56results about How to "Solve positioning" patented technology

The invention belongs to the field of multi-sensor fusion systems, and discloses a multi-sensor fusion system simultaneous positioning and mapping method and a related device, and the method comprises the steps: obtaining an environment image, a laser point cloud and IMU data; scene semantic information is extracted through a multi-modal large model; dynamically generating a weight matrix of vision and laser radar relative to an IMU (Inertial Measurement Unit) by utilizing a vision and point cloud condition deep neural network in combination with scene information and IMU observation; and performing local and global optimization by combining each sensor speedometer factor, an IMU pre-integration factor and the weight matrix, and finally outputting robot state estimation and map point coordinates. Through semantic understanding and a self-adaptive weight mechanism driven by a conditional deep neural network model, a fusion strategy can be intelligently adjusted before environment change or sensor degradation, the robustness, precision and consistency of a system in an unstructured scene are improved, and the problems of positioning drift and map failure caused by fixed weight in a traditional method are solved.

The invention discloses a video content retrieval method and device based on television system voice interaction and a terminal, and relates to the technical field of video processing, and the method comprises the steps: when a video is played for the first time, extracting a picture frame from the video at a preset frequency, converting the picture frame into a multi-dimensional image feature vector and a corresponding video timestamp, and carrying out hierarchical storage in a database, constructing vectorized data containing visual semantic information; obtaining a voice retrieval instruction, performing intention recognition and semantic understanding, extracting a detection keyword, and generating a multi-dimensional retrieval feature vector; calculating a matching degree between the multi-dimensional retrieval feature vector and a multi-dimensional image feature vector of a video picture frame stored in a database, and screening out picture frames of which the similarity is higher than a preset similarity threshold to form a retrieval candidate matching set; and determining matched picture playing. The video content retrieval method is efficient, accurate and high in interactivity, and retrieval experience and operation efficiency of the user in the video watching process are remarkably improved.

The application provides a manhole anti-falling net installation construction management system and method, and relates to the technical field of municipal pipe network construction management.The application realizes real-time monitoring of a manhole construction object, collects facility number, spatial coordinates, pre-construction, in-construction and post-construction images and timestamp information, constructs a unique identifier of the construction object and generates a standardized information body;on this basis, the spatial consistency coefficient of the construction object, the construction process integrity coefficient and the construction result reliability coefficient are calculated respectively, and compared with the corresponding threshold value for evaluation;when the evaluation result is unqualified, the control strategies such as automatic repositioning, supplementary shooting and recording, rework or manual review are triggered, the abnormal data flow is suspended, and the update and recording are completed until the evaluation meets the standard.The method realizes spatial verification, process constraint and result quality control of the manhole anti-falling net construction process, and improves the construction authenticity, data reliability and management efficiency.

The utility model relates to welding equipment technical field, concretely relates to a kind of intelligent gantry submerged arc automatic welding machine, it includes: piece platform, top is equipped with multiple piece grooves along length direction, and both sides are equipped with machine tool track respectively;Welding machine tool, welding machine tool is transversely placed above piece platform, and two ends are respectively equipped with with machine tool track cooperation walking mechanism, welding machine tool top is respectively equipped with welding machine power supply corresponding each piece groove;Movable crossbeam, movable crossbeam is horizontally arranged in welding machine tool middle part, and movable crossbeam is equipped with with each piece groove corresponding cross slide;Telescopic arm, respectively with each cross slide corresponding each cross slide connection, bottom end is respectively equipped with welding torch, welding torch is connected with corresponding welding machine power supply;Controller, controller is set in one side of welding machine tool, and walking mechanism, telescopic arm, welding machine power supply and welding torch are respectively electrically connected with controller.The utility model is integrated, and automation design has significantly improved the efficiency and quality of I-shaped steel structure welding.

The application relates to the technical field of deep learning, in particular to a traffic scene image small target detection method and device, equipment and a storage medium. The method comprises the following steps: acquiring a traffic scene image and a detection result of a small target in the traffic scene image, generating a training data set according to the traffic scene image and the detection result of the small target in the traffic scene image; constructing a convolutional neural network based on a feature extraction module, a feature fusion module, a sample distribution module and a positioning classification module, and optimizing the sample distribution module and the positioning classification module of the convolutional neural network to obtain an optimized convolutional neural network; training the optimized convolutional neural network by using the training data set, updating network parameters of the optimized convolutional neural network based on gradient back propagation, and stopping the training until a preset training stop condition is met; and detecting small targets in the traffic scene image by using the trained convolutional neural network. Thus, the problems of small target sample imbalance and poor positioning capability in the related art are solved.

