487 results about "Single frame" patented technology

The invention discloses a multi-mode counterfeit video detection method based on a multi-head addition cross attention mechanism, and belongs to the technical field of video counterfeit detection. The method comprises the following steps: preprocessing a video stream, decomposing a single-frame positioning face, and extracting an audio to generate a Mel spectrogram slice; the 3D convolutional network extracts video spatio-temporal features and motion differences, and the filter bank extracts audio features in combination with the residual network; audio features are mapped to a video alignment space through asymmetric projection, the video features are subjected to bidirectional interaction with an audio input multi-head addition cross attention module after being subjected to time sequence coding, and audio dominant and video dominant features are generated and are cascaded and fused with original features; and constructing cross-modal similarity loss constraint feature distribution, and fusing feature dynamic weighting and time sequence compression to output four classification probabilities of audio-visual double true, audio-visual double pseudo, video pseudo-audio true and video pseudo-audio pseudo. The multi-mode counterfeiting recognition precision is improved, and texture abnormity and audio and video mismatch features are captured.

The invention discloses a single-frame infrared thermal imaging building outer wall hollowing detection method and system based on heat flow evaluation. The method comprises the following steps: acquiring a single-frame infrared thermal image and a visible light image which are synchronously acquired; generating an infrared thermal imaging sequence based on the single-frame infrared thermal image and the Green function of the heat conduction control equation; performing time-space frequency domain transformation and column harmonic function inversion on the infrared thermal imaging sequence, and inversely transforming back to a time-space domain to generate a virtual heat flow image sequence; sequentially carrying out binarization processing and discrimination function enhancement processing on the single-frame virtual heat flow image to obtain an outer wall enhanced binarization image; identifying an interferent from the visible light image by adopting a target detection network; and carrying out registration and mapping on an identification result in the visible light image and the infrared thermal image, and removing interferents in the outer wall enhanced binarization image to obtain a pure outer wall binarization image. The method has the advantages of high environmental adaptability, high recognition precision and the like, and is suitable for safety detection and maintenance of building exterior walls.

The invention discloses a speed measurement method based on single-frame multiple exposure, and belongs to the field of hypersonic velocity field measurement. The implementation method comprises the following steps: recording equal time interval change of molecular marking in a flow field along with movement of the flow field through single-frame multi-time exposure imaging of an image enhancement camera, and realizing continuous and rapid acquisition of speed information within a short time after single-time marking. The morphological position after the time interval is determined according to the molecular marker, and the velocity field with spatial resolution in the flow field is extracted by mapping the marker before and after displacement, so that the velocity measurement precision is improved. Compared with other two-dimensional velocity field acquisition methods based on femtosecond laser molecular marker tracing velocity measurement, the method provided by the invention does not need a complex focusing lens combination, can obtain spatial positions of molecular markers at different moments only through isochronous continuous triggering exposure of a camera, thereby obtaining velocity flow field displacement information. And corresponding acceleration information can be further deduced according to the obtained velocity field. The method is especially suitable for hypersonic velocity field measurement.

The invention discloses a WebGL-based mass data particle system rendering method and device, and belongs to the technical field of computer graphic processing. The method comprises the following steps: firstly, selecting the type of a particle emitter according to scene requirements, then constructing a primitive object based on points or quadrangles, and configuring three-dimensional coordinates and texture parameters; aggregation rendering is carried out on a single-frame primitive array through a WebGL vertex shader and a fragment shader, and the drawing calling frequency is reduced; the method comprises the steps of dynamically adjusting particle attributes, including position compensation, color gradient and explicit-implicit control, determining a rendering form and playing time of an animation according to requirements of an actual application scene, and supporting fine configuration of a single particle, including texture animation, dynamic scaling and rotation parameters. When a large amount of data is processed, on the premise of ensuring the rendering quality, the rendering efficiency can be remarkably improved, the consumption of a memory and GPU resources can be reduced, the response time of the system can be optimized, and meanwhile, the interaction experience of a user can be improved.

