A method and apparatus for generating a luggage trajectory

Abstract

The application provides a luggage trajectory generation method and device. The method comprises the following steps: obtaining target image data of luggage; the target image data comprises original image data of the luggage and label data of the luggage; performing feature extraction according to the target image data to obtain feature map data; detecting a luggage position according to the feature map data to obtain luggage position data; determining initial luggage trajectory data according to the luggage position data; and correcting the initial luggage trajectory data to obtain target luggage trajectory. The application not only improves the accuracy of the luggage trajectory, but also helps airport staff to find the luggage according to the luggage trajectory, thereby saving labor cost and time.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of luggage trajectory, and further relates to a luggage trajectory generation method and device. BACKGROUND

[0002] Before an airplane takes off, a passenger cannot continue to board the airplane due to personal reasons, and needs to take away the luggage placed in the belly compartment of the airplane. However, since the trajectory of the luggage is not tracked and positioned, the airport staff needs to check the information of each luggage in the belly compartment of the airplane, which consumes a lot of time and easily leads to flight delay. In addition, due to the similar appearance of the luggage and the blurred luggage label, the luggage may be wrongly found or missed, which makes it difficult to guarantee the accuracy of manual luggage searching, and brings inconvenience and loss to the passengers and the airline company. SUMMARY

[0003] The present application aims to solve the technical problem of providing a luggage trajectory generation method to improve the accuracy of the luggage trajectory and save the cost of searching for the luggage position.

[0004] To solve the above technical problems, the technical scheme of the present application is as follows:

[0005] In a first aspect, the present application provides a luggage trajectory generation method, comprising:

[0006] obtaining target image data of the luggage; the target image data comprises original image data of the luggage and label data of the luggage;

[0007] performing feature extraction according to the target image data to obtain feature map data;

[0008] detecting the position of the luggage according to the feature map data to obtain luggage position data;

[0009] determining initial luggage trajectory data according to the luggage position data;

[0010] correcting the initial luggage trajectory data to obtain target luggage trajectory.

[0011] Optionally, the target image data of the luggage is obtained, comprising:

[0012] obtaining the original image data of the luggage through an image acquisition device;

[0013] obtaining the label data of the luggage by scanning the luggage shipping label; the label data of the luggage comprises luggage serial number data, flight data, destination data and passenger data;

[0014] determining the target image data of the luggage according to the original image data of the luggage and the label data of the luggage.

[0015] Optionally, feature extraction is performed according to the target image data to obtain feature map data, including:

[0016] The target image data is preprocessed to obtain input data;

[0017] The input data and 、 are used to obtain feature map data; the feature map data includes the height of the feature map, the width of the feature map, and the value of the feature map;

[0018] wherein C(i,j) represents the pixel value of the feature map at position (i,j), I represents the input data, K represents the convolution kernel, (i,j) represents the position of the feature map, m and n represent the dimensions of the convolution kernel, I(i+m,j+n) represents the pixel value of the input data at position (i+m,j+n), K(m,n) represents the weight value of the convolution kernel at position (m,n), N represents the width of the feature map, W represents the height of the feature map, F represents the width or height of the convolution kernel, P represents the size of the padding value, and S represents the size of the step.

[0019] Optionally, the luggage position is detected according to the feature map data to obtain luggage position data, including:

[0020] The feature map data is divided into a plurality of grid units;

[0021] A plurality of anchor points are determined according to the plurality of grid units;

[0022] The luggage position data is obtained according to the plurality of anchor points.

