the classification model, and inputting'images + prediction labels' into the multi-modal large language model to obtain a final report.

The system generates a three-dimensional model with layered depth images based on an input two-dimensional image. For training, layered depth images are derived from existing three-dimensional models. The system trains a machine learning model to predict multiple layered depth images from an input image of an object. The system compares the generated, multiple layered depth images to the derived layered depth images for the object to update the machine learning model during training. At inference time, the system receives an input image for an object. The system applies the machine learning model to the input image to output predicted layered depth images. The system generates a three-dimensional model from the predicted layered depth images.

This invention belongs to the field of medical image analysis technology and relates to a brain tumor segmentation method based on a boundary-aware mechanism. The method involves inputting T1, T1c, T2, and FLAIR images of the brain tumor into a trained image segmentation model, which outputs a predicted segmented image. The predicted segmented image is a brain tumor MRI image with a complete tumor region, a tumor core region, and an enhanced tumor region, obtained through prediction. The proposed brain tumor segmentation method based on a boundary-aware mechanism incorporates boundary information into the image segmentation model, improving the model's ability to distinguish features and achieving accurate segmentation of tumor subregions. A multimodal fusion method is used to integrate complementary information from different MRI sequences, providing a comprehensive understanding of tumor characteristics. Furthermore, uncertainty quantification and an uncertainty-based loss function are combined to provide confidence measurements for the segmentation results, enhancing the accuracy and reliability of the segmentation and assisting clinicians in evaluating the prediction results.

The invention discloses an intensive agricultural crop detection method and system based on deep learning. The method comprises the following steps: extracting different branch features of a dense agricultural image through a MobileNet pre-training model; obtaining a dense agricultural image rough prediction map, a dense agricultural image detail prediction map and a dense agricultural image prediction map model; and by sampling the rough prediction map, the detail prediction map and the prediction map to the size of an original image, constructing a loss function calculation error, and using the obtained error to reversely update the dense agricultural segmentation model. And inputting the dense agricultural image into the updated model, and obtaining and outputting a corresponding dense agricultural image segmentation prediction map. According to the method, the capturing capability of hidden crops in dense agriculture is enhanced, and the light weight of the model is kept while the segmentation precision is remarkably improved.

The application provides a kind of "based on constraint CaiT model's primary curing tobacco dry degree quantification method", including the following steps: S1: making primary curing tobacco image dry degree calibration sample;S2: constraint CaiT-transformer model construction and training;S3: primary curing tobacco image preprocessing;S4: obtain primary curing tobacco image predicted value Y″;S5: primary curing tobacco image predicted value Y″ Real-time feedback to baking control terminal, temperature stage regulation and control.The constraint CaiT-transformer model of the application is based on CaiT-transformer model, adopts assignment weight absolute erro weight to constrain the model, multiplies assignment weight and cross entropy as objective function, trains the model in a back propagation manner, ensures the continuity of primary curing tobacco image predicted value, improves the stability and accuracy of model prediction result.

The present application relates to a kind of image super-resolution enhancement algorithm and system, comprising: collection and processing high-definition image;Obtain training sample set;Build depth learning network, the depth learning network includes texture feature network NetV and color feature network NetC, first scale image SrcscaleIMG input texture feature network NetV, output first predicted image, with the difference between first predicted image and gray processing image GrayIMG minimum as goal iterative training, end iteration output optimal texture feature image prediction model, image result after being processed by optimal texture feature image prediction model and scale image SrcscaleIMG new superimposed image is generated by the way of pixel point superposition;Second, superimposed image is input to color feature network NetC, output second predicted image, with the difference between second predicted image and high-definition image SrcIMG minimum as goal iterative training, end iteration output optimal image super-resolution enhancement model.

The embodiment of the invention provides an image prediction method, an encoder, a decoder and a storage medium, if a current block uses an MIP mode to determine an intra-frame prediction value of the current block, the encoder sets the value of an MIP mode parameter to indicate to use the MIP mode and writes the value into a code stream; determining an MIP mode of the current block, and determining predicted values of a luminance component and a chrominance component corresponding to the current block according to the MIP mode; and writing the MIP mode of the current block into the code stream. The decoder analyzes the code stream and determines an MIP mode parameter of the current block; if the value of the MIP mode parameter indicates that the current block uses an MIP mode to determine an intra-frame predicted value of the current block, analyzing the code stream, determining the MIP mode of the current block, and determining predicted values of a luminance component and a chrominance component corresponding to the current block according to the MIP mode; and decoding the current block according to the predicted value.