The invention provides a page layout rearrangement / redrawing optimization method and system and computer equipment, and the method comprises the steps: carrying out real-time rearrangement / redrawing task triggering monitoring for a browser, and generating a rearrangement / redrawing task queue; performing dynamic priority grading on the rearrangement / redrawing tasks; monitoring the load proportion of the high-priority task in real time; meanwhile, determining a current frame interval and / or single-frame processing time; adjusting the task collection interval according to the current frame interval and / or the single frame processing time based on a dynamic interval adjustment rule; and according to the adjusted task collection interval, collecting the rearrangement / redrawing tasks according to task priorities, and performing unified layout calculation to generate an optimized rendering layout tree so as to execute page rendering according to the rendering layout tree. According to the method and the device, the task processing strategy is flexibly adjusted according to the real-time load condition, so that the page can still stably and smoothly run under the high load, the jamming phenomenon is avoided, and a more reliable browsing environment is provided for a user.

The invention relates to the technical field of artificial intelligence, can be applied to business scenes of financial science and technology, medical health and the like, and discloses a training and processing method, device, equipment and medium based on time sequence performance gains. Constructing ordered frame and disordered frame contents to generate an answer result, comparing the answer result to obtain a time sequence performance gain, generating a reward signal in combination with reward intensity and the time sequence performance gain, inputting a strategy optimization process to update strategy parameters, processing a training process to generate training output, and updating a training model until a target model is obtained. And completing the task by using the target model. According to the method, complex samples are highlighted through video entropy factors, and cross-frame reasoning is enhanced in combination with time sequence performance gain, so that reward signals simultaneously reflect sample difficulty and time sequence differences, a model is prevented from depending on a single frame, and the stability and generalization ability of video reasoning are improved.

The invention discloses a space-time speckle projection three-dimensional imaging method based on multi-frame optical flow alignment. Firstly, a projector based on DLP is used for projecting a space-time speckle pattern to a measured scene, and a binocular camera synchronously collects a three-dimensional space-time speckle image. The calibration parameters of the binocular camera are used to carry out stereo correction on an acquired original speckle image, and a parallax image is generated frame by frame in combination with a coarse-to-fine single-frame speckle matching strategy. And estimating a two-dimensional inter-frame displacement field between continuous disparity maps by using an optical flow method by taking an intermediate frame disparity map as a reference, compensating motion artifacts in a space-time speckle image, and ensuring strict space-time registration of a dynamic target. And based on the speckle image after motion correction, a speckle matching strategy is expanded to a time-space domain, and high-precision multi-frame three-dimensional measurement of a complex dynamic scene is realized. The method is suitable for performing rapid and high-precision three-dimensional modeling on a moving target in an unstructured environment, and can perform accurate three-dimensional measurement on a high-speed dynamic target undergoing any translation or rotation motion.

Methods, systems, and apparatus, including medium-encoded computer program products, for computer aided design of physical structures include: rendering, to a display device of a first user, a first user's view into an extended reality environment, the first user's view being generated using a first pose being tracked for the first user; identifying that a first avatar associated with the first user has entered a shared perspective mode with a second avatar associated with a second user of the XR environment, wherein the shared perspective mode has a single frame of reference in the XR environment that is shared by the first user and the second user; and performing, while in the shared perspective mode, the rendering to the display device of the first user.

The invention belongs to the technical field of computers, discloses an agile arrangement method for multi-node sub-animations, and aims to solve the problems of tedious animation editing process, low production efficiency, low maintainability and the like and realize efficient editing and display of multi-node animations. The method comprises the following steps: firstly, creating a node animation, drawing animation nodes and animation paths, and establishing association to form the node animation; then, constructing an animation array, forming the animation array based on a time axis, a node axis and a vernier, and after node animations are loaded, displaying the node animations in a static single-frame or dynamic sequence mode by combining the position of the vernier; afterwards, the animation is agilely arranged, and the starting and ending time of the animation is rapidly positioned based on time mapping. The duration of the node animation is agilely arranged, the comprehensive arrangement of the multi-node animation is supported, and a complex animation scene is quickly formed; and finally, performing multi-mode animation preview, selecting a single-node or multi-node animation preview mode, checking a single-node animation state, or displaying a multi-node comprehensive motion situation in a complex scene.