This invention relates to the field of book digitization and copyright management technology, and discloses a method and apparatus for intelligent recognition and structured information extraction of book copyright pages. The method includes: acquiring raw image data and performing preprocessing; locating multiple key regions using a target detection model; extracting text strings and recognition confidence scores using a sequence recognition model, executing a first feedback loop; if the recognition confidence is low or the semantics are incomplete, then performing adaptive adjustment of region coordinates; for text that passes the first loop verification, performing post-processing verification based on preset rules; if the post-processing verification fails, then activating a second feedback loop, locating potential erroneous characters and performing constrained re-recognition to correct character-level recognition errors; finally, integrating and storing the corrected text that passes all verifications. This invention significantly improves the accuracy, robustness, and automation level of information extraction by setting up dual dynamic correction loops for different error types.

The utility model discloses a printing equipment for printing fabric processing relates to printing fabric processing technical field, including the machine case, the printing mechanism is fixedly installed in the middle part in the machine case, the input of one side of machine case is equipped with preheating mechanism, and the output of the other side of machine case is fixedly installed with drying mechanism. The utility model adopts the above structure, and during the application of this technical scheme, the upper and lower printing plates are oppositely arranged through the staggered design of the printing mechanism, and the synchronous drive of the adjusting assembly is matched, so that one-time printing of both sides of the fabric can be realized during use, the fabric does not need to be turned over and reworked, the production process and time cost are reduced, positioning deviation caused by secondary processing is avoided, and the printing quality stability and production efficiency are improved, solving the problems that the existing device can only print one side, and secondary processing leads to low efficiency and positioning deviation.

The invention provides a method for detecting and positioning a large-current loop fault of a temporarily-arranged color board room on a construction site, and the method comprises the following steps: determining a fault type: determining the fault type, and carrying out the preprocessing; a range delimiting step of determining a laying path of a fault line and delimiting a scanning range of a fault area; a preliminary determination step: scanning the line in the scanning range, and preliminarily determining a fault position; and a determination step: detecting the fault position and determining a fault point. According to the method, through preprocessing, preliminary determination through scanning and final determination through detection, rapid and accurate identification of fault points can be realized, manual line disassembly is not needed, the detection time is saved, the troubleshooting efficiency is improved, and the operation safety can be ensured.

The application relates to the technical field of magnetic yoke punching sheet stacking, and particularly discloses a grabbing and stacking device for magnetic yoke punching sheets, which comprises a base frame, a storage module, a stacking module and a grabbing module driven by a gantry robot, the grabbing module comprises a visual positioning assembly for collecting visible light and near-infrared data, a solving unit for solving adsorption point coordinates based on a mask matrix, multi-modal fusion and a thermal deformation coefficient, a plurality of vacuum suction cups with a moving module, and an adsorption state detection assembly for detecting flatness and real-time pressure, and an adsorption force control unit for performing graded initial adsorption based on the flatness data and dynamic compensation based on the real-time pressure and a double safety threshold, the stacking module comprises a center positioning column and a stacking alignment unit with a radial synchronous telescopic mechanism, and the alignment rod of the stacking alignment unit is provided with auxiliary wheels, a flexible polishing belt and a pressure sensor; the application realizes high-precision positioning, self-adaptive adsorption and high-coaxiality stacking.

This utility model discloses a blanking and flattening device and operating mechanism, belonging to the field of blanking and flattening technology. The flattening device includes a fixed base, a flattening frame, a pressure mesh, and a driving assembly. The fixed base is located at the bottom of the overall flattening device; the flattening frame is placed on the upper end of the fixed base, with a hollow center, and the pressure mesh is placed within the hollow center and fixedly connected to the flattening frame; the pressure mesh protrudes from the lower end face of the flattening frame; the driving assembly is located on one side of the flattening frame, and its output component is fixedly connected to the flattening frame, enabling the flattening frame to move along the height direction. It can optimize the flattening effect and, when combined with a vision camera and an adsorption device, improve blanking accuracy.