This application relates to a method for training an image semantic segmentation model. Based on the ground truth labels of all pixels in the image to be segmented, edge detection is performed on the image to be segmented to obtain object edge images for each category. Image prediction results are obtained, and the gradient of each pixel in the prediction results is calculated to generate a pixel gradient map. The arctangent gradient map of each pixel is then obtained from the pixel gradient map. Based on the object edge images and the arctangent gradient maps, an auxiliary loss value is obtained for each pixel in the image to be segmented, predicted by the image semantic segmentation model to be trained. The overall loss value for each pixel in each category of the image to be segmented is calculated based on the auxiliary loss value and the loss value obtained from the main loss function. Based on the overall loss value, the parameters of the image semantic segmentation model are adjusted to obtain the trained image semantic segmentation model. This application can improve the training effect of the image semantic segmentation model and increase the accuracy of semantic segmentation.

The application relates to autonomous laser weed eradication, and provides a system for damaging or killing plants, the system comprising: a first camera, a second camera, a light source, a control system, and a computing system, the computing system is configured to: receive, at a first time, a first image of at least one plant in the field captured by the first camera; identifying a plant in the first image; predicting the position of the plant based on the first image; causing the second camera to capture a second image of an area of the field including the predicted position; predicting a target position of the plant in the second image at a second time after the first time, wherein movement of the second camera relative to the surface during the elapsed time is taken into account; causing the control system to direct an optical path of the light beam toward the predicted target position; the light source is caused to emit a light beam toward the predicted target location of the plant for a length of time sufficient to damage or kill the plant.

A detection method and system for a leaked gas based on multi-data fusion, and an electronic device, relating to the technical field of gas detection. The method comprises: acquiring a plurality of compressed images obtained by respectively compressing a plurality of first infrared images; on the basis of the plurality of compressed images, predicting a position of a potential leaked gas as a candidate position; extracting, respectively from a plurality of second infrared images, sub-images at the candidate position as infrared sub-images; and on the basis of all of the infrared sub-images, predicting a first classification result for the candidate position, wherein the first classification result is used for indicating the existence of a leaked gas at the candidate position or the absence of a leaked gas at the candidate position. By means of embodiments of the present application, rapid and accurate detection of a leaked gas can be achieved.

A method, device, storage medium and electronic equipment for image recognition are disclosed. The method comprises: determining a face image, human body region information and key point information of a human body skeleton of a to-be-recognized object from a to-be-recognized image; using a first pre-trained prediction model to predict the age of the to-be-recognized object based on the face image to obtain a predicted age of the to-be-recognized object; using a second pre-trained prediction model to predict an age interval in which the to-be-recognized object is located based on the human body region information and the key point information of the human body skeleton to obtain a predicted age interval of the to-be-recognized object; and determining whether the to-be-recognized object is a child based on the predicted age and the predicted age interval. The method overcomes the limitation of relying on single feature information for recognition, reduces the adverse effects of face or body occlusion on image recognition, and helps to improve the generalization, fault tolerance and accuracy of child recognition of image recognition.

The invention discloses an offshore wind power basic equipment stress monitoring and early warning system based on deep learning. The system comprises a data acquisition module, a mapping module, a stripe image prediction module, a fusion module, a model construction module, a stress module and an early warning module. According to the invention, the obtained multi-dimensional monitoring data is converted into a first stripe image which is consistent with a photoelastic stripe image in time and space through a mapping module and a stripe image prediction module, and then after the first stripe image and the photoelastic stripe image are fused, the stress distribution condition of the offshore wind power foundation equipment is predicted by using a stress module. And identifying abnormal characteristics in the stress distribution diagram, and in combination with a safety threshold value, judging potential risks in advance and sending out early warning. According to the method, the first stripe image and the photoelastic stripe image in the same time and space are fused to complement information loss of the first stripe image and the photoelastic stripe image, the fused image enables the stress prediction model to capture more key clues related to stress distribution at the same time, and the stress distribution can be predicted more accurately.

Systems and methods for evaluating progression of an anatomical object over a plurality of timepoints are provided. Longitudinal medical images of an anatomical object of a patient acquired over a plurality of timepoints are received. For each respective timepoint of the plurality of timepoints, features are extracted from the longitudinal medical images acquired at the respective timepoint using a machine learning based feature extractor network and the anatomical object in the longitudinal medical images acquired at the respective timepoint is analyzed based on the extracted features using a machine learning based prediction model. Progression of the anatomical object over the plurality of timepoints is evaluated based on results of the analyses using a machine learning based progression model. The evaluation of the progression of the anatomical object over the plurality of timepoints is output.