The invention discloses a video stream defogging method and system for 5G remote control, and relates to the technical field of image enhancement. The method comprises the following steps: acquiring a foggy video in real time; inputting the single-frame foggy video into a pre-trained monocular depth estimation neural network, and reasoning to obtain a depth map with the same size as the input single-frame foggy video; performing close-shot and long-shot region division on the scene of the current foggy video frame according to the depth map to obtain a region mask with the same size as the depth map; based on the region mask and the foggy video, calculating to obtain an atmospheric light parameter; calculating the transmissivity of each pixel according to the distance estimation result of each pixel in the depth map, and obtaining a transmissivity map with the same size as the depth map; based on the depth map, the atmospheric light parameters and the transmissivity map, adaptive calculation is performed on the foggy video to realize defogging, and a defogged clear video frame is obtained; the method can adapt to different depth-of-field fog effects, realizes different depth-of-field defogging, and outputs clear video frames.

The invention provides a cross-platform video decoding method and device, computing equipment and a computer readable storage medium, and the method is combined with a factory mode to completely package bottom layer differences of all platform types, and an application layer developer does not need to care about the bottom layer platform differences. The optimal implementation can be automatically selected on different engines and platforms through a uniform API (Application Program Interface); during rendering, a single-frame memory management strategy is used, and only one frame which is currently displayed is reserved in a memory; and in the aspect of performance, the advantages of a multi-thread platform and the parallel computing characteristics of the GPU are fully utilized. According to the embodiment of the invention, through deep fusion of a cross-platform soft decoding architecture, single-frame memory optimization, multi-thread parallel, GPU acceleration and other technologies, perfect balance of cross-platform, high performance and low memory is realized, and the development efficiency of the whole project is improved.

A method for training a model for generating a representation of long-term motion from a text input comprises: training a motion encoder of an autoencoder to compress and map an input motion into a latent representation comprising a sequence of latent vectors in a discrete latent space, each latent vector representing a fixed length of motion; training a quantization module to quantize the latent vectors to a sequence of quantized latent vectors in quantized latent space; and training a motion decoder to reconstruct the quantized sequence as a sequence of single-frame pose representations. A text encoder is trained to predict a latent sequence conditioned on a text input and a duration using the mapped latent representation as a target.

A method for controlling a vehicle in an environment includes generating, via a cross-attention model, a cross-attention cost volume based on a current image of the environment and a previous image of the environment in a sequence of images. The method also includes generating combined features by combining cost volume features of the cross-attention cost volume with single-frame features associated with the current image. The single-frame features may be generated via a single-frame encoding model. The method further includes generating a depth estimate of the current image based on the combined features. The method still further includes controlling an action of the vehicle based on the depth estimate.

The invention provides a video prediction method, device and equipment based on dynamic routing and a readable medium, and belongs to the technical field of computers. The method comprises the following steps: inputting an original frame into a dynamic routing video prediction model to obtain an output prediction frame; specifically, in a dynamic routing video prediction model, an encoder is used for extracting spatial features from an original frame, and a dynamic routing unit is used for outputting a selection probability corresponding to each subunit in a predictor according to the spatial features; the predictor determines a selected target subunit according to the selection probability, and performs fusion processing on the spatial features output by the encoder by adopting the target subunit so as to predict deep features; and the decoder reconstructs the image frame based on the deep features and outputs a prediction frame corresponding to the original frame. The subunits, participating in feature calculation, of the predictor are dynamically adjustable, different scenes can be flexibly adapted, the prediction task of a single frame or multiple original frames can be processed, and the error accumulation effect in the multi-frame prediction task is effectively relieved.

The invention relates to the technical field of motion capture, in particular to a running posture continuous capture analysis method for training, which comprises the following steps of: predicting a joint position and speed of a current frame based on a joint state vector of a previous frame and a currently processed single-frame image, acquiring a priori motion posture vector, and acquiring a priori motion posture vector; performing 2D key point positioning on the current single-frame image in parallel, extracting an incomplete joint coordinate measurement value caused by limb swing shielding, and combining the prior motion attitude vector with the joint coordinate measurement value; according to the method, the joint position and speed of the current frame are predicted based on cooperative processing of the previous frame joint state and the current image feature information, continuous and stable prior attitude data are obtained, meanwhile, the missing joint position measurement value influenced by limb shielding in the single frame image is extracted in parallel, and the prediction data and the actual measurement value are fused and compensated. Therefore, the technical problem that joint coordinates are not complete due to shielding is solved, and the continuity and stability of attitude capture are improved.