The invention belongs to the technical field of robots, and provides a control method, device, equipment, medium and product, the control method is applied to a robot, and a pile outlet route and / or a pile return route of the robot are / is demonstrated by receiving remote control operation of a user; environment maps corresponding to the pile exiting route and / or the pile returning route are constructed; the pile outlet route is a route starting from the charging pile and proceeding to the transit point in the first direction, the pile return route is a route proceeding to the charging pile from the transit point, and the transit point is an end point of a first route proceeding in the first direction after the robot can obtain a real-time dynamic positioning fixed solution. According to the embodiment, the technical problem of positioning failure when the robot autonomously goes out of the pile and / or returns the pile can be solved, so that the use convenience and user experience of the robot are improved.

The invention discloses a Beidou / 5G fusion positioning terminal for indoor and outdoor seamless switching, and belongs to the technical field of satellite navigation and wireless communication fusion positioning. Comprising a main control module, a Beidou positioning module, a 5G communication positioning module, an IMU inertial measurement module, a storage module, a power supply module and a positioning result output module. The master control module comprises a data preprocessing unit, a three-level vertical feature extraction unit, a fusion positioning decision model unit and an indoor and outdoor seamless switching execution unit. Through a depth feature extraction system of a three-level vertical processing architecture, progressive mining from a shallow single-source feature to a middle scene association feature and then to a deep switching decision feature is realized, and a vertical processing rule that a second-level feature is based on a first level and a third-level feature is based on a second level is strictly followed, so that cross-level feature interference is avoided, and the processing efficiency is improved. The depth and accuracy of multi-source data feature mining are greatly improved, and high-reliability feature support is provided for scene recognition and switching decision.

This invention discloses a method and system for injecting knowledge into large-scale models for vertical industries, belonging to the field of large-scale model knowledge injection technology. The method involves: collecting professional knowledge data from vertical industries, performing denoising, structured transformation, and terminology standardization; dividing the data into knowledge modules and associating them with vectorized indexes and metadata tags to form a vertical industry knowledge base; parsing the semantic feature vector of the input question using a dynamic routing classifier and routing it to the target knowledge module in the vertical industry knowledge base; employing a hybrid retrieval strategy to recall relevant document sets from the target knowledge module and deduplicate them to generate enhanced retrieval results; based on this, concatenating the enhanced retrieval results with the input question using industry-specific prompt templates; and inputting the LoRA-tuned large-scale model to generate accurate answers conforming to industry standards. This invention solves the problem of insufficient knowledge adaptability of large-scale models in vertical industries, improving the accuracy and applicability of large-scale models in vertical industries.

The utility model discloses a kind of ultramicro spectrophotometer cuvette detection structure, including operation platform, the rotating seat on the top of operation platform is rotatably connected with flip cover, main road optical fiber is equipped on the operation platform, the bottom of operation platform is also installed with cuvette placing box.The utility model passes through adopting double optical path design, pedestal and cuvette light path are independent, integrated design, without physical replacement spare part, by software selection "pedestal" or "cuvette card slot" mode, can effectively avoid cross contamination, when cuvette detection, directly cuvette is placed in the fixed slot beside pedestal, light path passes through incident optical fiber, aligns cuvette light transmission hole, the utility model structure is simple, without complex light path switching system, effectively solve the positioning problem and light energy loss problem of physical switching, improve the range of measurement, accuracy and reliability of equipment, convenient operation, convenient to carry, with higher practical value.

The present application relates to ceramic matrix composite forming die, tooling and method, specifically relates to large complex closed cavity thin-walled structure ceramic matrix composite component overall forming tooling and method, in order to solve the existing technology exists ceramic matrix composite component parts assembly gap, overall strength is insufficient, service life is reduced and other insufficient, the present application large complex closed cavity thin-walled structure ceramic matrix composite component overall forming tooling includes shaping auxiliary tooling and graphite forming die, shaping auxiliary tooling includes M skeleton box shaping tooling and first bottom plate, wherein, M is greater than or equal to 1, graphite forming die includes M inner mold forming die, annular forming die and outer mold forming die, the present application large complex closed cavity thin-walled structure ceramic matrix composite component overall forming method is based on the above tooling, in turn includes preform shaping, preform densification forming, component processing, realizes overall forming.