[0023] Optionally, initial luggage trajectory data is determined according to the luggage position data, including:

[0024] The initial state of the luggage is determined according to the luggage position data;

[0025] The initial state of the luggage is input into a state prediction model 、 to obtain a luggage state prediction result;

[0026] The luggage position data and the luggage state prediction result are input into a state update model to obtain initial luggage trajectory data;

[0027] wherein the luggage state prediction result at time k=1 is obtained by ;

[0028] wherein the prediction covariance at time k=1 is obtained by ;

[0029] wherein is a preset state transition matrix, is a preset process noise covariance matrix, is a baggage state prediction result at time k, is a baggage state prediction result at time k-1, is a control input matrix, is a control input, is a prediction covariance at time k, is a prediction covariance at time k-1, is is a transpose matrix of is a baggage state prediction result at time k=1, is a preset state transition matrix at k=1, is an initial value state, is a control input matrix at k=1, is a control input at k=1, is a prediction covariance at time k=1, is an initial covariance, is a preset process noise covariance matrix at k=1.

[0030] Optionally, the initial baggage trajectory data is corrected to obtain a target baggage trajectory, including:

[0031] According to the initial baggage trajectory, matching data is determined;

[0032] According to the matching data, the initial baggage trajectory data is corrected to obtain a target baggage trajectory.

[0033] Optionally, the method further includes:

[0034] According to the target baggage trajectory, modeling processing is performed to obtain a baggage storage model;

[0035] According to the received baggage position query request and the baggage storage model, baggage position information is determined.

[0036] In a second aspect of the present application, a baggage trajectory generation device is provided, including:

[0037] An acquisition module is configured to acquire target image data of baggage; the target image data includes original image data of baggage and label data of baggage;

[0038] A processing module is configured to perform feature extraction according to the target image data to obtain feature map data; detect a baggage position according to the feature map data to obtain baggage position data; determine initial baggage trajectory data according to the baggage position data; and correct the initial baggage trajectory data to obtain a target baggage trajectory.

[0039] In a third aspect, the present application provides a computing device, comprising: a processor, a memory storing a computer program, when the computer program is run by the processor, the method according to the first aspect is executed.

[0040] In a fourth aspect, the present application provides a computer readable storage medium storing instructions, when the instructions are run on a computer, the computer executes the method according to the first aspect.

[0041] The above scheme of the present application has at least the following beneficial effects:

[0042] The above scheme of the present application, by acquiring target image data of the luggage and performing feature extraction, obtains feature map data, then detects the luggage position according to the feature map data to obtain luggage position data, further determines initial luggage trajectory data, and finally corrects the initial luggage trajectory data to obtain the target luggage trajectory, not only improves the accuracy of the luggage trajectory, but also is beneficial for the airport staff to find the luggage according to the luggage trajectory, has the advantages of saving labor cost and time. BRIEF DESCRIPTION OF DRAWINGS

[0043] Figure 1 is a flow diagram of the luggage trajectory generation method in the embodiment of the present application;

[0044] Figure 2 is a structural diagram of the luggage trajectory generation device in the embodiment of the present application. DETAILED DESCRIPTION

[0045] Exemplary embodiments of the present application will be described hereinafter with reference to the accompanying drawings. Although exemplary embodiments of the present application are shown in the drawings, it should be understood that the present application can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided so that the present application can be more thoroughly understood and the scope of the present application can be accurately conveyed to those skilled in the art.

[0046] As Figure 1 shown, the embodiment of the present application proposes a luggage trajectory generation method, comprising the following steps:

[0047] Step 101, acquiring target image data of the luggage; the target image data comprises original image data of the luggage and label data of the luggage;

[0048] Step 102, performing feature extraction according to the target image data to obtain feature map data;

[0049] Step 103, detecting the luggage position according to the feature map data to obtain luggage position data;

[0050] Step 104, determining initial luggage trajectory data according to the luggage position data;

[0051] Step 105, correcting the initial luggage trajectory data to obtain target luggage trajectory.

[0052] The luggage trajectory generation method of the embodiment of the present application, by obtaining target image data of the luggage and performing feature extraction, obtaining feature map data, then detecting the luggage position according to the feature map data to obtain luggage position data, and further determining initial luggage trajectory data, and finally correcting the initial luggage trajectory data to obtain the target luggage trajectory, not only improves the accuracy of the luggage trajectory, but also facilitates the airport staff to find the luggage according to the luggage trajectory, has the advantages of saving labor cost and time.