The invention discloses a single-frame three-dimensional imaging method based on multi-frequency RGB (red, green and blue) stripe compression. According to the method, three groups of structured light stripe patterns with relatively prime frequencies and different directions are respectively embedded into RGB channels by utilizing channel characteristics of a color camera, the patterns are switched at a high speed by a projector, dynamic superposition is completed within single-frame exposure time of the camera, space compression acquisition of time domain information is realized, and a color superposition image fused with multiple pieces of time slice information is obtained. According to the invention, a multi-stage deep neural network architecture is designed to realize separation of fringe images of a color superposition image at different moments, phase information is accurately extracted from the decomposed fringe images, and finally unambiguous high-speed three-dimensional imaging in dynamic scenes at different moments is realized. According to the method, the time resolution of three-dimensional imaging can be remarkably improved without improving the frame rate of the camera, meanwhile, the space precision and the system stability are considered, and good universality and expandability are achieved.

The invention provides a block rendering method and system based on a video stream. The method comprises the following steps: acquiring a task; performing task preview by adopting a dependency decoupling model, and judging rendering elements and dependency among the elements; selecting a blocking mode according to the complexity of the rendering elements and the dependency relationship among the elements; splitting task blocks according to the blocking mode; rearranging the task blocks, rendering the task blocks without the dependency relationship in parallel, and rendering the task blocks with the dependency relationship according to a sequence or a time sequence; and combining the rendered task blocks according to an initial task sequence, and carrying out smooth transition processing on a connection part. By constructing the dependency decoupling model and splitting the task blocks, the rendering process is simplified, the number of parallel rendering task blocks is increased, and the rendering time of a single frame is remarkably reduced, so that the overall rendering efficiency and the computing resource utilization rate are improved.

According to the foresight sonar target detection fusion method provided by the invention, the target detection performance in a foresight sonar image is improved through fusion of a model framework and multi-frame information; the method comprises the following steps: collecting a forward-looking sonar image sequence; image preprocessing including coordinate conversion, denoising and the like; a target detection model is constructed based on deep learning, and single-frame target detection is completed; carrying out detection result fusion on the single-frame image, wherein the detection result fusion comprises coordinate transformation, image slicing, distance matrix construction, fusion threshold setting and the like; performing target tracking and multi-frame fusion on different frames of images; outputting the processed target state information; the detection robustness is improved through a model framework, and detection fusion is performed by using multi-frame information, so that the false alarm phenomenon in forward-looking sonar target detection is reduced; experimental results show that the method can effectively improve the forward-looking sonar target detection performance and reduce the false alarm rate, and has practical application value.

A system comprises an encoder configured to compress and encode data for a three-dimensional mesh. To compress the three-dimensional mesh, the encoder predicts, for a current frame of a three-dimensional mesh, vertex values of the current frame using location information from one or more preceding frames or using multiple vertex values from a single frame. Predictors and residuals for determining the current frame may be signaled in a bitstream to a decoder to decompress the three-dimensional mesh.

The invention relates to the technical field of information processing, and particularly discloses an automatic mapping method and system for an autonomous vehicle. An automatic mapping method for an autonomous vehicle comprises the following steps: constructing a generation and preprocessing model of a point cloud map, and fusing image, point cloud and position information to obtain a complete lane line map; single-frame lane information extraction is carried out on the lane line map, lane information around the vehicle at each moment in the vehicle driving process is calculated, and the lane information comprises lane line information and lane center line information; and carrying out multi-frame lane association and post-processing on the lane line information and the lane center line information, and carrying out inter-frame matching association and alignment on the lane information generated by a single frame to obtain a vector map for an automatic driving vehicle. The existing lane line information in the display environment is given, and the concept of lanes and the topological connection relation between the lanes can be represented.

