48 results about "Visual marking" patented technology

The application discloses a kind of based on natural language's traffic event's question and answer method and system, belong to intelligent question and answer field, first acquisition roadside image and user query, respectively by visual encoder and natural language encoder extraction multiscale visual feature set and text feature;Then utilize pre-training cross-modal intent alignment and focus module, with text feature as guide to spatial attention anchor and cross-modal intermingling to visual feature, output joint representation feature;Again, the feature is input light-weight decoder obtained by structured logic decoupling and migration strategy distillation, so that it executes step-by-step logical inference and coordinate regression based on spatial response graph in parallel, synchronously output natural language answer text and target bounding box coordinates;Finally generate traffic event question and answer result containing semantic answer and visual mark.The application effectively solves the problem that prior art cannot accurately and efficiently based on natural language to traffic event intelligent question and answer.

This invention discloses an AI-interactive AR (Augmented Reality) cultural wall system for cultural tourism, belonging to the field of AR system technology. The surface of the cultural wall is equipped with a pre-embedded feature set, including visual markers, differentiated texture features, 3D depth variation areas, and specifically designed interactive hotspots. These features provide the system with stable visual benchmarks, semantic distinction criteria, depth cues, and precise positioning anchors. The system includes: a feature point acquisition module, used to collect the feature set on the cultural wall through a multimodal feature extraction algorithm to construct a 3D spatial map with semantic annotations; a user positioning and pose estimation module; an AI content generation engine; and a spatial adaptation and rendering module. Through the cultural wall feature set and the multi-module collaborative system, precise spatial integration of virtual content and the physical wall surface is achieved, along with intelligent dynamic content generation based on user behavior, effectively enhancing the immersion, interactivity, and personalization of the AR cultural experience.

Method and apparatus for training multimodal large model and method and apparatus for image question answering are disclosed, which relates to artificial intelligence technologies such as large models, deep learning, natural language processing, and computer vision. The method for training multimodal large model includes: obtaining an initial sample image, a sample object in the initial sample image, and a location information of the sample object; obtaining a target sample image including a sample visual marker based on the initial sample image and a target image region corresponding to the initial sample image; obtaining a sample question corresponding to the target sample image based on the sample visual marker, and obtaining a sample answer corresponding to the sample question; training an initial multimodal large model based on a target training sample constituted by the target sample image, the sample question and the sample answer to obtain a target multimodal large model. The method for image question answering includes: obtaining a target image including a target visual marker and a target question; inputting the target image and the target question into the target multimodal large model to obtain a target answer. The present disclosure enables the target multimodal large model to effectively understand the target visual marker in the target image, thereby improving the accuracy of the target answer.

This invention relates to the field of automation technology and proposes a semantic disambiguation click method and system based on RPA and context-aware visual tags. The method involves receiving user natural language prompts and interface screenshots; identifying UI elements and generating basic visual tags using computer vision technology; injecting contextual features into the tags based on spatial geometric calculations to form enhanced visual tags; semantically expanding the enhanced tags using a first-stage multimodal large model to supplement functional labels and contextual descriptions; then, using a second-stage large language model to perform deep matching and disambiguation between user intent and semantic tags to accurately select target elements; finally, mapping the element position to actual screen coordinates to drive RPA to execute the click operation, and forming an operation loop through visual verification. The corresponding system includes a series of functional modules that perform the above steps.

A parking control method, electronic device, and storage media. The method comprises the following operations. A position of a reflective target is determined by a LiDAR. A vehicle is controlled to move to a recognition area of a visual marker. A pose relationship between the vehicle and the visual marker is determined based on a visual sensor within the recognition area of the visual marker. A parking target point is determined based on the pose relationship. The vehicle is controlled to move toward the parking target point.

Machine learning models for autonomously navigating the web and / or apps are provided that receive (i) visual input that includes one or more images representing the current, and optionally one or more prior, visual states of the website or app being navigated, and (ii) textual input that represents a user request to be implemented by the model, the current HTML or other code defining the website or app, and optionally additional text like prior model outputs that represent prior steps in the navigation of the website or app by the model. The visual input and textual input are transformed into respective sets of visual and textual tokens, and this multi-modal token set is applied to an encoder. The output of the encoder is applied to a decoder to generate textual output that represents an action to take, relative to the code defining the website or app, to pursue the user request.

