A tool and equipment management method and device

Abstract

The present application discloses a tool management method and device, the method comprising: obtaining a facial image of the current operator and the tool status of each current tool in the tool library, the tool status including in-library status and out-of-library status; extracting the facial feature vector corresponding to the facial image using a preset PCA algorithm; tracking the current tool using a preset DCF algorithm, and updating the tool status in real time; performing facial recognition on the current operator based on the facial feature vector and the difference threshold to obtain an operator recognition result; establishing an association between the operator recognition result and the current tool whose tool status is out-of-library status, and obtaining a tool borrowing record. The present application can solve the technical problem that the identity card in the existing RFID-based tool management technology is easy to be lost or replaced, resulting in poor tool management effect.

Description

Technical Field

[0001] The present application relates to the technical field of tool management, and in particular to a tool management method and device. Background Art

[0002] The management of tools and equipment in prefecture-level power companies has always been a difficult problem. There are many types of tools and equipment in large quantities, and they are frequently used and returned. The degree of wear and tear during use is difficult to count, which leads to many problems such as no or incomplete records of use and return, failure to test or scrap after the expiration date, and untimely updating of ledgers. The traditional paper-based warehouse entry and exit registration management method is labor-intensive, inefficient, and prone to loss on site and improper placement.

[0003] At present, the company's tool management is still mainly at the stage of manual registration of incoming and outgoing records. Whether from the perspective of storage space or comprehensive management, the manual recording method is far from keeping up with the needs of tool management. For this reason, the existing technology proposes to realize intelligent tool management based on RFID (radio frequency identification technology). However, in the early stage, each employee and each tool needs to be equipped with a unique corresponding identity card. When the recipient wears the identity card to enter the warehouse, the wireless radio frequency device automatically receives the information in the identity card and records the recipient's information; when the recipient carries the tools out of the warehouse, the radio frequency device automatically reads the tool information and the recipient's identity information, and performs data association, thereby realizing intelligent management of tools.

[0004] The current RFID-based tool management technology requires that employees and tools be equipped with unique corresponding identity cards in advance. Such identity cards are easy to lose, and when identity cards are accidentally exchanged or confused, tool management chaos can easily occur. Summary of the invention

[0005] The present application provides a tool management method and device, which are used to solve the technical problem in the existing RFID-based tool management technology that the identity card is easy to be lost or replaced, resulting in poor tool management effect.

[0006] In view of this, the first aspect of the present application provides a tool management method, comprising:

[0007] Acquire a facial image of a current operator and a tool status of each current tool in a tool library, wherein the tool status includes a library status and an out-of-library status;

[0008] A preset PCA algorithm is used to extract a facial feature vector corresponding to the facial image;

[0009] Tracking the current tool through a preset DCF algorithm and updating the tool status in real time;

[0010] Performing face recognition on the current operator based on the face feature vector and the difference threshold to obtain an operator recognition result;

[0011] An association relationship is established between the operator identification result and the current tool whose tool status is the out-of-warehouse status, so as to obtain a tool borrowing record.

[0012] Preferably, the tracking of the current tool by a preset DCF algorithm and updating the tool status in real time includes:

[0013] Extracting optimal state information of the next frame according to the state information of the previous frame of the current tool by a preset DCF algorithm, wherein the state information includes position information and image size;

[0014] The current tool state is determined according to the optimal state information and the preset area range of the current tool, and the tool state is updated according to the determination result.

[0015] Preferably, the step of performing face recognition on the current operator based on the face feature vector and the difference threshold to obtain an operator recognition result further includes:

[0016] Convert each target face image in the preset face database into a face column vector to obtain a face matrix;

[0017] Performing preset covariance calculation according to the face matrix to obtain a transformation matrix;

[0018] The target projection column vector corresponding to the target face image is calculated by the transformation matrix to obtain a target projection column vector set.

[0019] Preferably, the step of converting each target face image in the preset face database into a face column vector to obtain a face matrix further includes:

[0020] Obtaining employee information of all target operators, the employee information including employee face images, names, and work numbers;

[0021] The employee face image is preprocessed, and a preset face database is constructed based on the obtained target face image. The preprocessing operation includes histogram equalization, normalization, geometric correction and filtering.