[0053] In an optional embodiment of the present application, the step 101 of obtaining target image data of the luggage comprises:

[0054] Step 1011, obtaining original image data of the luggage through an image acquisition device;

[0055] Specifically, the image acquisition device records video monitoring or the camera acquires the image of the luggage in the process of security check, loading into the aircraft belly compartment and unloading the luggage from the aircraft belly compartment. Here, the original image data of the luggage can include picture data and / or video data of the luggage.

[0056] Step 1012, obtaining label data of the luggage by scanning the luggage shipping label; the label data of the luggage includes luggage number data, flight data, destination data and passenger data;

[0057] Specifically, by using an automatic identification device such as an RFID (Radio Frequency Identification) reader or an embedded control board in combination with multiple cameras, the luggage carrying an RFID chip is scanned, the bar code on the luggage shipping label is scanned and information is identified, and at least the label data including the luggage number data, the flight data, the destination data and the passenger data is obtained.

[0058] Step 1013, determining target image data of the luggage according to the original image data of the luggage and the label data of the luggage.

[0059] Specifically, the original image data of the luggage and the label data of the luggage are bound or associated to obtain the target image data of the luggage, so as to facilitate the subsequent luggage searching according to the passenger information or the original image data to obtain the label data of the luggage, and improve the efficiency of luggage searching and comparison.

[0060] It should be noted that the subsequent steps are all processing of the original image data in the target image data.

[0061] In an optional embodiment of the present application, step 102 comprises:

[0062] Step 1021, pre-processing the target image data to obtain input data;

[0063] Specifically, the original image data in the target image data can be pre-processed in at least one of the following ways: scaling, cropping and normalizing, to reduce noise in the image and adjust the image size to adapt to the network input

[0064] Step 1022, obtaining feature map data according to the input data and 、 ; the feature map data comprises the height of the feature map, the width of the feature map and the value of the feature map;

[0065] Wherein, C(i,j) represents the pixel value of the feature map at position (i,j), I represents the input data, K represents the convolution kernel, (i,j) represents the position of the feature map, m and n represent the dimensions of the convolution kernel, I(i+m,j+n) represents the pixel value of the input data at position (i+m,j+n), K(m,n) represents the weight value of the convolution kernel at position (m,n), N represents the width of the feature map, W represents the height of the feature map, F represents the width or height of the convolution kernel, P represents the size of the padding value, and S represents the size of the step.

[0066] Specifically, in another optional embodiment, step 1022 obtains first feature map data, and step 102 can further comprise:

[0067] Step 1023, performing linear processing on the first feature map data by a preset activation function ReLU(x)=max(0,x) to obtain second feature map data; wherein x is the output value of the convolution layer, and ReLU(x) is the output value after applying the preset activation function;

[0068] Specifically, the preset activation function is used to increase the non-linear characteristics of the network, and the preset activation function changes all negative values to zero, thereby introducing non-linearity. This step enables the network to learn complex features and improves the accuracy of the feature map.

[0069] Step 1024, scaling the second feature map data by maximum pooling or average pooling to obtain the feature map data.

[0070] Specifically, maximum pooling selects the maximum value in each pooling window as the output, while average pooling calculates the average value in each pooling window as the output. The purpose is to reduce the size of the feature map, reduce the amount of calculation and prevent overfitting.

[0071] In an optional embodiment of the present application, step 103 comprises:

[0072] Step 1031, grid division is performed on the feature map data to obtain a plurality of grid cells;

[0073] Specifically, the feature map data can be divided into SxS grid cells, and if the center point coordinate of the luggage falls within a certain grid cell, the grid cell is responsible for determining the position of the luggage.