The application discloses a transcranial magnetic stimulation navigation calibration system and a control method, a control device and a medium thereof, relates to the technical field of medical rehabilitation, and comprises a treatment paddle and a calibration structural member installed on the treatment paddle. A fixed spatial offset relationship exists between the geometric center of the visual mark of the calibration structural member and the stimulation function center of the treatment paddle. According to the spatial offset relationship, the visual mark pose recognized by each depth camera is converted into a first pose of the stimulation function center in the corresponding depth camera coordinate system, instead of being based on the approximate estimation of the paddle surface geometric center, so that the offset error between the stimulation function center and the paddle surface geometric center is avoided, and the calibration accuracy of the stimulation function center of the treatment paddle is improved. Meanwhile, the spatial transformation relationship between the depth camera coordinate system and the mechanical arm base coordinate system is independently calculated for each depth camera. In combination with the spatial transformation relationship and the spatial offset relationship, each depth camera establishes an independent calibration link, does not depend on the synchronous work of multiple cameras, and meets the demand of a complex scene with frequent occlusion in clinical application.

This invention discloses a text-guided spectral image description method, system, device, and storage medium, relating to the field of image processing technology. The method includes acquiring a target image; preprocessing the target image to obtain an image tensor; performing image visual encoding on the image tensor to obtain an image visual marker sequence; generating a spectral feature tensor based on the image tensor; and converting the spectral feature tensor into a spectral marker sequence. This invention simultaneously extracts the visual marker sequence and spectral marker sequence of the target image and constructs a joint global representation. It introduces guiding text to enhance the spectral-visual relationship under text guidance. Furthermore, through step-by-step processing of initial generation and candidate text refinement, it can improve the image description's ability to capture detailed information to a certain extent, optimize the fusion effect of different modal features, and improve the accuracy and detail richness of the final image description result.

Embodiments of the present disclosure disclose a point cloud processing method and device, electronic equipment and storage medium, the method comprising: obtaining a reconstructed point cloud and an ideal point cloud, the reconstructed point cloud being obtained by reconstructing an image taken for a target screen, and the ideal point cloud being generated based on design data of the target screen, the target screen being provided with at least one visual marker for carrying corresponding identification information; matching the reconstructed point cloud with the ideal point cloud based on the identification information, determining an initial point set in the reconstructed point cloud that matches successfully, and determining a target point set in the ideal point cloud that matches successfully; determining a transformation parameter according to the initial point set and the target point set, correcting the reconstructed point cloud based on the transformation parameter; scaling the ideal point cloud based on the transformation parameter, completing the corrected reconstructed point cloud based on the scaled ideal point cloud, and obtaining a target point cloud; the embodiments of the present disclosure can improve the point cloud processing efficiency.

This invention discloses a phase sequence error prevention and verification method for densely insulated busbar trunking, belonging to the field of power transmission equipment installation technology. Addressing the shortcomings of existing technologies that rely on special hardware or visual markings and cannot verify phase sequence segment by segment, this method proposes: selecting the first segment to establish a reference phase sequence; aligning the near end of the unit to be connected with the reference unit without tightening it; connecting the first test end of the verification tool to the reference phase busbar and the second test end to each busbar at the far end of the unit to be connected, determining the near-end phase sequence based on continuity; tightening after confirming matching; and iteratively completing the entire line installation. This invention utilizes the axial continuity of the busbars within the busbar trunking to independently verify the phase sequence using far-end continuity detection before tightening, without relying on hardware or markings, achieving simultaneous installation and verification, eliminating the accumulation of incorrect phases from the source, significantly reducing rework costs, and is applicable to complex structures such as straight lines, bends, and branch circuits.

The application discloses a kind of test time self-adaptive target detection methods based on pre-training large model, pre-trained text encoder and image encoder are used to extract the features of input text and image respectively, then input pre-training large model to obtain initial target proposal set, after refining initial target proposal set, the visual label of each refined target proposal is generated, then it is spliced with text features to obtain extended text features, the bounding box of new target proposal is obtained by inputting extended text features and original image features into pre-training large model again, the classification probability vector and confidence based on prediction score are calculated, and the final target proposal set is obtained, finally, post-processing is carried out to obtain the final target detection result.The application realizes fine-grained, context-aware visual-language real-time alignment in the test stage, so as to improve the positioning accuracy and classification robustness of target detection.