The invention discloses a polarization imaging device and a polarization analysis method based on single-frame polarization collinear holography, and relates to the field of polarization imaging.According to the polarization imaging device, through combination of optical elements such as a half-wave plate, a Wollaston prism and a CCD camera, spatial multiplexing of a detector plane is carried out, and the polarization collinear holography is obtained. And four polarization collinear holograms can be simultaneously acquired by a single frame. According to the polarization analysis method, a Fourier transform auxiliary angular spectrum method is adopted to analyze an intensity distribution diagram of a polarization collinear hologram to estimate initial wavefront distribution, and an iterative finite support constraint selection method is adopted to solve the problem of twin images in the polarization collinear hologram. And further quantizing a complete Jones matrix and a complete Stokes parameter from the finally reconstructed complex amplitude information. The invention solves the problems of high complexity and long processing time caused by independent reconstruction of amplitude, phase and polarization in the traditional polarization holographic technology, and effectively solves the problem that the traditional off-axis polarization holographic imaging system limits the size of an imaging field of view.

A method for decoding a plurality of tensors forming a hierarchical representation of feature maps for a single frame from a bitstream. The method comprises: decoding a first unit of information from the bitstream; decoding a second unit of information from the bitstream; and determining a first plurality of tensors, feature maps of at least one tensor of the first plurality of tensors having a different spatial resolution from feature maps of other tensor(s). The method also comprises determining a second plurality of tensors, feature maps of at least one tensor of the second plurality of tensors having a different spatial resolution from feature maps of other tensor(s). Feature maps of each tensor of the first plurality of tensors have different spatial resolution from feature maps of each tensor of the second plurality of tensors, and the tensors correspond to the hierarchical representation of feature maps for the single frame.

The invention discloses a dam infrared thermal imaging leakage intelligent identification method fused with dynamic frame difference, and the method comprises the steps: building a frame difference graph based on an improved OTSU algorithm through a thermal infrared image sequence, so as to enhance the temperature micro-change perception capability, and achieving the high-precision identification of a leakage abnormal region through combining with a YOLOv8 model; meanwhile, the area of an abnormal region in a frame difference image sequence is extracted, the leakage evolution process is dynamically described, and a quantitative judgment mechanism is constructed in combination with fitting of an area change trend and t test, so that the accuracy and robustness of leakage outlet recognition are improved. Compared with a traditional visible light detection means, the method has earlier response capacity under various complex working conditions, the misjudgment rate and the missed judgment rate in single-frame recognition are remarkably reduced, and powerful technical support and engineering reference are provided for intelligent recognition and safety management of earth and rockfill dam leakage.

A single-frame speckle deflection measurement method based on subarea wavefront correction belongs to the technical field of optical measurement, and comprises the following steps: based on a monocular deflection measurement system, projecting a standard speckle image to a to-be-measured mirror surface from a screen, and collecting a single-frame deformed speckle image reflected by the to-be-measured mirror surface; processing the reflected image by using a Zernike polynomial projection distortion correction method based on a mark point, and correcting the projection distortion of the reflected image caused by a non-orthogonal light path; obtaining sub-pixel mapping of the reflection image and the standard image based on a sub-region wavefront fitting correction and iterative registration algorithm; and based on system calibration and camera calibration, establishing a spatial incident light model and a spatial reflected light model, obtaining surface gradient distribution of the to-be-measured mirror surface, and reconstructing a three-dimensional shape of the to-be-measured mirror surface. According to the method, a reference plane is not needed, high-reflection surface shape measurement of only collecting a single-frame image can be achieved, the dynamic range and precision of speckle deflection measurement are improved through subarea wavefront correction, and the surface reconstruction precision can reach hundreds of nanometers.

A system and method of encoding at least a plurality of tensors forming a hierarchical representation for a single frame into a bitstream. The method comprises deriving a first unit of information from a plurality of tensors forming the hierarchical representation, the plurality of tensors including at least a first tensor and a second tensor, feature maps of the first tensor having a larger spatial resolution than feature maps of the second tensor and encoding, in a first mode, at least the first unit of information into the bitstream. In a second mode, the method also comprises deriving a second unit of information derived from the first tensor; and encoding, the second unit of information and the first unit of information into the bitstream.

