214 results about "Coordinate difference" patented technology

The invention relates to the technical field of image state recognition, in particular to an unhooking and rehooking AI accurate recognition grabbing system and method based on pattern recognition, in the system, node construction is conducted through contour changes, boundary difference values and gray level dynamic states of a hook assembly in an image sequence frame, edge displacement accumulation analysis is combined, meanwhile, through a graph neural network, an image sequence frame is obtained, and the image sequence frame is obtained. Cosine values and coordinate difference values between nodes are subjected to combined comparison, and a path hopping sequence is constructed, so that the response sensitivity to state abrupt change is enhanced, a key path of morphological evolution can still be stably extracted under the condition of complex background interference or local shielding, the anti-interference performance and fault tolerance of space path identification are effectively improved, and the space path identification accuracy is improved. Statistical modeling is further carried out on state rate sudden change points through a hidden Markov model, paragraph merging and invalid fragment removing operation are carried out on abnormal point segments by matching a standard state mode, a state label sequence is constructed, and accurate division of high-confidence and multi-segment continuous states is achieved.

The invention discloses a GNSS (Global Navigation Satellite System) / INS (Inertial Navigation Satellite System) anti-deception navigation method based on ESKF (Enhanced Shift Keying Function) and dynamic window detection, which comprises the following steps of: 1, acquiring observation data of an INS and a GNSS, performing time synchronization and coordinate system unification on the observation data of the INS and the GNSS, and respectively eliminating a time error and a coordinate difference; 2, a dynamic window detection mechanism is designed, an error state dynamical model is constructed, the dynamic window detection mechanism is used for detecting whether GNSS cheating exists or not, and meanwhile the error state dynamical model can output stable navigation parameters; and 3, outputting a final high-precision navigation result according to the stable navigation parameters. According to the method, the problems of error accumulation and covariance distortion of nonlinear system model linearization are effectively solved, pollution of GNSS deception signals can be effectively inhibited, high-precision positioning and anti-deception capability are kept in complex environments (such as urban shielding and electromagnetic interference), and a navigation system with high precision and high reliability is formed.

The invention discloses a contact type three-coordinate measuring machine and a measuring method, and belongs to the technical field of three-coordinate measuring machines. The measuring method comprises the following steps: S1, acquiring coordinate values of standard shapes of all detection paths of a standard workpiece in different temperature ranges and coordinate values of corresponding contact type measuring heads through a three-coordinate measuring machine; s2, obtaining coordinate values of a to-be-detected shape of a preset detection path of the to-be-detected workpiece through a three-coordinate measuring machine; s3, acquiring a current temperature and a current detection path, and comparing the current detection path with the preset detection path; s4, presetting a coordinate deviation threshold range between the coordinate value of the standard shape and the coordinate value of the to-be-measured shape; s5, whether the coordinate difference between the coordinate value of the to-be-detected shape of the current detection path and the coordinate value of the corresponding standard shape is within the coordinate deviation threshold range or not is judged; according to the measuring method of the contact type three-coordinate measuring machine, the workpiece can be measured more accurately.

This disclosure describes systems, methods, and devices related to object detection in images. A device may input an image, representing an object, to a manual labeling learner system; identify, using the system, first coordinates of an upper left corner of a bounding box representing the object based on a heatmap indicative of a probability of the first coordinates representing the upper left corner; identify, using the system, second coordinates of a bottom right corner of the bounding box based on the first coordinates and a first distance regression map indicative of coordinate differences between the second coordinates and ground truth coordinates input to the machine learning model as training data; generate, using the system, adjustments to the first coordinates and the second coordinates based on a second regression map; and generate, using the system, the adjusted first and second coordinates, the bounding box.

The invention provides a distributed multi-target microwave transmission control method and system, and relates to the technical field of data processing, and the method comprises the steps: obtaining the geographic coordinate information of each target device, constructing the motion track of each target device, sorting the coordinate information of each target device according to a time node sequence, and storing the sorted coordinate information of each target device; coordinate difference values between adjacent time nodes are accumulated to obtain a predicted coordinate data set; coordinate distance values between target devices are extracted; the coordinate distance values are subjected to grouping statistics to obtain a position merging data set; beam direction range information and beam angle interval parameters of a current time period are extracted; and obtaining a beam candidate data set, performing coordinate comparison on each target device to obtain beam binding data, generating a control instruction, and ensuring that the beam adjusts the beam emission direction according to the control instruction. The coordinate of the target equipment can be predicted, the beam emission direction can be controlled, and microwaves can be stably transmitted to the target equipment.