[0022] Preferably, performing face recognition on the current operator based on the face feature vector and the difference threshold to obtain an operator recognition result includes:

[0023] Compare the facial feature vector with each of the target projection column vectors in the target projection column vector set one by one to obtain a difference value;

[0024] If the difference value is lower than the difference threshold, the operator corresponding to the target projection column vector is used as the operator identification result.

[0025] A second aspect of the present application provides a tool management device, comprising:

[0026] A first acquisition module is used to acquire a facial image of a current operator and a tool status of each current tool in a tool library, wherein the tool status includes a library status and an out-of-library status;

[0027] A feature extraction module, used to extract a facial feature vector corresponding to the facial image using a preset PCA algorithm;

[0028] A target tracking module, used to track the current tool through a preset DCF algorithm and update the tool status in real time;

[0029] A face recognition module, used to perform face recognition on the current operator based on the face feature vector and the difference threshold, and obtain an operator recognition result;

[0030] The association building module is used to establish an association relationship between the operator identification result and the current tool whose tool status is the out-of-warehouse status, so as to obtain a tool borrowing record.

[0031] Preferably, the target tracking module is specifically used for:

[0032] Extracting optimal state information of the next frame according to the state information of the previous frame of the current tool by a preset DCF algorithm, wherein the state information includes position information and image size;

[0033] The current tool state is determined according to the optimal state information and the preset area range of the current tool, and the tool state is updated according to the determination result.

[0034] Preferably, it also includes:

[0035] A matrix construction module is used to convert each target face image in a preset face database into a face column vector to obtain a face matrix;

[0036] A feature calculation module, used to perform preset covariance calculation according to the face matrix to obtain a transformation matrix;

[0037] The projection calculation module is used to calculate the target projection column vector corresponding to the target face image through the transformation matrix to obtain a target projection column vector set.

[0038] Preferably, it also includes:

[0039] The second acquisition module is used to acquire the employee information of all target operators, wherein the employee information includes the employee's face image, name and work number;

[0040] The preprocessing module is used to perform preprocessing operations on the employee's face image and build a preset face database based on the obtained target face image. The preprocessing operations include histogram equalization, normalization, geometric correction and filtering.

[0041] Preferably, the face recognition module is specifically used for:

[0042] Compare the facial feature vector with each of the target projection column vectors in the target projection column vector set one by one to obtain a difference value;

[0043] If the difference value is lower than the difference threshold, the operator corresponding to the target projection column vector is used as the operator identification result.

[0044] It can be seen from the above technical solutions that the embodiments of the present application have the following advantages:

[0045] In the present application, a tool management method is provided, including: obtaining a facial image of a current operator and the tool status of each current tool in a tool library, the tool status including an in-library status and an out-of-library status; extracting a facial feature vector corresponding to the facial image using a preset PCA algorithm; tracking the current tool using a preset DCF algorithm, and updating the tool status in real time; performing facial recognition on the current operator based on the facial feature vector and a difference threshold to obtain an operator recognition result; establishing an association between the operator recognition result and the current tool whose tool status is an out-of-library status to obtain a tool borrowing record.

[0046] The tool management method provided by the present application does not require operators and tools to be equipped with unique identity cards or label cards. Instead, face recognition matching is performed on operators entering the tool library, and then each tool in the library is tracked in real time, and the tool status is updated. When the tool leaves the library, the tool status information is associated with the current corresponding personnel identification information and recorded, so that tool management can be implemented in a targeted manner to avoid confusion of tools; it can clarify the removal operation of specific operators on specific tools, thereby improving the management efficiency of tools. Therefore, the present application can solve the technical problem that the identity card in the existing RFID-based tool management technology is easy to be lost or replaced, resulting in poor tool management effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 A schematic diagram of a process flow of a tool management method provided in an embodiment of the present application;

[0048] Figure 2 Another flowchart of a tool management method provided by an embodiment of the present application;

[0049] Figure 3 A schematic diagram of the structure of a tool management device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0050] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0051] For easier understanding, see Figure 1 , the present application provides a tool management method embodiment 1, including:

[0052] Step 101: Acquire a facial image of the current operator and a tool status of each current tool in a tool library, where the tool status includes an in-library state and an out-of-library state.