This invention discloses a sheet metal processing device, including a processing platform, a sheet metal clamping mechanism, and multiple elastic pre-tensioning components. The sheet metal clamping mechanism includes a support, a first telescopic rod, and a clamping plate. The support is connected to the processing platform, the first telescopic rod is connected to the support, and the clamping plate is connected to the free end of the first telescopic rod. The elastic pre-tensioning components are connected to the lower surface of the clamping plate and are used to adapt to the micro-undulations and macro-deformations of the sheet metal surface, and to clamp the sheet metal. The clamping plate has a cavity, and the elastic pre-tensioning components include a connecting rod, a contact head, a limiting plate, and a second spring. The connecting rod is slidably connected to the lower wall of the clamping plate. The contact head is connected to the lower end of the connecting rod. This invention, through multiple independently floating elastic pre-tensioning components, can adaptively clamp the sheet metal surface unevenness, avoiding local overpressure damage or insecure fixation; the clamping force detection and visual marking functions facilitate sheet metal quality control and subsequent sheet metal processing.

The present invention relates to an image recognition performance optimization method and system, and provides an image recognition performance optimization method and system, in which a visual token is re-tokenized in units of semantics in order to improve image recognition efficiency.

The present application provides for synchronizing supplemental extended reality (XR) media content with media content on a primary device. A media server may stream media content on a client device and then receive a request to provide XR content supplementing the media content. The media server may then transmit to the XR device (a) an image of a visual marker for determining dimensions and a location of the client device displaying the media content, and (b) timing information. The media server may then transmit to the client device instructions to display the visual marker at a time that is based on the timing information. The XR device may then capture an image of the client device at the time that is based on the timing information and determine the dimensions and the location, and then display the XR content supplementing the media content based on the determined dimensions and location.

This invention discloses a laser marking device for impeller imbalance, relating to the field of rotor dynamic balancing technology. The invention aims to solve the problems of visual deviation and inaccurate positioning caused by human factors in existing impeller balancing tests that rely on manual visual marking of the center of gravity. The device includes a balancing test apparatus and a laser device independently of the impeller's rotation axis. The laser device includes a fixed base and a testing device, which slides laterally on a slide rail of the fixed base via pulleys. A multi-degree-of-freedom adjustment mechanism, including a clamping plate, grippers, a shaft, a connecting rod, and a pin, is mounted on the fixed platform of the testing device. The laser emitting device is installed at the end of the connecting rod. This invention achieves non-contact laser-based point marking, eliminating errors caused by manual visual marking, effectively improving the accuracy of impeller imbalance location marking, and enhancing the efficiency of impeller weight reduction and correction while improving equipment operational safety.

The application discloses a PCB and an automatic double-rejection marking system and method thereof, comprising an AOI detection module, an OK plate conveying station, a scrap PCB shunting mechanism, an MES system, a laser marking module, a copper sticker attaching mechanism and an automatic shunting module. The AOI detection module determines whether the PCB is scrapped through high-resolution optical scanning and local electrical property detection; the MES system is connected with the AOI detection module and the shunting mechanism in data, and automatically generates a scrap serial number and records data. The laser marking module ablates characters on the scrap PCB according to the scrap serial number, forming a permanent visual mark that cannot be erased; the copper sticker attaching mechanism attaches a copper sticker beside the characters, forming an electrical short circuit, and realizing automatic failure of electrical property testing. The application solves the problems of unstable ink marking, easy cleaning and tampering, and high false marking rate, and completely eliminates the risk of mixed flow of scrap boards through visual permanent marking and electrical isolation double protection.

The present disclosure provides a large language model pruning method, an electronic device and a program product, and relates to the technical field of computers. The large language model pruning method comprises: performing visual coding on an input video to obtain first visual markers of each video frame image; based on the overlap rate between adjacent video frame images, the input video is divided into multiple video segments; the feature vector and the final pruning rate of each video segment are determined; the processing process of the large language model on the input video is divided into multiple stages to obtain second visual markers of each video frame image; based on the first similarity between the second visual markers of adjacent video frame images, the original kernel matrix for the determinant point process of each stage is constructed, and the sub-kernel matrix is extracted; for each sub-kernel matrix, the optimal visual marker subset is obtained by maximum a posteriori inference; the optimal visual marker subset is combined as the pruning result of each stage, and the pruning result is taken as the input of the next stage.