[0074] Step 1032, a plurality of anchor points are determined according to the plurality of grid cells;

[0075] Specifically, a plurality of anchor points are set in each grid cell. Anchor points are possible object regions preset on a feature map, and they have different scales and proportions, corresponding to objects of different sizes and shapes. Anchor points can be set based on the size and shape of the luggage. Anchor points are predefined reference boxes used to cover possible objects in a feature map. Each anchor point is assigned a class label and a bounding box regression value.

[0076] Step 1033, luggage position data is obtained according to the plurality of anchor points.

[0077] Specifically, by calculating the intersection over union between anchor points and real luggage bounding boxes, it is determined which anchor points match the luggage. Here, only when the intersection over union between an anchor point and an object exceeds a predetermined threshold, they are considered to match. For anchor points that match the luggage, their position data (such as center point coordinates, bounding box dimensions, etc.) is extracted. These data represent the position of the luggage in the image. Therefore, the luggage position data can include position data (such as center point coordinates and bounding box dimensions, etc.).

[0078] In an optional embodiment of the present application, step 104 comprises:

[0079] Step 1041, determining the initial state of the luggage according to the luggage position data;

[0080] Specifically, the luggage position data determined according to the first frame of image in the original image data of the luggage is the initial position of the luggage, i.e. the initial state of the luggage.

[0081] Step 1042, inputting the initial state of the luggage into a state prediction model to obtain a luggage state prediction result;

[0082] Specifically, the initial state of the luggage (i.e. the initial position of the luggage) is input into the formula of the state prediction model 、 to obtain the prediction result of the luggage state, i.e. the luggage state prediction result. The specific steps include:

[0083] The initial state of the luggage is taken as the initial value state. For the initial covariance Perform random initialization;

[0084] According to the preset state transition matrix Preset process noise covariance matrix Initial value state Initial covariance and , We obtain the baggage status prediction results and covariance prediction results;

[0085] Among them, the baggage status prediction result at time k=1 is obtained through get;

[0086] The prediction covariance at time k=1 is obtained through... get;

[0087] in, The preset state transition matrix is ​​used. The preset process noise covariance matrix, Here is the baggage status prediction result at time k. The baggage status prediction result at time k-1, To control the input matrix, To control the input, Let k be the prediction covariance. Let the prediction covariance be at time k-1. for The transpose of the matrix, This is the baggage status prediction result at time k=1. This is the preset state transition matrix when k=1. This is the initial state. The control input matrix when k=1 For control input when k=1, Let the prediction covariance be at time k=1. The initial covariance, The preset process noise covariance matrix when k=1.

[0088] Step 1043: Input the baggage location data and the baggage status prediction result into the status update model to obtain the initial baggage trajectory data.

[0089] Specifically, baggage location data also includes real-time location data of the baggage, that is, the current location data of the baggage;

[0090] Based on the current location data of the luggage, the luggage status prediction results, and , to obtain a residual error;

[0091] wherein, is a residual error, is current position data of the baggage at time k in the baggage position number, is a baggage state prediction result at time k;

[0092] According to , a gain is calculated;

[0093] wherein, is a gain, is a prediction covariance at time k, H is an observation matrix, and R is a measurement noise covariance matrix;

[0094] According to the residual error, the gain, and , a baggage state estimation value is obtained;

[0095] wherein, is a final baggage state estimation value at time k, is a baggage state prediction value at time k, is a gain, is a residual error;

[0096] According to the baggage state estimation value at each time, the residual error, and the gain, an initial baggage trajectory data can be determined, and here, the initial baggage trajectory data includes the residual error, the gain, and the baggage state estimation value, which are connected to form

[0097] In an optional embodiment of the present application, step 105 comprises:

[0098] Step 1051, determining matching data according to the initial baggage trajectory data;

[0099] Specifically, a similarity is calculated according to the initial baggage trajectory data and , wherein d is a similarity at time k, is a residual error in the initial baggage trajectory data, is a residual error covariance matrix, , wherein H is an observation matrix, is a prior error covariance matrix, R is a measurement noise covariance matrix, each time similarity is filled into a blank matrix to obtain a correlation matrix, and the correlation matrix is the matching data.