[0053] The current operator is the employee currently entering the tool library. The facial image can be acquired in real time using a camera. The judgment of the tool status also requires analysis of the specific image. A small camera can be installed at the storage location of each tool and set within the library range. The status of the tool can then be determined based on the acquired image.

[0054] Step 102: Use a preset PCA algorithm to extract a facial feature vector corresponding to the facial image.

[0055] The PCA algorithm is the principal component analysis algorithm, which can analyze the main influencing factors from multivariate things, reveal the essence of things, and simplify complex problems; that is, it projects high-dimensional data into a lower-dimensional space, and captures the main features of complex things that can only be described by multiple variables, which are reflected in a small number of main variables, or a linear combination of the original variables, to minimize the required variables. In this embodiment, the PCA algorithm is used to extract features from the face, which is to find the most representative feature space, and then map the face image to the space to obtain a set of features corresponding to the face image.

[0056] Step 103: Track the current tool through a preset DCF algorithm and update the tool status in real time.

[0057] The DCF algorithm is a decision correlation filter. The principle of target tracking through the DCF algorithm is to extract the best target position of the next frame based on the current frame. In this embodiment, the position of the current tool needs to be tracked in real time. When the position of the tool changes, the tool status needs to be updated, that is, from the in-stock state to the out-stock state. If the operator brings in the tool from outside, the out-stock state is updated to the in-stock state.

[0058] Step 104: Perform face recognition on the current operator based on the face feature vector and the difference threshold to obtain an operator recognition result.

[0059] The difference threshold is an adaptive threshold set for comparison and matching. Different matching methods can be selected to compare and match the facial feature vector with the benchmark feature vector calculated by the PCA algorithm, so as to determine the specific warehousing operator.

[0060] Step 105 : Establish an association between the operator identification result and the current tool whose tool status is out of stock, and obtain a tool borrowing record.

[0061] The identification result of the current operator is associated with the current tool status, and each tool can be uniquely associated with an operator to avoid confusion caused by secondary binding. The tool will be unbound when it returns to the tool library. That is, when the tool leaves the library, it will be bound to the current operator and will not disappear out of thin air. When it enters the library, it will be in a non-bound or non-associated state. In this way, both employee information and tool usage information can be clearly recorded in the record information.

[0062] It is understandable that for tools entering the tool library for the first time, the operation flow of entering information is adopted, which is not the tool management process flow described in this embodiment and will not be described in detail here.

[0063] The tool management method provided by the embodiment of the present application does not need to configure unique identity cards or label cards for operators and tools. Instead, it performs facial recognition matching on operators entering the tool library, and then tracks each tool in the library in real time, updates the tool status, and associates the tool status information with the current corresponding personnel identification information when the tool leaves the library. This can achieve targeted tool management and avoid confusion of tools; it can clarify the removal operation of specific operators on specific tools, thereby improving the management efficiency of tools. Therefore, the embodiment of the present application can solve the technical problem that the identity card in the existing RFID-based tool management technology is easy to be lost or replaced, resulting in poor tool management effect.

[0064] For easier understanding, see Figure 2 The present application provides a second embodiment of a tool management method, including:

[0065] Step 201: Acquire the facial image of the current operator and the tool status of each current tool in the tool library, where the tool status includes an in-library state and an out-of-library state.

[0066] In the real-time video, if there is an operator entering the tool library, the facial image of the operator can be extracted, and the tools in the tool library are monitored in real time, that is, the tool status is updated in real time, and can be monitored whether it is stored or taken out.

[0067] Step 202: Use a preset PCA algorithm to extract a facial feature vector corresponding to the facial image.

[0068] The specific process of extracting the features of face images using the PCA algorithm is as follows: first, the face image is expressed as an s×1 column vector A, and then the projection vector corresponding to the face image is calculated: B=W T (A-μ), which is a k×1 column vector, where μ is the mean of the feature vector of the database face image and W is the preset transformation matrix.