This invention belongs to the interdisciplinary field of computer vision and 3D reconstruction, specifically relating to a generative 3D reconstruction method and system based on global constraints of location information. The method includes the following steps: S1, inputting multi-view UAV image sequences into a pre-trained feedforward neural network, which predicts sparse point clouds and the camera pose for each image; S2, extracting GNSS location information from the image sequence and using it as a global constraint to jointly optimize the sparse point cloud and camera pose, obtaining optimized point clouds and optimized camera poses in an absolute geographic coordinate system; S3, based on the optimized point cloud, employing a 3D Gaussian splashing method and utilizing the optimized camera pose for differentiable rendering optimization, generating a realistic 3D model with absolute geographic coordinates. This method, while ensuring 3D reconstruction accuracy, rapidly generates a geographic coordinate system 3D model, providing an efficient and scalable technical solution for city-level realistic 3D construction.

The present application belongs to the technical field of mechanical processing, and relates to a positioning and processing device and a processing method for a composite material sheet part. The composite material sheet part product is a sheet ring part with positioning holes on the periphery. The sheet part is within 5 mm, and the length-width ratio and thickness ratio are above 60. The processing device is composed of a base, an inner type processing block, an outer type processing block, an outer block positioning pin, an inner block positioning pin, a pressing plate, and an upper pressing plate. The device uses two pairs of positioning pins for alignment and verification, and uses the combination of the pressing plate and the base, the combination of the inner type processing block and the base, and the combination of the outer type processing block and the base to achieve three times of accurate positioning, thereby greatly improving the convenience and processing qualification rate of the composite material sheet product.

The invention provides a triangular three-dimensional amorphous alloy transformer copper foil winding fault detection method, and relates to the technical field of copper foil winding fault detection, and the method comprises the steps: obtaining the real-time magnetic induction intensity of a first winding of a transformer at a plurality of radially preset positions; and determining and positioning the turn-to-turn short circuit fault based on the magnetic induction intensity of each position and the corresponding magnetic induction intensity threshold. According to the method, based on the electromagnetic physical characteristics of the turn-to-turn short circuit of the copper foil winding, the turn-to-turn short circuit fault of the copper foil winding is judged and positioned through the magnetic induction intensity measured by a plurality of sensitive detection parts of the transformer winding; the method can effectively counteract influences brought by changes of magnetic induction intensity caused by turn-to-turn short circuits of different degrees, can adapt to fault positioning of various turn-to-turn short circuit degrees, effectively solves the problem that turn-to-turn short circuit faults of a copper foil winding type transformer are difficult to recognize and inaccurate in positioning at present, and is higher in accuracy.

The utility model discloses a kind of bolt processing clamping manipulator, including base mounting frame, support column, control device, rotary drive mechanism, joint arm, hydraulic telescopic arm and clamping assembly, support column is fixed in base mounting frame, joint arm is connected drive motor, hydraulic telescopic arm is equipped with hydraulic control system, rotary drive mechanism contains permanent magnet synchronous motor, drive shaft body etc., ensure rotation accurate stability, joint arm is connected with hydraulic telescopic arm by rotating joint, rotate flexibly, hydraulic telescopic arm is multistage nested type, telescopic range is big, clamping assembly has jaw holder, multi-joint hydraulic clamping jaw etc., jaw head is with antiskid toothed plate and pressure sensor, bolt can be stably clamped, control device realizes accurate control and data interaction.The utility model can efficiently and stably complete bolt processing clamping handling, improve automation level and production efficiency.

The utility model provides a kind of floating body type double channel field fish environment monitoring device, including sensor unit, data remote telemetry unit, power supply unit and floating body structure unit, sensor unit is used to collect water data, including a variety of sensors, data remote telemetry unit is used to collect sensor data and transmission, power supply unit is used to power supply device, including solar panel and lithium battery;Floating body structure unit includes the water part and the water entry part of upper and lower settings.This application realizes in complex, remote, unattended field water environment, to fish ecological environment key parameter carries out long-term, stable, continuous, automatic remote monitoring, provides extremely powerful and reliable technical means for fish resources protection, aquaculture management, water ecological research, solves the key bottleneck problem of communication, power supply, platform stability in long-term stable monitoring in field.