[0100] Step 1052, modifying the initial baggage trajectory data according to the matching data to obtain a target baggage trajectory.

[0101] Specifically, by subtracting the minimum value of each row (or column) of the matching data (i.e., the association matrix), the row transformation and the column transformation are performed on the association matrix, so that each row and each column contains at least one zero element; the least number of rows and columns are used to cover all the zero elements, so as to find the rows and columns that can be matched; if the number of rows and the number of columns covered by the current zero elements are not equal, the element values in the rows and columns that are not covered need to be adjusted, and new zero elements are added, so that the number of rows and the number of columns covered by the current zero elements are equal, which can be achieved by finding the minimum value in the uncovered elements, and then performing subtraction operation on the uncovered rows and columns, and performing addition operation on the intersection points (i.e., the points on the selected rows and columns); until the minimum zero element set covering all rows and columns is found, i.e., the optimal matching; the baggage trajectory point at each time point in the initial baggage trajectory data is connected with the trajectory point at the next time point matched therewith, to form a continuous trajectory, and if a trajectory point at a certain time point cannot find a match (i.e., the point is regarded as an isolated point in the optimal matching), the continuity of the trajectory is maintained by interpolation, smoothing or deleting the point. Through the above steps, the initial baggage trajectory data can be corrected to obtain the target baggage trajectory.

[0102] As shown in Figure 1 the optional embodiment of the present application, the method further comprises the following steps:

[0103] Step 106, modeling processing is performed according to the target baggage trajectory, to obtain a baggage storage model;

[0104] Specifically, when the position of the baggage in the target baggage trajectory remains unchanged within a preset time period, it indicates that the baggage has stopped moving (e.g., is placed in the belly compartment of the aircraft), at this time, a three-dimensional modeling software is used to model all the baggage according to the position (e.g., coordinates) of the baggage in the target baggage trajectory, to obtain the baggage storage model, so that the position of the baggage to be searched can be intuitively seen according to the baggage storage model, thereby saving searching time and labor cost.

[0105] Step 107, determining the baggage position information according to the received baggage position query request and the baggage storage model.

[0106] Specifically, after receiving the baggage position query request, the tag data and the original image data of the matched baggage can be found from the target image data of the baggage according to the passenger information or the baggage image information in the baggage position query request, and then the position of the baggage, i.e., the baggage position information, can be found from the baggage storage model according to the original image data, so as to take out the baggage from the belly compartment of the aircraft according to the baggage position information subsequently.

[0107] One specific embodiment of the baggage trajectory generation method of the embodiment of the present application includes:

[0108] Step 111, acquiring baggage data;

[0109] The video monitoring is recorded by the image acquisition device throughout the process, or the video data and image data of the luggage are acquired by the camera, and the luggage carrying the RFID chip is scanned by the automatic identification device to obtain label data including at least luggage serial number data, flight data, destination data and passenger data.

[0110] In step 112, feature map data is determined.

[0111] The original image data is preprocessed in at least one of scaling, cropping and normalization, and then the feature map data is obtained according to a preset formula.

[0112] In step 113, luggage position data is determined.

[0113] The feature map data is divided into a plurality of grids of a preset size, a plurality of anchor points are set in each grid unit based on the size and shape of the luggage, and the luggage position data is determined by calculating the intersection over union between the anchor points and the real luggage bounding box.

[0114] In step 114, initial luggage trajectory data is determined.

[0115] The initial position of the luggage is determined according to the first frame image in the original image data of the luggage, and then the initial luggage trajectory data is obtained by using a state prediction model and a state update model.

[0116] In step 115, target luggage trajectory is determined.

[0117] The initial luggage trajectory is corrected by a preset luggage trajectory optimization model to calculate the correlation matrix, and the target luggage trajectory is obtained.

[0118] The luggage trajectory generation method of the embodiment of the present application generates the luggage trajectory by acquiring the video and image data of the luggage in the process of luggage security inspection and transportation, which is simple and easy to implement, can accurately locate the luggage position, helps users to find the luggage, and reduces the workload of manual searching.