[0069] Step 203: Acquire employee information of all target operators, where the employee information includes employee face images, names, and work numbers.

[0070] To build a facial database for operators, in addition to obtaining employee facial images, it is also necessary to complete the basic information of employees, such as name and work number, etc. Other relevant information can also be added according to personalized design requirements, which will not be elaborated here. Employee facial images can be documents with clear faces, such as operator ID photos.

[0071] Step 204: preprocess the employee's face image and build a preset face database based on the obtained target face image. The preprocessing operation includes histogram equalization, normalization, geometric correction and filtering.

[0072] Preprocessing can improve the quality of employee face images and facilitate the improvement of subsequent image analysis results; after preprocessing, the target face image can be obtained, and after all operator information is extracted, the overall preset face database can be constructed for comparison and matching. In addition to equalization, normalization, geometric correction and filtering, preprocessing operations can also add other image processing methods to improve image quality, remove image noise, etc., which are not limited here.

[0073] Step 205: Convert each target face image in the preset face database into a face column vector to obtain a face matrix.

[0074] In the PCA algorithm, the final matching step is the comparison between the current target vector to be identified and the feature vector corresponding to the database. Those with higher similarity are identified as the same target. Therefore, before matching, it is necessary to calculate the feature vector of each face image in the preset face database. In this embodiment, the target face image can be converted into a face column vector first. The entire database is a face matrix, which is convenient for subsequent calculation and analysis.

[0075] Step 206: Perform preset covariance calculation based on the face matrix to obtain a transformation matrix.

[0076] It is understandable that each person may have more than one photo sample. Assuming there are y employees and each employee has z photos, the face matrix formed by the preset database is a y×z matrix. The sample library can be expressed as:

[0077] X=(X1,X2,......,X n ),n=y×z;

[0078] Among them, the face column vector is expressed as X i , that is, the face image of the i-th employee, the size can be expressed as s×1, s is the pixel of the employee's face image; then calculate the mean vector:

[0079]

[0080] Then we can get the centralized sample matrix:

[0081] C=(X1-μ,X2-μ,......,X n -μ), n = y × z;

[0082] Then the covariance matrix is ​​calculated:

[0083] ∑=C·C T ;

[0084] Find the eigenvalue of the covariance matrix ∑, select the largest k value, and find the corresponding eigenvector e i ; Combine and arrange the k eigenvectors to obtain the s×k transformation matrix W=(e1,e2,……,e k ).

[0085] Step 207: Calculate the target projection column vector corresponding to the target face image through the transformation matrix to obtain a target projection column vector set.

[0086] Each target face image can be transformed into the corresponding target projection column vector by using the transformation matrix:

[0087] Y i =W T (Xi -μ);

[0088] The size of this vector is k×1. Based on these target projection column vectors, a target projection column vector set can be constructed and used in the matching process.

[0089] Step 208: extract the best state information of the next frame according to the state information of the previous frame of the current tool through a preset DCF algorithm, where the state information includes position information and image size.

[0090] The specific prediction principle of the DCF algorithm is:

[0091] Now assume that the target box f is composed of a d-dimensional feature vector f(n)∈R d In the process of learning the relevant filter, the RGB value can be extracted for each pixel in the target box, but in fact, any feature representation based on pixel blocks can be considered; then, use f l ,l∈{1,...,d} represents one of the feature channels of f, and a filter h is obtained by learning each feature channel separately l The specific learning process is to transform the relevant response L 2 This is achieved by subtracting the error from the expected related output g to the minimum, the formula is as follows:

[0092]

[0093] Among them, Θ represents the loop-related operation, is a regularization term, where λ is a weight parameter. The desired correlation output g is usually selected as a Gaussian function with a parameterized standard deviation. In addition, f l 、h l , g have the same dimension and size.

[0094] The above formula is a linear least squares problem. It can be effectively solved by converting it to the Fourier domain through the Parseval formula. Finally, the minimized filter formula is as follows:

[0095]

[0096] Capital letters represent the discrete Fourier transform (DFT) of the corresponding lowercase variable, and overscores (such as ) indicates complex conjugation.