The invention, which belongs to the technical field of artificial intelligence, discloses a multi-modal cross-view-angle air-ground cooperation big model system comprising a heterogeneous agent module, a multi-modal data processing module, a cross-view-angle fusion module, a big language model reasoning module and an action decision module. The heterogeneous intelligent agent module comprises an unmanned aerial vehicle, a ground robot and an unmanned vehicle, the unmanned aerial vehicle is provided with a high-resolution camera, an infrared imaging device, a laser radar and a meteorological sensor, the ground robot is provided with the laser radar, a camera and a force sensor, and the unmanned vehicle is provided with a camera, a radar and an inertial measurement unit. The multi-view features are fused through a cross attention mechanism, precise mapping from the features to real coordinates is realized in combination with a ScaleToRealClip model, and scale and coordinate differences of air and ground data are eliminated; meanwhile, the cross-view fusion module can dynamically update environment coordinates, help an intelligent agent to perceive temporary obstacles in real time, and construct a comprehensive and accurate environment cognition model.

The invention relates to the technical field of resolution improvement, in particular to a video definition improvement method and system.According to the video definition improvement method and system, through window accumulation of adjacent frame gray difference and coordinate difference, minimum displacement is selected, mapping is updated, boundary neighborhood weighted correction is carried out, cross-frame dislocation is reduced, and motion area connection is guaranteed; capturing a three-frame gray scale sequence to calculate a difference value, writing a replacement value according to a monotone and amplitude condition, inputting the replacement value and an alignment frame difference into Fourier transform, decomposing amplitude and phase distribution in a residual field in a frequency domain, and performing amplitude and phase consistency elimination on an abnormal block during frame-by-frame superposition to obtain an abnormal block; row difference and column difference absolute values of neighborhood blocks in the residual iteration texture matrix are accumulated into energy, the energy and a center difference are combined with a threshold value to output a mark, and a texture region and an edge region are distinguished; the brightness of the low-frequency position is updated according to the weight mean value, the maximum difference and the direction difference are combined for the high-frequency position, the edge gradient is extracted through the convolutional neural network, a correction value is generated, edge steps and fuzziness are suppressed, and high-frequency details are continuously kept.

The invention discloses a real-time thickness measuring method and a thickness measuring device, and relates to the technical field of computer vision. The real-time thickness measurement method comprises the steps that a first image set is obtained, and the first image set is shot image frames of a first measurement plane, and the shot image frames are continuously shot by a camera with a calibration plane as the reference along an initial point; recording the shooting coordinates of each frame of image in the first image set, and calculating the definition of each frame of image in the first image set frame by frame; determining a first clear image, and recording a shooting coordinate of the first clear image as a first shooting coordinate; and determining a second clear image, and recording the shooting coordinate of the second clear image as a second shooting coordinate. And respectively calculating the coordinate difference between the first shooting coordinate and the initial point and the coordinate difference between the second shooting coordinate and the initial point to obtain the thickness value of the to-be-measured object. By adopting the technology provided by the invention, the non-contact high-precision measurement of the thickness of the object can be realized at low cost.

To provide a positioning device capable of improving positioning accuracy, a positioning method, a program, and a storage device.SOLUTION: A positioning device 20 includes a first position estimation unit 31, a second position estimation unit 33, and an information integration unit 36. The first position estimation unit 31 and the second position estimation unit 33 output first information and second information by positioning in different coordinate systems (a world geodetic system and a map coordinate system), respectively. The information integration unit 36 performs predetermined optimization processing relative to a combination of the first information and the second information recognized as the same time. The information integration unit 36 acquires difference (inter-coordinate difference) regarding information such as a position and a posture of a moving object between mutually different coordinate systems of the first information and the second information by predetermined optimization processing.SELECTED DRAWING: Figure 1

The invention discloses a target measurement method and device, electronic equipment and a storage medium, and relates to the technical field of holders. The method comprises the following steps: acquiring a shot image, wherein the shot image comprises a shot target; determining a device parameter and a pixel coordinate difference value, and calculating a correction factor by using the device parameter and the pixel coordinate difference value; and determining a target rotation angle by using the correction factor, and controlling the holder to adjust based on the target rotation angle. Compared with the prior art, the method has the advantages that the correction factor is calculated only through the basic parameters of the image and the equipment under the condition that modes such as laser ranging do not need to be introduced, the measurement precision is improved compared with a mode of trigonometric function conversion, errors caused by lens distortion, projection deformation or environmental factors are reduced, and meanwhile, the hardware cost is also reduced.