[0119] As shown in Figure 2 The embodiment of the present application proposes a luggage trajectory generation device 200, which comprises:

[0120] The acquisition module 201 is used to acquire target image data of the luggage; the target image data comprises original image data of the luggage and label data of the luggage.

[0121] The processing module 202 is configured to perform feature extraction on the target image data to obtain feature map data, detect a luggage position according to the feature map data to obtain luggage position data, determine initial luggage trajectory data according to the luggage position data, and correct the initial luggage trajectory data to obtain target luggage trajectory data.

[0122] Optionally, the target image data of the luggage is obtained, and the obtaining the target image data of the luggage comprises:

[0123] The original image data of the luggage is obtained by using an image acquisition device;

[0124] The label data of the luggage is obtained by scanning a luggage shipping label, and the label data of the luggage comprises luggage serial number data, flight data, destination data and passenger data;

[0125] The target image data of the luggage is determined according to the original image data of the luggage and the label data of the luggage.

[0126] Optionally, the feature extraction is performed on the target image data to obtain the feature map data, and the feature extraction comprises:

[0127] The target image data is preprocessed to obtain input data;

[0128] The feature map data is obtained according to the input data and 、 ; the feature map data comprises a height of the feature map, a width of the feature map and a value of the feature map;

[0129] wherein C(i, j) represents a pixel value of the feature map at a position (i, j), I represents the input data, K represents a convolution kernel, (i, j) represents the position of the feature map, m and n represent dimensions of the convolution kernel, I(i+m, j+n) represents a pixel value of the input data at a position (i+m, j+n), K(m, n) represents a weight value of the convolution kernel at a position (m, n), N represents the width of the feature map, W represents the height of the feature map, F represents a width or a height of the convolution kernel, P represents a size of a padding value, and S represents a size of a step.

[0130] Optionally, the luggage position is detected according to the feature map data to obtain the luggage position data, and the detecting the luggage position comprises:

[0131] The feature map data is divided into a plurality of grid units;

[0132] A plurality of anchor points are determined according to the plurality of grid units;

[0133] The luggage position data is obtained according to the plurality of anchor points.

[0134] Optionally, based on the luggage location data, an initial luggage trajectory is determined, including:

[0135] Based on the luggage location data, determine the initial state of the luggage;

[0136] Input the initial state of the luggage into the state prediction model , The baggage status prediction results are obtained.

[0137] The luggage location data and the luggage status prediction results are input into the status update model to obtain the initial luggage trajectory data.

[0138] Among them, the baggage status prediction result at time k=1 is obtained through get;

[0139] The prediction covariance at time k=1 is obtained through... get;

[0140] in, The preset state transition matrix is ​​used. The preset process noise covariance matrix, Here is the baggage status prediction result at time k. The baggage status prediction result at time k-1, To control the input matrix, To control the input, Let k be the prediction covariance. Let the prediction covariance be at time k-1. for The transpose of the matrix, This is the baggage status prediction result at time k=1. This is the preset state transition matrix when k=1. This is the initial state. The control input matrix when k=1 For control input when k=1, Let the prediction covariance be at time k=1. For the initial covariance, The preset process noise covariance matrix when k=1.

[0141] Optionally, the initial baggage trajectory data is corrected to obtain the target baggage trajectory, including:

[0142] Based on the initial luggage trajectory input, matching data is determined;

[0143] The initial luggage trajectory is corrected based on the matching data to obtain the target luggage trajectory.

[0144] Optionally, the processing module 202 is further configured to:

[0145] perform modeling processing according to the target luggage trajectory, to obtain a luggage storage model;

[0146] determine luggage position information according to the received luggage position query request and the luggage storage model.

[0147] The luggage trajectory generation device provided by the embodiment of the application improves the accuracy of the luggage trajectory, and is beneficial to the airport staff to search for the luggage according to the luggage trajectory, and has the advantages of saving labor cost and time.