This invention discloses a method, apparatus, and storage medium for cataloging and locating space targets based on optical observation. The method includes the following steps: acquiring multiple observation images taken consecutively at the same location; performing noise reduction filtering on the acquired observation images and uniformly outputting corresponding astronomical images; extracting all identifiable stars from the astronomical images and outputting a list of stars; identifying stars in the astronomical images and outputting a list of stars using the movement characteristics of stars and satellites; filtering out all identified stars and identifying space targets from the remaining stars; locating the named stars using a lookup table method; and locating the identified space targets using a triangulation geometric algorithm. This invention can use mainstream astronomical telescopes on the market to achieve target discovery and location. Compared with methods that rely on radar systems to identify and locate space targets, this invention is easier to implement and can effectively reduce cataloging costs.

The invention relates to a safety early warning method and device for a railway tunnel, computer equipment, a computer readable storage medium and a computer program product. The method comprises the following steps: the beeper sends a position signal of a target person to the wired signal box; in response to the position signal, the wired signal box sends the position information of the target person to a central control board through the base station; the central control board determines the positioning information of the target personnel based on the position information, sends the positioning information to a target train, determines the prediction time when the target train arrives at the target personnel, and sends an early warning signal to the base station according to the prediction time; and in response to the early warning signal, the base station sends early warning information to the pager through the wired signal box, so that the pager carries out early warning prompt on the target person. By adopting the method, the safety can be improved.

The invention belongs to the technical field of safety protection, and discloses a real-time safety protection method based on a sensor array, and the method is based on a multi-mode sensor array system combining digital twin modeling, scene adaptation design and precise deployment, and is matched with a scene adaptive mixed domain attention reinforcement learning fusion model. High-dimensional features of sensing data are extracted from four dimensions of space, time, frequency and semantics, data weights of all sensors are dynamically distributed according to scene characteristics, and environmental interference is inhibited while threat associated features are enhanced; a scene-divided threat sample data set is constructed by combining a virtual threat sample generated by a digital twinning model with a real sample, and a lightweight model is optimized through transfer learning, model pruning and incremental training, so that the threat identification accuracy in a complex scene is improved, the adaptability of a system to novel threats is enhanced, and the threat identification efficiency is improved. The identification deviation caused by scene difference is reduced, and the accurate monitoring requirements of different protection scenes are met.

This application discloses a damage detection method, electronic device, storage medium, and computer program product, relating to the field of nondestructive testing technology. The method includes: scanning a composite material to be tested using a terahertz time-domain spectroscopy system to obtain the original time-domain echo signal of the composite material, wherein the composite material to be tested is a glass fiber reinforced polymer composite material; preprocessing the original time-domain echo signal to obtain target image data; extracting features from the target image data to obtain damage features; predicting the geometric offset of the damage features; and locating the damage target in the composite material to be tested based on the geometric offset and the damage features. This application solves the technical problem of poor detection effect in existing nondestructive testing technologies.

This utility model relates to the field of iron core tilt detection technology, and in particular to a device for detecting the tilt of a skewed rotor iron core. The technical solution includes: a base frame, a support mechanism mounted on the base frame, a drive mechanism mounted on the support mechanism, and an iron core to be inspected mounted on the drive mechanism; a first detection mechanism and a second detection mechanism are respectively mounted on both sides of the support mechanism. The first detection mechanism is used to detect the included angle of the aluminum groove edge of the iron core to be inspected, and the second detection mechanism is used to detect the cleanliness of the aluminum groove of the iron core to be inspected. This utility model effectively solves the problems of traditional detection methods, such as the inability to detect 100% accuracy, the inability to detect embedded scrap and abnormal tilting in the middle section of the iron core, low production efficiency, and high labor intensity.

The utility model relates to the technical field of backstops, and discloses a permanent magnet roller self-locking backstop which comprises a backstop body, a mounting hole is formed in the center of the interior of the backstop body, flat key grooves are formed in the upper side and the lower side of the mounting hole respectively, a roller shaft is slidably connected into the mounting hole, and the roller shaft is connected with a permanent magnet roller. The upper side and the lower side of the roller shaft are fixedly connected with protruding blocks, the two sides of the inner walls of the flat key grooves in the two sides are each provided with a plurality of limiting assemblies, each limiting assembly comprises a limiting ball, the side wall of each limiting ball is fixedly connected with a limiting plate, and the side edge of each limiting plate is provided with a limiting spring. According to the utility model, the roller shaft is inserted into the mounting hole, the convex block is ensured to be aligned with the flat key groove, and after the roller shaft is completely inserted, the limiting spring reversely pushes the limiting ball to be clamped into the limiting groove, so that the roller shaft is limited, the effect of automatically locking the position of the roller shaft is achieved, and the problem of easy axial movement or positioning deviation in the traditional roller shaft mounting process is solved; and the installation efficiency and the connection stability are improved.

