404 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 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.

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.

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.

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.

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.

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.

A medical image generation method is provided. A PET dynamic image of a target part of a scanned object is obtained, a quality analysis is performed on the PET dynamic image to determine a first single frame image that meets quality enhancement requirements, the first single frame image corresponding to an initial single frame duration. A second set of scanning data corresponding to a first single frame duration of the PET dynamic image are obtained when a count of coincidence events corresponding to the first single frame image is less than a count threshold. The first single frame duration is obtained by extending the initial single frame duration. A target single frame image of the target part is constructed based on the second set of scanning data, and the count of coincidence events corresponding to the second set of scanning data is greater than or equal to the count threshold.

The present invention relates to the technical field of molds, and provides a tire mold insert assembly structure and an application method thereof, which include a shaping frame, an insert and a tightening member. Among them, the shaping frame is composed of a plurality of single frames arranged in a circular array, and each single frame is provided with an installation hole; the insert is inserted into the installation hole, one end of the insert abuts against the inner wall of the shaping frame, the other end extends to the outside of the shaping frame, and a limiting portion is provided; the tightening member is located outside the shaping frame, the tightening member is in a ring structure, and the tightening member is used to apply a force away from the center of the shaping frame to the limiting portion, so that the insert and the shaping frame are relatively fixed. By arranging a plurality of single frames to form the shaping frame and providing installation holes in the single frames, the installation of the insert can be realized. By replacing the inserts with different patterns, different models of tires can be produced by this mold without resetting the whole set of molds, which effectively reduces the mold opening cost while meeting the requirements of product replacement.

The invention discloses a lightweight traffic sign recognition method fusing time sequence characteristics in a shielding scene, and relates to the field of computer vision and intelligent traffic, in particular to a traffic sign recognition technology. Respectively constructing a single-frame data set with slight occlusion and a time sequence data set with severe occlusion; for a street scene image containing a complex background, training a lightweight target detector YOLOv8-N based on an existing bounding box label in a data set; secondly, a depth separable convolution structure with MobileNetV2 as a trunk network is adopted, and a high-dimensional feature map is extracted from the candidate mark image; then, a space attention mechanism is introduced, weighted correction is carried out on input features, an unshielded region is strengthened, and a shielded and noise region is suppressed; and finally, extracting a feature sequence of a historical frame of a heavily shielded sample, and inputting the feature sequence into an extended long-short-term memory network xLSTM for time sequence fusion.

The invention discloses a single-frame unmanned aerial vehicle image pixel positioning method and system based on an elevation map, and relates to the technical field of unmanned aerial vehicle image processing and geographic positioning, and the method comprises the steps: obtaining a single-frame original image shot by an unmanned aerial vehicle, unmanned aerial vehicle attitude information during shooting, camera internal and external parameters, and a digital elevation map of a shooting region; according to the internal reference of the camera, converting the pixel coordinate of the target pixel in the original image into a normalized plane coordinate system of the camera, and carrying out coordinate distortion correction to obtain the back projection point coordinate of the target pixel; converting a camera-unmanned aerial vehicle body coordinate system and a body-geographic inertial reference coordinate system according to camera external parameters and unmanned aerial vehicle attitude information to obtain normalized ray direction vectors corresponding to target pixels; an earth surface height constraint is constructed based on a digital elevation map, and an intersection point of a ray and the ground is determined by performing iterative approximation on a space point along a ray direction. According to the invention, rapid and accurate conversion from the single-frame unmanned aerial vehicle image pixel to the geographic coordinate can be realized.

The invention discloses a structured light single-frame reconstruction method based on layered stereo matching. A structured light active measurement system is regarded as a binocular passive measurement system, random speckles are adopted as projection patterns, a single-frame depth estimation model is provided based on the stereo matching principle, and the single-frame depth estimation model mainly comprises a depth initialization module, a depth updating module and a depth refinement module. Compared with a traditional structured light three-dimensional reconstruction algorithm, the method has the advantages that three-dimensional reconstruction of an object can be completed only through a single-frame image, and the method has higher dynamics; compared with other deep learning based on a phase shift method, the method does not need to solve three-dimensional data through intermediate variables such as a wrapped phase and an absolute phase, and avoids the influence of an accumulation error on a final result. In addition, system calibration parameters are used as a priori construction model, and a simulation data set is adopted to assist training, so that the model has higher generalization performance.

The invention provides a servo mechanism control system and method based on TSN network communication, and relates to the field of servo mechanism control. According to the system, a TSN network communication interface is adopted, and 1Gbps instruction receiving and sending can be achieved. Through optimization mechanisms such as time synchronization, flow scheduling and frame preemption, it is guaranteed that key data streams arrive at destinations within strictly specified time, delay and jitter are extremely low compared with RS422, only the us level exists, and state feedback of a servo mechanism is more real-time and accurate; in a communication period of 1ms, the maximum transmission of a single frame at a rate of 1Gbps can achieve 1500 bytes, and the capacity is greatly improved compared with that of RS422, so that a servo mechanism can return more state information; an optical cable is adopted as a communication medium, electromagnetic interference is avoided, and the real-time performance and stability of the communication process of the servo mechanism are guaranteed; and for a high-dynamic precise control scene, the communication period can be shortened to 500us, the state feedback of the servo mechanism is real-time, and further precise control is facilitated. The problems that a traditional servo mechanism is low in communication rate, limited in single-frame capacity, poor in anti-interference capacity and the like are solved.

The invention discloses an intelligent electric meter sensor sensing control method and system based on an improved algorithm, and relates to the technical field of intelligent electric meter sensor sensing control, and the method comprises the steps: setting an MCU timer to be in an automatic heavy load mode, starting interruption, carrying out the dual-channel scanning of CT / PT, taking a mean value as static bias, and presetting a spine elimination threshold value. And correcting the fluctuation threshold in real time, adaptively switching the resolution according to the interference intensity, and synchronously collecting data and storing the data in an annular buffer area. And performing single-frame median rejection, resampling when the deviation of continuous frames exceeds the limit, and accumulating residual energy in a sliding window through median filtering and self-adaptive low-pass filtering. And dynamically adjusting the noise covariance based on residual statistics, and outputting a high-confidence current / voltage estimated value through lightweight Kalman filtering. And mapping the normalized error energy into a perception behavior of CT / PT. According to the invention, adaptive optimization can be realized under different load conditions in an environment with static resolution and bandwidth setting.

The embodiment of the invention provides a data fusion method and device of a sensor, equipment, a medium and a product, and the method comprises the steps: obtaining a target recognized by a current frame of the sensor, calculating the correlation degree of the target and each track based on a historical data set of each track, which comprises the correlation relationship between the historical target and the track, and obtaining the correlation degree of the target and each track; and finally, selecting the target with the highest association degree to be fused with the track. According to the scheme, the association relationship between the target and the track is determined by introducing the association data of the historical target under the track, the short view defect of single-frame similarity matching is made up, and the continuity of association between the target and the track can be enhanced based on the association data to avoid mismatching and track breakage, so that the accuracy of a sensor data fusion result is improved.