The invention provides an intelligent ward nursing information acquisition and processing method and system. The method comprises the following steps: acquiring mattress pressure data to construct a pressure grey-scale map sequence, calculating a pressure barycentric coordinate difference value between continuous frames to generate a patient movement vector set and a body barycentric movement path, extracting a continuous pixel region exceeding a pressure threshold in a single-frame image as a high-pressure region, and counting the duration of the high-pressure region; calculating gradient distribution by using the pressure data, decomposing the gradient distribution into direction and intensity data, and generating a pressure topological graph through dual-channel fusion; integrating the bending characteristics of the moving path, the time distribution characteristics of the duration of the high-pressure area and the spatial gradient characteristics of the pressure topological graph, and constructing a body position dynamic space-time diagram with the central point of the high-pressure area as a node and the moving distance and angle of the node as edges; when it is recognized that the high-risk static body position features exist in the image, a turning-over reminding process is triggered, and a control instruction is generated. According to the invention, accurate recognition and automatic nursing response of high-risk body positions are realized.

The invention provides a point cloud semantic segmentation method and system based on multi-information fusion coding, and belongs to the technical field of computer vision, and the method comprises the steps: inputting preprocessed point cloud data into a pre-constructed feature encoder, and the pre-constructed feature encoder comprises a first stage, a second stage and a third stage, the second stage, the third stage, the fourth stage, the fifth stage and the fifth stage respectively comprise an SA layer and an LIA layer, and the LIA layer is used for carrying out multi-information fusion coding on the aggregated point cloud data; calculating local center point vector information, a local center point three-dimensional coordinate, a coordinate difference between the local center point and a neighborhood point, a neighborhood point feature, and a local center point and neighborhood point feature, and carrying out fusion coding; and decoding and segmenting the output of the pre-constructed feature encoder to obtain a segmentation result. According to the method, the defect of a single feature is overcome, the segmentation precision is improved, and the real-time requirement is met on the premise that hardware equipment is not required to be added.

The present application provides an automatic focusing method for a fluorescence imaging device, including: S1, obtain a relationship function curve Z-Fv; S2, Gaussian fitting is performed on the relationship function curve Z-Fv, to obtain a standard Gaussian function f(x); S3, the object lens is arranged at three different travel coordinates, to obtain three groups of coordinate-clarity data (Z, Fv); S4, any two groups of the data are taken from the three groups of the coordinate-clarity data (Z, Fv) obtained in S3 and perform the Gaussian fitting with a standard deviation σ, to obtain a Gaussian function f1(x); and S5, an average value μ1 of the Gaussian function f1(x) is calculated, as a preferred travel coordinate Z′ for this time of the focusing object lens. The present application solves problems of large focusing clarity error and slow focusing speed of the fluorescence imaging device in existing technologies.

The invention provides a positioning method and device and a storage medium, and the method comprises the steps: a first cloud server obtains environment information uploaded by a terminal, carries out the environment reconstruction through the environment information, and obtains a map under a global coordinate system, the global coordinate system is a position coordinate system of map data after environment reconstruction; in response to starting of the terminal, the first cloud server obtains a first coordinate of the terminal in a local coordinate system every preset time interval, and determines a target coordinate difference value corresponding to the first coordinate in a global coordinate system; and the terminal receives the target coordinate difference value sent by the first cloud server, and corrects the first coordinate based on the target coordinate difference value to obtain a second coordinate positioned by the terminal. According to the method and the device, deviation accumulation or positioning loss of the positioning position is avoided, the positioning precision is improved, the situations of mode crossing and drifting in the virtual picture are further avoided, a user does not need to position through a computer backpack, and the user experience is improved.

The invention relates to the technical field of laser ranging, in particular to a novel railway tunnel deformation real-time surveying and mapping system based on a three-dimensional scanning technology, which comprises a laser sampling module, an echo evaluation module, a space repositioning module, a residual evolution module and a deformation surveying and mapping module. According to the method, after continuous distance information of a target position is obtained, measuring point sequence data is constructed based on a time sequence, a signal abnormal area is recognized through echo intensity fluctuation, the signal abnormal area is further converted into a space coordinate structure, track differences are extracted, vector data are generated, deviation characteristics are analyzed in combination with the tunnel space position, and the deformation amplitude is quantified. Compared with a periodic observation mode depending on manual distribution of measuring points, the method has the advantages that fine coordinate difference data can be continuously output, the deformation monitoring frequency and the space coverage range can be improved, manual interference errors can be reduced, and the dynamic mastering capability of the tunnel structure evolution process can be enhanced.