The invention discloses an identification method suitable for retinal artery occlusion. The method comprises the following steps: S10, acquiring a CFP image and an OCT image; s20, carrying out differential preprocessing, and carrying out standardization, enhancement and normalization processing on the obtained image; s30, performing feature fusion by using a double-model freezing mode, including the steps of performing general feature extraction by using a basic model, performing local feature enhancement by using a small model, and performing feature fusion; and S40, carrying out multi-modal dynamic decision making, predicting the fused data through a classifier to obtain a prediction result, and integrating multi-image prediction results by adopting a maximum probability selection strategy to obtain an identification result of the retinal artery occlusion. According to the method, the advantages of the multi-modal image can be effectively fused, and the powerful deep learning model is utilized to realize the new method of accurate and automatic identification, so that the defects of the prior art are overcome, and efficient and reliable technical support is provided for early identification of the RAO.

The present disclosure provides an abnormal tissue growth prediction method, device, electronic equipment and storage medium, the method comprising: obtaining an abnormal tissue growth prediction model; obtaining a historical sample sequence and a target growth duration, the historical sample sequence being arranged in chronological order by M historical samples obtained by examining abnormal tissues; performing a time sequence feature extraction operation; inputting the time sequence features of the historical sample sequence and the Mth historical sample with added noise data into a generative model, outputting post-growth image noise and post-growth abnormal tissue segmentation results; based on a preset denoising formula, using the Mth historical sample with added noise data and post-growth image noise to obtain post-growth image prediction results. In this way, the recurrent neural network for extracting time dimension information is embedded into the generative model for extracting spatial dimension information, improving the performance of the model and making the prediction results of abnormal tissue growth more accurate.

The application relates to the field of embodied intelligence and autonomous navigation, in particular to a wheelchair navigation method and device, equipment and medium, wherein the method comprises the following steps: acquiring a current environment image, a navigation instruction and a historical environment image of a wheelchair; extracting an environment structure feature of the current environment image, and determining whether a target environment is an indoor environment or an outdoor environment based on the environment structure feature; when the target environment is the indoor environment, generating a bird's-eye view feature map according to the current environment image, and combining the navigation instruction to navigate; when the target environment is the outdoor environment, predicting a three-dimensional point cloud, a relative pose and a driving trajectory based on the current environment image and the historical environment image, and combining the navigation instruction to navigate. Therefore, the problems of poor unified representation capability of perception, mapping and decision, insufficient spatial geometric expression and real-time performance, and lack of continuous environment updating and memory capability in the related art are solved.

Systems and techniques are disclosed for determining discrepancies between conditions and parameters associated with an image. An image processing model may generate output indicating predicted values for an image using the image as input, while a text processing model may generate output indicating corresponding predicted values for an image using textual data associated with the image as input. A comparison of the output data may be performed to determine discrepancies between values. Discrepancies that are sufficiently significant and relevant may be reported for additional analysis.

To provide a picture prediction method, encoder, decoder and storage medium that use a Matrix-based Intra Prediction (MIP) mode.SOLUTION: An encoder sets a value of an MIP mode parameter as indicating the use of the MIP mode and writes it into a bitstream; determines the MIP mode of a current block; determines, based on the MIP mode, prediction values for luma and chroma components corresponding to the current block; and writes the MIP mode of the current block into the bitstream. The decoder parses the bitstream and determines an MIP mode parameter of the current block; if the MIP mode parameter value indicates that the current block uses the MIP mode to determine an intra prediction value of the current block, parses the bitstream, determines the MIP mode of the current block and prediction values of luma and chroma components corresponding to the current block based on the MIP mode; and decodes the current block based on the prediction values.SELECTED DRAWING: Figure 7

The application discloses a cell sorting method and system based on an image acoustic flow control cell sorting model and a computer readable storage medium, the image acoustic flow control cell sorting model comprises a cell image recognition module and a cell ejection module, and the cell sorting method comprises the following steps: acquiring an original cell image set and a predetermined sample cell image set; inputting the sample cell image set into the cell image recognition module for feature extraction to obtain image category information of the cell image; inputting the image category information and the sample image set into the image acoustic flow control cell sorting model for training; inputting the original cell image into the trained image acoustic flow control cell sorting model for image prediction to determine a target image corresponding to the image category; and ejecting a cell corresponding to the target image to a preset collection area based on the cell ejection module. In the embodiment of the application, the automatic classification of cells can be performed by using a label-free method, so that the purification and collection of target cells are realized.