This invention relates to the fields of intraoperative functional monitoring in neurosurgery, invasive brain-computer interfaces, and multimodal synchronous data acquisition. It discloses a device for synchronous acquisition of multimodal hand movements during awakening surgery and brain-computer interface calibration. The device includes: a flexible sensing finger sleeve or finger back strip, fixed to one or more fingers of the patient, with a flexible bending sensor and local visual markers integrated on it; a thin task board, thick enough to be inserted through gaps in the patient's hand restraints during surgery, with global visual calibration points, a pressure sensing module, and synchronous light-emitting diodes (LEDs); and a control and acquisition module that, when a set motor event is detected, simultaneously triggers three synchronous hardware outputs: outputting a behavior log to a task computer, controlling the flashing of the synchronous LEDs, and sending a TTL event code to a neural signal acquisition system. This invention enables rapid deployment in confined spaces and solves the problem of low time alignment accuracy of multimodal data.

The present invention relates to a laundry washing machine (100) comprising a machine body (200); a dispenser (300) having a plurality of compartments (310) for loading cleaning agents in fluid form; a dispenser housing (210) provided in the machine body (200) for the placement of the dispenser (300); and a plurality of visual markers (400) for indicating the type of cleaning agent in at least one compartment (310).

The disclosure provides a natural language processing and artificial intelligence-based tender automatic writing method and system, an electronic device, a storage medium and a program product to solve the problems of low efficiency, omissions and compliance risks in tender writing. The method comprises: importing a tender document and parsing, performing deep semantic understanding on the parsed tender document, and extracting structured key information; based on the extracted format specification information, automatically standardizing the layout of the tender document; based on the structured key information, intelligently retrieving and matching related text content from a pre-constructed industry knowledge base, and automatically filling into the corresponding chapters of the tender document; identifying parts that cannot be automatically generated or need to be customized, and performing visual marking to prompt manual writing; performing multi-dimensional automatic checking on the tender content after preliminary filling and writing; and outputting the checked tender template file. The disclosure effectively improves the efficiency and standardization of tender writing.

The present application belongs to the field of visual monitoring, and particularly relates to a visual marker automatic selection method and system for a mobile robot, comprising: acquiring monocular image frame sequences and odometer data at continuous time points; detecting and decoding visual markers for each image to obtain measurement poses of the visual markers, and adaptively assigning uncertainty covariance matrices according to pixel areas, detection confidence or distances of the visual markers; calculating traces of the covariance matrices corresponding to the measurement poses, screening out effective observations satisfying an accuracy threshold, and selecting the one with the minimum trace as a representative observation when there are multiple effective observations at the same time; calculating a pose change between adjacent representative observation time points according to the odometer and assigning a motion uncertainty covariance matrix; taking the pose at the representative observation time point as an optimization variable, constructing a weighted optimization objective function, and solving the objective function to obtain an optimized pose and output a current positioning result.

An online system automates the generation and presentation of video content for physical objects. The system stores object data and 3D models for users, including models of objects for sale. Users can generate and manage visual markers on 3D models to highlight specific features. The system generates videos depicting the physical objects using the stored models and object data, and produces synchronized audio voiceovers describing the objects' features. Script generation prompts are created based on object data and input to a generative language model to generate scripts for the voiceover. The system generates voiceover audio data using text-to-speech or generative AI models and combines the audio with the video. The resulting video and audio data are transmitted to a client device for user presentation and interaction. The system provides an integrated workflow for creating, editing, and presenting detailed, object-specific video content with automated narration.

This application provides a method, apparatus, system, device, medium, and product for controlling the landing of a drone, relating to the field of drone control technology. The method includes: acquiring image data from a visual marker tag, radio frequency data transmitted by multiple radio frequency tags (RFID tags) affixed to the visual marker tag, and environmental data of the drone; determining target control parameters based on the image data, RFID data, and environmental data; and controlling the drone to land based on the target control parameters. This application uses RFID tags affixed to the visual marker tag. These RFID tags are unaffected by lighting conditions, enabling accurate drone positioning even in low-light conditions or when the visual marker tag is partially obscured. Furthermore, utilizing multimodal data to determine the parameters for controlling the drone improves the accuracy of the control parameters compared to a single sensor or tagging system, thereby enhancing landing accuracy.