The invention relates to the field of component generation, and discloses a grid component generation method and device with a visual interface and a storage medium. The method comprises the steps that gridding processing is conducted on an interface according to preset grid parameters, and the interface with N grid units is obtained; reading coordinate data of the dragging component; traversing and selecting a target grid unit from the N grid units; calculating a horizontal and longitudinal difference value between the coordinate data and the target grid unit to obtain a horizontal coordinate difference value and a longitudinal coordinate difference value; judging whether components of occupied grid units exist in the target grid units or not; if not, performing distance detection processing on the target grid unit to generate an effective grid unit; after the N grid units are traversed, size constraint alignment processing is carried out on the dragged component according to the effective grid units, and a visual component adsorbing the grid units is generated. In the embodiment of the invention, the difference between actual effects of Web preview and applets is reduced, and the efficiency and precision of layout design are improved.

The invention relates to the technical field of composite machine tools, in particular to a multi-axis composite machining machine tool and an intelligent control method thereof.The method comprises the steps that the machining machine tool is subjected to pilot run, and a tool spindle coordinate system and an inner circle grinding head spindle coordinate system are established; determining the error level of the machine tool positioning precision according to the coordinate difference of the origin of the coordinate system; determining an adjusting amplitude and an adjusting object according to the vector coordinates, and adjusting the fast moving speed and the feeding speed of the moving module; acquiring a plurality of coordinate differences, judging the positioning grade of the machine tool, and calling back the judgment standard for judging the positioning precision error grade of the machine tool; cutting force borne by a workpiece spindle is detected, the abrasion degree of a tool is judged, and corresponding measures are taken; the change trend of the cutting force is analyzed, the consumption grade of abrasion consumption of the tool is judged according to the change trend of the cutting force, and the ratio of the spindle rotating speed to the feeding speed is adjusted. Multiple procedures are completed through one-time clamping, machining parameters of the machine tool are adjusted, and the positioning precision and machining efficiency of the machine tool are remarkably improved.

The invention relates to the technical field of limb rehabilitation training, in particular to an artificial intelligence-based cognitive impairment assessment and intervention training system, which comprises a path construction module, an action acquisition module, a matching recognition module, a rhythm regulation and control module and a cognitive intervention module. According to the method, the standard path is constructed in the virtual task and combined with the three-dimensional action data of the user to form the path section response result, so that the spatial integrating degree of the action of the user and the target path can be identified, and quantitative description of the action execution stability and continuity is realized through coordinate difference analysis; a cooperative path record of task execution is obtained in combination with a trajectory coverage rate and a matching relationship, and a trend analysis and parameter regulation mechanism is introduced, so that a task pushing rhythm and a user response rhythm are dynamically matched, and tracking correction of an intervention effect in a training stage and fine evaluation of cognitive limb disorder performance are realized. The analysis granularity of cognitive state and action collaboration is improved, and the mapping precision between the intervention content and the user execution characteristics is enhanced.

The invention relates to a stability analysis method for control points in a tunnel in the technical field of tunnel measurement, which comprises the following steps of: S1, establishing a wire control network and a leveling network in the tunnel, and remeasuring the control points; s2, selecting a traverse point as an analysis point, comparing the two groups of horizontal angle data obtained after retest, and judging whether the two groups of horizontal angle data meet the standard or not; s3, establishing a deviation limiting system of two adjacent control points, wherein the deviation limiting system comprises a side length deviation standard, an elevation deviation standard and a difference relative precision standard of coordinate differences; s4, taking two control points adjacent to the analysis point as two target points, comparing the deviation limiting systems of the three control points pairwise, and integrating the judgment condition that the horizontal angle is relatively poor to obtain the stability condition of the analysis point; according to the method, comprehensive analysis can be carried out from multiple aspects, and the accuracy of stability judgment is improved.

The invention provides a table topological relation establishing method and device and a computer program product, and relates to the technical field of electric digital data processing. Vertex coordinates of cells are recognized, matching point positions are selected to calculate a coordinate difference value, and the coordinate difference value is compared with a length error threshold value and a width error threshold value; according to the method, the association relationship between the cells in the row direction and the column direction is identified, the coordinate values of the specified mark point positions are sorted in the row or the column, and row and column division and spatial relationship reconstruction are directly performed based on geometric rules, so that the problem of performance reduction caused by overlarge image size is reduced, and the method does not depend on the integrity or definition of a table frame; the method can be suitable for processing a frameless table and a complex table scene containing a large number of empty rows and empty columns, and the processing speed is remarkably improved.