[0148] It should be noted that the device corresponds to the above method, and all implementation manners in the above method embodiments are applicable to the device embodiments and can achieve the same technical effects. Details are not described herein again.

[0149] The embodiment of the application further provides a computing device, including a processor and a memory storing a computer program, when the computer program is executed by the processor, the method in any one of the above embodiments is executed. All implementation manners in the above method embodiments are applicable to the device embodiments and can achieve the same technical effects. Details are not described herein again.

[0150] The embodiment of the application further provides a computer readable storage medium, which stores instructions, when the instructions are executed on a computer, the computer executes the method in any one of the above embodiments. All implementation manners in the above method embodiments are applicable to the device embodiments and can achieve the same technical effects. Details are not described herein again.

[0151] It should be noted that in the device and method of the application, each component or step can be decomposed and / or recombined. These decompositions and / or recombination should be regarded as equivalent solutions of the application. Moreover, the steps of performing the above series of processes can be executed in time sequence according to the order of description, but do not need to be executed in time sequence. Some steps can be executed in parallel, cross or independently of each other.

[0152] It is to be understood that the terminology "including", "comprising", or any other variation thereof, is intended to cover a non-exclusive inclusion such that process, method, article, or apparatus that comprises a list of elements does not include only those elements but can also include other elements not expressly listed or inherent to such process, method, article, or apparatus. Exclusion of such elements is only present if it is expressly stated that these elements are excluded. Further, it is to be understood that the description of the embodiments of the present application is not limited to the order of the steps of the methods described herein and that unless otherwise specified, the steps of the methods described herein can be performed in any order. Additionally, features described with respect to certain examples can be combined in other examples.

[0153] The above description is considered that the preferred embodiments of the present application, it should be noted that for those skilled in the art, without departing from the principles of the present application described, can also make a number of improvements and refinements, these improvements and refinements should also be considered the scope of protection of the present application.

Claims

1. A method of generating a baggage trajectory, characterized by, The method comprises the following steps: acquiring target image data of the luggage; the target image data comprises original image data of the luggage and label data of the luggage; performing feature extraction according to the target image data to obtain feature map data; detecting the position of the luggage according to the feature map data to obtain luggage position data; determining initial luggage trajectory data according to the luggage position data; correcting the initial luggage trajectory data to obtain a target luggage trajectory; wherein, determining the initial luggage trajectory data according to the luggage position data comprises: determining the initial state of the luggage according to the luggage position data; inputting the initial state of the luggage into a state prediction model , , obtaining a luggage state prediction result; inputting the luggage position data and the predicted result of the luggage state into a state update model to obtain the initial luggage trajectory data; wherein the baggage state prediction result at time k = 1 is obtained by ; where the prediction covariance at time k = 1 is obtained by = 1 is obtained by wherein, is a preset state transition matrix, is a preset process noise covariance matrix, is a baggage state prediction result at time k, is a baggage state prediction result at time k-1, is a control input matrix, is a control input, is a prediction covariance at time k, is a prediction covariance at time k-1, is a transpose matrix of is a baggage state prediction result at time k=1, is a preset state transition matrix at k=1, is an initial value state, is a control input matrix at k=1, is a control input at k=1, is a prediction covariance at time k=1, is an initial covariance, is a preset process noise covariance matrix at k=1;​ wherein, correcting the initial luggage trajectory data to obtain the target luggage trajectory comprises: According to the initial luggage trajectory data, determine matching data; according to the initial luggage trajectory data and Calculate the similarity, wherein, is the similarity at time k, is the residual in the initial luggage trajectory data, is the residual covariance matrix, wherein H is an observation matrix, is the prior error covariance matrix, R is the measurement noise covariance matrix, fill in the blank matrix every time similarity, get the correlation matrix, which is the matching data; correcting the initial luggage trajectory data according to the matching data to obtain the target luggage trajectory.