The invention discloses a cutting machine tool wear detection method and system based on image recognition, and relates to the technical field of wear detection, and the method comprises the steps: collecting multiple tool nose images in a short time window through triggering signal control, obtaining the time sequence information of a tool state, and eliminating the interference through combining with an image preprocessing and time sequence alignment technology. A U-Net model is used for precisely segmenting a single-frame image to obtain a tool abnormal area, dynamic association among multiple frames of images is deeply excavated through connected component analysis and time sequence consistency verification, a real abrasion area is screened out from a multi-frame sequence, and finally a tool abrasion detection result is obtained. Therefore, non-wear interference can be effectively distinguished and eliminated, and the defects that a traditional method is insufficient in precision and prone to false detection and missing detection are overcome, so that the accuracy and robustness of a detection result are remarkably improved, false alarm and missing alarm are avoided, and the intelligent production requirement of the modern industry is met.

Methods, systems, and apparatus, including medium-encoded computer program products, for computer aided design of physical structures include: rendering, to a display device of a first user, a first user's view into an extended reality environment, the first user's view being generated using a first pose being tracked for the first user; identifying that a first avatar associated with the first user has entered a shared perspective mode with a second avatar associated with a second user of the XR environment, wherein the shared perspective mode has a single frame of reference in the XR environment that is shared by the first user and the second user; and performing, while in the shared perspective mode, the rendering to the display device of the first user.

The invention provides a single-frame infrared small target detection method based on a variant convolutional neural network. The single-frame infrared small target detection method specifically comprises the steps that a picture input module carries out standardized preprocessing and primary feature extraction on an original image; the efficient channel attention module extracts channel descriptors through global pooling, efficiently calculates weights by using a lightweight network, and performs weighted adjustment on input channels; the FPN module realizes efficient fusion of cross-scale features through a space cyclic shift and channel grouping technology; the residual error convolution module adopts a structural design of combining a convolution path and jump connection, so that the problem of gradient disappearance in deep network training is effectively solved; the feed-forward module integrates an efficient channel attention mechanism, and enhances the response capability of a key feature channel through global information aggregation, channel weight generation and re-calibration operation; the model output module converts high-level features into high-precision mask images and calculates detection performance indexes in real time, all the modules work cooperatively to form a complete feature extraction-fusion-prediction processing chain, and the performance of the infrared small target detector is remarkably improved while the real-time reasoning efficiency is guaranteed.

The invention discloses a video-based method for supervising behaviors of people in a bank, which comprises the following steps of: acquiring a video sequence by using a camera in the bank, performing single-frame segmentation on the video sequence to obtain a current frame image of the people in the bank, and performing attitude estimation on the people in the bank in the current frame image by using an attitude positioning model; wherein the attitude positioning model comprises a first attitude positioning module and a second attitude positioning module; the first posture positioning module is used for acquiring the spatial relationship of key points of people in the bank in the current frame image, and the second posture positioning module is used for acquiring the current frame human body posture of the people in the bank by using the spatial relationship of the key points of the people in the bank; and finally, the current frame of human body posture is used for supervising the abnormal behavior of the personnel in the bank at present, the spatial relationship in the image is effectively utilized, the accuracy of the result is improved, and in addition, the precision and efficiency of supervising the personnel in the bank are improved by improving the training strategy.

The invention belongs to the technical field of radar data processing, and relates to a ground target elimination method for two-dimensional phased array low-altitude detection, a medium and electronic equipment. The method comprises the following steps: transmitting a single-frame waveform composed of an airspace detection wave position and a ground sidelobe rejection wave position; selecting the detection point with the maximum power as an effective detection point of the airspace detection wave position, and judging the other detection points as ground side lobe detection points and removing the ground side lobe detection points; and rejecting effective detection points with ground sidelobe rejecting wave positions of which the power is higher than that of the effective detection points from the detection points of the airspace detection wave positions. According to the method, the unmanned aerial vehicle batch starting time is greatly shortened, and unmanned aerial vehicle track batch starting is realized; the contradiction between the detection power and the batch starting time is decoupled, and the ground side lobe target is eliminated; the beam resource utilization rate is improved, and the detection response speed of the radar system to the unmanned aerial vehicle target is improved.