The invention discloses a method for generating a match condition top view based on multi-camera information fusion, which comprises the following steps of: deploying cameras which can cover the whole stadium and form an overlapped region at four corners of the stadium, separating and extracting straight lines and intersection points in an image by utilizing an HSV color space, and calculating and constructing a rectangular feature point set of the stadium; based on geometric features and a coordinate transformation matrix, mapping a local view angle into a unified overlook coordinate system, screening out an optimally matched rectangular coordinate through a rectangular matching score, and calculating the coordinate difference of the athletes in the overlapping region transformed into the overlook from different view angles; weight distribution is carried out on the rectangular matching error and the coordinate error of the athlete through a CRITIC weight method, the objective function is solved, and an optimal dimension transformation matrix is determined. According to the invention, real-time top view generation of the sports event can be realized.

The invention discloses a target detection method and device based on visual identification. The method comprises the following steps: acquiring relative coordinates between an image acquisition part and a detection surface of a detection object; when the relative coordinate is not consistent with the reference coordinate, determining a first coordinate difference value; when the relative coordinate is not within the coordinate error range, adjusting the image acquisition component until the relative coordinate is consistent with the reference coordinate; when the relative coordinates are consistent with the reference coordinates, controlling an image acquisition component to perform image acquisition on the detection object to obtain a plurality of images; determining second relative coordinate differences between the plurality of images and the reference image, and selecting a target image; and performing target detection according to the target image. The technical problem that in the prior art, a machine vision system adopts depth-of-field limitation of a fixed camera, so that all targets cannot be ensured to be within the depth-of-field range of the camera when the heights of different models of recognized targets are changed, and recognition cannot be possibly caused is solved.

The invention provides a network RTK base station coordinate precision detection method and system, and the method comprises the steps: carrying out the base station networking based on a Delaunay triangulation algorithm, and determining a baseline list; obtaining satellite precision orbit information, selecting a to-be-solved baseline vector from the baseline list, and resolving the to-be-solved baseline vector by adopting a fast static positioning algorithm to obtain a preset high-precision relative coordinate result; the preset high-precision relative coordinate result is converted into coordinate values of all reference stations, the coordinate values of all the reference stations are compared with preset known coordinate values in a system, coordinate difference values of all the reference stations on all the base lines are obtained through calculation, gross error elimination is conducted based on the coordinate difference values, and processed coordinate difference values are obtained; and integrating the processed coordinate difference values of each reference station on the plurality of base lines, performing consistency analysis and statistical judgment, determining the reference station with abnormal coordinates in combination with a preset threshold value, and outputting a diagnosis alarm result. According to the invention, the reliability and precision of the network RTK service are effectively guaranteed.

The invention discloses an aircraft structure finite element calculation unit connection method and device, equipment and a medium, and relates to the field of structure finite element analysis, and the method comprises the steps: determining a slave node and a master node based on the interface of different calculation units of a finite element model, and determining the node freedom degrees of the master node and the slave node; determining a coordinate difference value of each master node and each slave node, and respectively determining each first transformation relation matrix of the force and torque transmitted from each master node to the force and torque of the slave node according to each coordinate difference value; determining an intermediate variable matrix based on the node degree of freedom and the first transformation relation matrix, and determining each second transformation relation matrix of the force and torque transmitted from the slave node to each master node through the node degree of freedom, the first transformation relation matrix and the intermediate variable matrix; and determining a displacement relationship between the master node and the slave node according to the second transformation relationship matrix, and determining a calculation unit connection mode by using the displacement relationship. Connection between different calculation units can be realized.

The invention relates to the technical field of document processing, in particular to a multimedia document online editing method and system.The method comprises the following steps that image-text coordinates and difference values are obtained to generate a path sequence, labels and attributes are matched to construct a structure chart, paths are compared to generate offset identification, editing information is collected to extract a trigger sequence, and synchronous deviation is calculated to generate a task index. And reconstructing the sequence to generate a rearrangement result. According to the method, the logic structure is constructed through the image-text coordinate difference and the offset path, the combination state change is accurately judged in combination with the angle and distance change, the time sequence and label characteristics in the editing behavior are collected, the operation link is established, the time sequence consistency of content recombination is improved, and the task synchronization rhythm is calculated through the delay and response difference. According to the method, the rhythm control of content loading is realized, the combination sequence is reconstructed according to the logic distance, the continuity and accurate positioning after structure rearrangement are kept, and the stability and response efficiency of a multimedia document in a complex editing scene are enhanced.