[0097] Given the target frame f of the previous frame, the optimal correlation filter h is obtained. However, in order to learn a more robust correlation filter h, it is necessary to comprehensively consider multiple target frames in multiple frames. This can be done by calculating the value of all target boxes f1,...,f in the above formula. t The linear least squares problem can be diagonalized by DFT, and then the final result H is obtained by solving N d×d linear systems (where N is equal to the filter h l The number of elements in ), but this will result in too much calculation. For this reason, the exact solution obtained by a single target box can be used to calculate a robust approximation, so the target box f of the current frame is used t Contains information to update the filter Molecules and denominator B t , as the filter required for the next frame prediction. The specific formula is as follows:

[0098]

[0099]

[0100] Among them, η is the learning rate parameter, and B t-1 Both are the numerator and denominator of the filter updated in the previous frame.

[0101] In order to apply the filter to the new frame t, it is necessary to extract f from the candidate frame by t The target feature z is extracted using the same method as the target feature in t , in the standard shift correlation filter, z t is an image patch centered on the predicted target location, then the relevant score y t The DFT of is expressed in the Fourier domain as follows:

[0102]

[0103] Through Y t De-inverse Fourier transform: y t =F -1 {Y t} to calculate z t The candidate box with the largest correlation score is the best predicted target box, that is, the best state information.

[0104] In this embodiment, for the tools that are put into storage for the first time, it is necessary to select the initial position and image size of the tools, that is, the initial frame image I0, the initial center position p0 of the target and the initial size s0; then extract the initial feature f0 of the tools; then calculate the initial filters A0 and B0; for the current frame image I0 input in the subsequent continuous tracking process, t , according to the target position p of the previous frame image t-1 and size s0 to extract the feature z of the current framet , and then calculate the corresponding y t , and the current frame image I t The target position p t Set to maximize y t The position of; finally update the filter, that is, according to p t and s0 to extract the target feature f of the current frame t , and calculate the numerator A of the filter t and denominator B t .

[0105] Step 209 : determine the current tool state according to the optimal state information and the preset area range of the current tool, and update the tool state according to the determination result.

[0106] Each tool is set with an adaptive preset area range. Once the tool leaves the preset area range, it is considered that the current tool is out of stock. It is understandable that the size and storage space of different tools are not consistent, so the size of the preset area range must be different. It can be set according to actual conditions, which will not be described here. The optimal status information contains the current position information, which can be judged based on the position comparison to detect whether the tool exceeds the preset area range, and then update the tool status according to the judgment result.

[0107] Step 210: Compare the facial feature vector with each target projection column vector in the target projection column vector set one by one to obtain a difference value.

[0108] It is known that the face image is expressed as an s×1 column vector A, and the projection vector corresponding to the face image is calculated: B=W T (A-μ), that is, the facial feature vector; then the vector B is combined with each target projection column vector Y i By comparing, we can get:

[0109] α i =||Y i -B||;

[0110] where α i is the difference value; the number of target projection column vectors is the number of difference values.

[0111] Step 211: If the difference value is lower than the difference threshold, the operator corresponding to the target projection column vector is used as the operator identification result.

[0112] The difference threshold can be set adaptively, but the target projection column vector corresponding to the smallest difference value can also be directly selected from multiple difference values, and the corresponding operator is the target recognition operator.

[0113] Step 212: Establish an association between the operator identification result and the current tool whose tool status is out of stock, and obtain a tool borrowing record.

[0114] In order to detect in time when an operator enters the tool warehouse, a detection device can be installed at the door of the tool warehouse. Real-time target tracking and monitoring will be carried out only when the detection device at the door is triggered and started, otherwise the monitoring can be stopped.

[0115] It can be understood that the positions of all tools can be tracked during the tool tracking process. When a tool disappears from the image, its status will change from in-warehouse to out-of-warehouse, and the tracking mechanism can continue to execute and continuously track it based on the image of its first entry into the warehouse until it reappears within the preset area. The status will then change from out-of-warehouse to in-warehouse, thus avoiding the problem of misjudgment caused by the camera being blocked by passing employees. When a tool is taken out (or brought in) by an employee and disappears (or appears) in the image, its status will change from in-warehouse to out-of-warehouse (or from out-of-warehouse to in-warehouse), and the employee goes out and closes the door, the system will associate the employee information with the tool information status, thereby realizing automatic management of tool taking in and out.