The application discloses a method for rapid prediction of existing building damage based on image-text multimodal data, comprising the following steps: obtaining post-earthquake component damage images and corresponding text descriptions, pre-processing, and then respectively using a pre-trained image feature extraction model and a pre-trained text feature extraction model for encoding to obtain image features and text features, and using a bottom freezing and upper fine-tuning strategy to adapt to post-disaster damage representation; cross-attention mechanism is used to realize cross-modal information interaction and fusion to form fusion features; the image features, text features and fusion features are respectively input into an image prediction head, a text prediction head and a multimodal prediction head, and a component damage level prediction result is output; in the training stage, modal discarding and knowledge distillation are introduced, so that the single-modal prediction head learns multimodal joint discrimination information, and stable prediction is realized; the application improves the accuracy and robustness of the damage prediction.

Disclosed are a system and method for analyzing a stool image, which derive the state of the large intestine of a user by analyzing a stool image by using a pre-trained deep learning model. The system for analyzing a stool image may include an input unit configured to receive stool images of a user, a data set generation unit configured to generate a data set by grouping the stool images in chronological order, an analysis unit configured to derive information on a stool state of the user or information on a state of the large intestine of the user by analyzing the data set, and an output unit configured to output the information on the stool state or the information on the state of the large intestine. The endoscopic activity of ulcerative colitis can be predicted from a camera image of stool even without performing a colonoscopy.

The invention provides an image compression method and device, electronic equipment and a readable medium, and relates to the technical field of image processing, and the image compression method comprises the steps: obtaining an input image and a target code rate; inputting the input image into the content feature analysis model to input a feature vector of the image, the feature vector being capable of reflecting a perception feature of the input image; and inputting the target code rate and the feature vector into a joint parameter decision model to generate a spatial scaling parameter and a quantization parameter adaptive to the input image and the target code rate. According to the embodiment of the invention, under the limitation of the target code rate, the details of the visual salient region are reserved preferentially according to the adaptive control of the image prediction optimal compression parameter, and the method is widely applied to application scenes such as extremely-low bandwidth image transmission, terminal front-end compression and Internet of Things image perception.

The application provides a transparent object reconstruction method and system based on RGB-D camera consistency depth prediction, acquires an RGB image, a depth image and camera internal parameter information of an RGB-D camera containing a transparent object; uses a transparent object segmentation network to obtain a mask of the transparent object based on the RGB image; applies the mask on the depth image to acquire a depth value of a non-transparent object region, and uses the camera internal parameter information to acquire point cloud of the image in a three-dimensional space; uses a pre-trained consistency depth prediction neural network to perform depth image prediction based on the mask, the RGB image and the three-dimensional space point cloud to obtain a restored depth image; and performs three-dimensional reconstruction based on the restored depth image, the RGB image and the camera internal parameter information to obtain a final result. The application can accurately reconstruct a scene with a transparent object as a foreground.

The embodiment of the application discloses an image prediction method, an encoder, a decoder and a storage medium. The method comprises the following steps: obtaining an initial prediction value of a to-be-predicted image component of a current block in an image through a prediction model; and performing filtering processing on the initial prediction value to obtain a target prediction value of the to-be-predicted image component of the current block.

The embodiment of the invention discloses an image prediction method, an encoder, a decoder and a storage medium, and the method comprises the steps: obtaining an initial prediction value of a to-be-predicted image component of a current block in an image through a prediction model; and filtering the initial prediction value to obtain a target prediction value of the to-be-predicted image component of the current block.

The application provides a simulation prediction analysis method for laser polishing of a microtopography and roughness of a diamond surface, and comprises the following steps: creating a mapping relationship data table of laser parameters and laser spot size and power density; establishing a mapping relationship model diagram of laser spot power density, pulse width, laser incidence angle and diamond surface ablation pit depth; constructing an analysis model of a laser spot stacking state; generating a simulation image of the microtopography of the laser polished diamond surface; and predicting the roughness of the laser polished diamond surface. The application takes the machining parameters of laser spot power, pulse width, repetition frequency, defocusing amount, incidence angle, transverse scanning rate and longitudinal stepping interval as inputs, takes the microtopography and roughness of the laser polished diamond surface as outputs, and establishes a mapping relationship between the inputs and the outputs through simulation analysis of the size of the ablation pit, the spot scanning path and the spot stacking state, so as to guide the optimization of the laser polishing process parameters.