The invention discloses an unmanned aerial vehicle positioning method, device and system during signal interruption, and a storage medium. The method comprises the following steps: receiving a broadcast ephemeris broadcasted by a satellite, observation value data and a real-time orbital clock error correction number; judging whether the real-time orbital clock error correction is interrupted or not; if the real-time orbital clock error correction is interrupted, performing short-term forecasting on the correction by establishing a quartic polynomial; combining the correction number with the broadcast ephemeris to generate a real-time precise ephemeris and a real-time track; if the predicted correction number is used, an observation equation is established through epoch difference to settle a coordinate difference; if the received correction number is used, an observation equation is established through a PPP model, and the real-time coordinates of the unmanned aerial vehicle are settled. By adopting the technical scheme of the invention, high-precision positioning of the unmanned aerial vehicle during short-term signal interruption is realized.

The invention discloses an obstacle distance measurement method and system based on panoramic tiled display. Four-way or six-way fisheye cameras are installed on a vehicle, images around a vehicle body are collected, 3D rendering splicing is carried out based on calibration data (internal parameters / external parameters), and a 360-degree panoramic aerial view image is generated. A user clicks two points of a screen through a display terminal, a system converts screen coordinates into 3D normalized vertex coordinates through viewport inverse transformation, relative coordinate differences are calculated and then expanded into homogeneous coordinates, the homogeneous coordinates are converted into physical coordinate differences in combination with a pre-stored rendering transformation matrix inverse matrix (generated by model / projection / view transformation combination), and the 3D normalized vertex coordinates are converted into 3D normalized vertex coordinates. And finally calculating the actual physical distance between the two points. The system supports dynamic refreshing of a ranging result and renders a distance marking line on a panoramic image. According to the method, through multi-camera calibration, coordinate space mapping and a matrix transformation algorithm, interactive obstacle distance measurement in a vehicle-mounted panoramic display scene is realized, and the method is suitable for intelligent driving, vehicle auxiliary driving and other scenes.

The application provides a kind of based on remote sensing image's directed lane center line generation method and system, method includes: the remote sensing image data is input into trained directed lane center line segmentation model, obtains the position prediction image and direction prediction image of lane center line;Line segment detection method is used to obtain line segment information from the position prediction image, the actual geographic coordinates corresponding to line segment are obtained using the coordinate information of remote sensing image data;The RGB data in the same place of the direction prediction image with the coordinate position of the line segment is extracted, and the coordinate difference (dx, dy) between the end and the beginning of the line segment is calculated according to the RGB data;According to the coordinate difference (dx, dy), the direction information of the current line segment is calculated;Iterate position prediction image and direction prediction image, according to the coordinate information and direction information of each line segment, adjust the head-tail connection relationship of each line segment, and obtain directed lane center line data.The application reduces the production cost of lane center line data.

This disclosure describes systems, methods, and devices relating to object recognition in images. A device can input an image representing an object into a manual labeling learning system; using the system, identify the first coordinates of an upper-left corner of a bounding box representing the object, based on a heatmap indicating the probability that the first coordinates represent the upper-left corner; use the system to identify the second coordinates of a lower-right corner of the bounding box, based on the first coordinates and a first distance regression map indicating the coordinate differences between the second coordinates and ground-truth coordinates, which are input into the machine learning model as training data.Using the system, generate adjustments to the first and second coordinates based on a second distance regression map; and using the system, generate the bounding box with the adjusted first and second coordinates.

The application provides a zero position calibration method and device of a camera, electronic equipment and a storage medium, and relates to the technical field of camera shooting. The camera comprises a main camera and at least one auxiliary camera. The method comprises the following steps: when the main camera is at a first target zero position, controlling the main camera to shoot a first image; determining a first target position in the first image; controlling the auxiliary camera to rotate to a second target zero position based on the first target position and a preset position difference value corresponding to the auxiliary camera; the preset position difference value is a coordinate difference value between the first target position and a second target position; the second target position is a target position in a second image shot by the auxiliary camera and located in the first image. The application realizes the calibration of the second target zero position of other auxiliary cameras based on the first target zero position of the main camera. The calibration of the zero position of each auxiliary camera does not require the participation of an optical interrupter, thereby saving cost and reducing the probability of hardware failure.