This invention discloses a multimodal anomaly inference method for distribution network equipment based on the Mamba model, belonging to the field of multimodal anomaly detection technology. The key points of the scheme are: multimodal feature extraction from images and text; construction of a text-conditional Kalman projector, using a prediction-update mechanism combined with text prior gating to dynamically denoise visual features; building a pre-trained Mamba model and freezing the backbone parameters, achieving lightweight fine-tuning through custom plugins; concatenating multimodal features and feeding them into Mamba for autoregressive inference, extracting hidden states when decoding dedicated positioning markers, and outputting strings representing coordinates via bounding box regression heads; and employing a two-layer learning rate and cosine decay optimization. This invention combines the linear inference efficiency advantage of Mamba with text-guided visual filtering, effectively suppressing background noise pollution of the Mamba state space, and achieving efficient classification, accurate positioning, and inference of distribution network equipment anomalies.

The utility model discloses a multi-directional positioning railway track gantry hydraulic supporting device which comprises a gantry body, hoisting equipment is installed at the upper end of the gantry body, a walking mechanism is arranged at the lower end of the gantry body, installation plates are fixed to the two sides of the gantry body, bearing blocks are fixed to one sides of the installation plates, and the bearing blocks are fixed to the other sides of the installation plates. A sliding groove is formed in one side of the bearing block, a forward and reverse screw rod is jointly and rotationally connected between the opposite side walls in the sliding groove, and two sliding blocks are installed in the sliding groove. The portal frame can be accurately positioned to a position where the portal frame needs to work, improves the construction efficiency and the installation precision of track parts, meanwhile can enable supporting force to be evenly distributed, avoids the problem that a traditional portal frame is uneven in supporting force on a curve track section, further enhances the supporting stability, and reduces the construction cost. Risks of shaking and inclining of the portal frame in the working process are reduced, and life safety of constructors and normal operation of construction equipment are guaranteed.

The utility model provides a panel surface heat resistance detection device, belongs to the technical field of automobile central control panel detection, and aims to solve the problems that the adjusting mode of the dial indicator position of an existing tester needs to repeatedly tighten and tighten a fastening bolt for fixing the dial indicator, the process is tedious, and the manual control of the fastening force of the bolt is easy to cause deviation. Comprising a tester main body, a sliding block, a plugging column, a rapid adjusting assembly and a separation maintenance assembly. The three sliding blocks are connected to the interiors of the three fixed stand columns in a sliding mode correspondingly. The three inserting columns are inserted into the front ends of the three sliding blocks correspondingly. The three groups of quick adjusting assemblies are respectively arranged in the three fixed upright posts; and the three groups of separating and overhauling assemblies are respectively arranged above the three inserting columns. According to the utility model, the digital display meter can be quickly adjusted to a required position, the problems of complexity and positioning deviation of manual adjustment are solved, the disassembly, assembly and maintenance efficiency of the dial indicator is remarkably improved through the separation and maintenance assembly, and detection data distortion caused by equipment adjustment errors is effectively reduced.

The application discloses an intravascular ultrasound image segmentation method based on fuzzy fusion and contrast distillation, and comprises the following steps: preprocessing an acquired intravascular ultrasound image, performing multi-scale feature extraction on the preprocessed image, and generating a pixel-level uncertainty map corresponding to each scale feature; constructing a fusion uncertainty map generation module to realize adaptive weighted fusion of cross-scale pixel-level uncertainty maps and obtain a fused uncertainty map; constructing a fused feature generation module to generate fused features and fused disturbance features; constructing a multi-objective optimization module to obtain an optimal segmentation model; and segmenting the collected intravascular ultrasound image to obtain a final segmentation result. The application can improve the segmentation accuracy and robustness of the intravascular ultrasound image, solve the problems of inaccurate boundary positioning and incorrect area judgment caused by noise interference, structure blur and low-efficiency multi-scale feature fusion in the intravascular ultrasound image segmentation, and realize high-precision analysis of complex pathological characteristics.