This invention discloses an efficient high-resolution image visual tag generation method for multimodal large-scale models. The invention includes the following steps: First, the input original image is segmented using a minimum-fill dynamic slicing method to obtain several image patches, which are then added to a set of image patches to be processed. The original image is preprocessed and also added to the set of image patches to be processed. Next, a visual encoder is used to extract multi-layer visual features from each image patch in the set of image patches to be processed. Then, the penultimate layer of features is downsampled to low-resolution features and cross-attention is performed with the high-resolution features to obtain a visual tag subsequence for the current image patch. Finally, the final compressed visual tag sequence is obtained. This invention can efficiently generate visual tag sequences and is applicable to various visual reasoning tasks such as visual question answering, document question answering, and optical character recognition, providing an efficient and accurate visual context representation method for multimodal large-scale language models.

Methods and systems are provided for diagnosing chronic spinal pain of a patient that include receiving and storing an image of the patient's back or neck. The image includes a first set of visual indicia corresponding to anatomical landmarks of the patient's back or neck as well as a second set of visual indica corresponding to locus of pain experienced by the patient. The image is processed to generate feature data representing spatial location and pain characteristics of point(s) corresponding to locus of pain experienced by the patient. The feature data is input to an expert system that generates first output data representing a predicted diagnosis of chronic pain of the patient and optional second output data representing a confidence level associated with the predicted diagnosis as indicated by the feature data input. A lookup table provides a treatment corresponding to the first output data and the optional second output data.

A modular combat sport system is provided, comprising a strike-responsive target assembly, a flexible melee training weapon, a wearable armor platform, and an integrated team-based gameplay framework. The target assembly includes a housing that emits a visual marking substance—such as paint, powder, or dye—and may optionally produce an audible tone. The weapon features a resilient core with an energy-absorbing sheath and a compressible tip, enabling safe physical engagement. The modular armor assembly supports attachment of targets and provides protection. The sport system enables competitive team combat with defined player roles, fog-of-war mechanics, and objective scoring based on visual and / or auditory confirmation of legal strikes. The system allows for specialist player roles, referee oversight, and scalable rule variations, creating a hybrid between medieval melee reenactment and modern tactical games like paintball.

The application discloses a kind of visual mark robust positioning method and system for complex illumination environment, belong to robot positioning technical field, its technical scheme main point is including, obtain the current corresponding multi-source data set of inspection robot and carry out pre-processing, multi-source data set includes RGBD data and point cloud data;According to the laser composite feature matrix obtained according to pre-processed point cloud data;According to the visual composite feature matrix obtained according to pre-processed RGBD data and residual network model;According to laser composite feature matrix and visual composite feature matrix, obtain fusion matrix;According to the position of the current inspection robot where according to the fusion matrix corresponding to adjacent last time fusion matrix, the present application extracts visual composite feature matrix by means of residual network model, based on laser composite feature matrix, the geometric structure information of environment is completely retained, through the fusion of laser and visual composite feature matrix, improve the position perception accuracy of inspection robot in complex scene.

This invention relates to the field of medical device technology, specifically to a wound repair traction device based on visual recognition. It includes a circular frame, a movable clamp, and two arc-shaped plates movably disposed inside the circular frame. A horizontal plate is disposed on the side of the circular frame, and two compression pads that contact the leg are disposed at the bottom of the horizontal plate. An air bladder column is disposed between the horizontal plate and the movable clamp, and a pneumatic detection mechanism connected to the movable clamp is disposed at the end of the air bladder column. A clamping and mounting mechanism is disposed between the circular frame and the horizontal plate. This wound repair traction device based on visual recognition, through the use of components such as screw one, screw two, knob one, and knob two, drives the arc-shaped plates and compression pads to perform multi-degree-of-freedom adjustments, enabling them to closely and stably adapt to the anatomical shape of different patients' legs. This mechanical rigid clamping reduces reliance on easily interfered visual markers, provides a stable physical reference for the visual recognition system, and avoids accidental interruption or deviation of the traction force.