2. The method of claim 1, wherein, acquiring target image data of the luggage comprises: acquiring original image data of the luggage through an image acquisition device; acquiring label data of the luggage by scanning the luggage shipping label; the label data of the luggage comprises luggage serial number data, flight data, destination data and passenger data; determining the target image data of the luggage according to the original image data of the luggage and the label data of the luggage.

3. The method of claim 1, wherein, performing feature extraction according to the target image data to obtain feature map data comprises: preprocessing the target image data to obtain input data; According to the input data and , , obtain feature map data; the feature map data includes the height of the feature map, the width of the feature map and the value of the feature map; wherein, C(i,j) represents the pixel value of the feature map at position (i,j), I represents the input data, K represents the convolution kernel, (i,j) represents the position of the feature map, m and n represent the dimensions of the convolution kernel, I(i+m,j+n) represents the pixel value of the input data at position (i+m,j+n), K(m,n) represents the weight value of the convolution kernel at position (m,n), N represents the width of the feature map, W represents the height of the feature map, F represents the width or height of the convolution kernel, P represents the size of the padding value, and S represents the size of the step.

4. The method of claim 1, wherein, detecting the position of the luggage according to the feature map data to obtain luggage position data comprises: dividing the feature map data into a plurality of grid units; determining a plurality of anchor points according to the plurality of grid units; obtaining luggage position data according to the plurality of anchor points.

5. The method of claim 1, wherein, The method further comprises the following steps: modeling processing according to the target luggage trajectory to obtain a luggage storage model; determining luggage position information according to the received luggage position query request and the luggage storage model.

6. A luggage trajectory generating apparatus characterized by comprising: The method comprises the following steps: an acquisition module is configured to acquire target image data of the luggage; the target image data comprises original image data of the luggage and label data of the luggage; a processing module is configured to perform feature extraction according to the target image data to obtain feature map data; detect the position of the luggage according to the feature map data to obtain luggage position data; and determine initial luggage trajectory data according to the luggage position data; correcting the initial luggage trajectory data to obtain a target luggage trajectory; wherein, determining the initial luggage trajectory data according to the luggage position data comprises: According to the luggage position data, a luggage initial state is determined; inputting the initial state of the luggage into a state prediction model , , to obtain a luggage state prediction result; The luggage position data and the luggage state prediction result are input into a state update model to obtain initial luggage trajectory data; wherein the baggage state prediction result at time k = 1 is obtained by = 1 is obtained by where the prediction covariance at time k = 1 is obtained by = 1 is given by wherein, is a preset state transition matrix, is a preset process noise covariance matrix, is a baggage state prediction result at time k, is a baggage state prediction result at time k-1, is a control input matrix, is a control input, is a prediction covariance at time k, is a prediction covariance at time k-1, is a transpose matrix of is a baggage state prediction result at time k=1, is a preset state transition matrix at k=1, is an initial value state, is a control input matrix at k=1, is a control input at k=1, is a prediction covariance at time k=1, is an initial covariance, is a preset process noise covariance matrix at k=1;​ Wherein, the initial luggage trajectory data is corrected to obtain a target luggage trajectory, including: According to the initial luggage trajectory data, determine matching data; according to the initial luggage trajectory data and Calculate the similarity, wherein, is the similarity at time k, is the residual in the initial luggage trajectory data, is the residual covariance matrix, wherein H is an observation matrix, is a prior error covariance matrix, R is a measurement noise covariance matrix, fill in the blank matrix with each time similarity, and obtain a correlation matrix, which is the matching data; According to the matching data, the initial luggage trajectory data is corrected to obtain a target luggage trajectory.

7. A computing device, comprising: Including: A processor, a memory storing a computer program, when the computer program is run by the processor, the method of any one of claims 1 to 5 is executed.

8. A computer-readable storage medium, characterized in that, An instruction is stored, when the instruction is run on a computer, the method of any one of claims 1 to 5 is executed.

Images