[0116] For easier understanding, see Figure 3 The present application also provides an embodiment of a tool management device, including:

[0117] The first acquisition module 301 is used to acquire the face image of the current operator and the tool status of each current tool in the tool library, where the tool status includes a library status and an out-of-library status;

[0118] A feature extraction module 302 is used to extract a facial feature vector corresponding to the facial image using a preset PCA algorithm;

[0119] The target tracking module 303 is used to track the current tool through a preset DCF algorithm and update the tool status in real time;

[0120] A face recognition module 304 is used to perform face recognition on the current operator based on the face feature vector and the difference threshold to obtain an operator recognition result;

[0121] The association building module 305 is used to establish an association relationship between the operator identification result and the current tool whose tool status is out of stock, so as to obtain a tool borrowing record.

[0122] Furthermore, the target tracking module 303 is specifically used for:

[0123] By using a preset DCF algorithm, the optimal state information of the next frame is extracted according to the state information of the previous frame of the current tool, and the state information includes position information and image size;

[0124] The current tool state is determined according to the optimal state information and the preset area range of the current tool, and the tool state is updated according to the determination result.

[0125] Furthermore, it also includes:

[0126] A matrix construction module 306, used to convert each target face image in the preset face database into a face column vector to obtain a face matrix;

[0127] A feature calculation module 307, used to perform preset covariance calculation according to the face matrix to obtain a transformation matrix;

[0128] The projection calculation module 308 is used to calculate the target projection column vector corresponding to the target face image through the transformation matrix to obtain a target projection column vector set.

[0129] Furthermore, it also includes:

[0130] The second acquisition module 309 is used to acquire the employee information of all target operators, the employee information including the employee's face image, name and work number;

[0131] The preprocessing module 310 is used to perform preprocessing operations on the employee face images and build a preset face database based on the obtained target face images. The preprocessing operations include histogram equalization, normalization, geometric correction and filtering.

[0132] Furthermore, the face recognition module 304 is specifically used for:

[0133] Compare the facial feature vector with each target projection column vector in the target projection column vector set one by one to obtain a difference value;

[0134] If the difference value is lower than the difference threshold, the operator corresponding to the target projection column vector is taken as the operator recognition result.

[0135] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0136] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0137] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0138] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for executing all or part of the steps of the method described in each embodiment of the present application through a computer device (which can be a personal computer, a server, or a network device, etc.). The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (full name in English: Read-Only Memory, English abbreviation: ROM), random access memory (full name in English: Random Access Memory, English abbreviation: RAM), disk or optical disk and other media that can store program codes.

[0139] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A tool management method, characterized in that: include: The facial image of the current operator and the tool status of each current tool in the tool library are obtained, and the tool status includes the library status and the out-of-library status. The specific process is as follows: Use the camera to obtain the facial image of the current operator in real time; A small camera is installed at the storage location of each tool and set within the library, and the status of the tool is determined based on the acquired image; A preset PCA algorithm is used to extract a facial feature vector corresponding to the facial image; The current tool is tracked by a preset DCF algorithm, and the tool status is updated in real time. The specific process is as follows: Extracting optimal state information of the next frame according to the state information of the previous frame of the current tool by a preset DCF algorithm, wherein the state information includes position information and image size; Determine the current tool state according to the optimal state information and the preset area range of the current tool, and update the tool state according to the determination result; Performing face recognition on the current operator based on the face feature vector and the difference threshold to obtain an operator recognition result; An association relationship is established between the operator identification result and the current tool whose tool status is the out-of-warehouse status, so as to obtain a tool borrowing record.

2. The tool management method according to claim 1, characterized in that: The face recognition of the current operator is performed based on the face feature vector and the difference threshold to obtain the operator recognition result, and the method further includes: Convert each target face image in the preset face database into a face column vector to obtain a face matrix; Performing preset covariance calculation according to the face matrix to obtain a transformation matrix; The target projection column vector corresponding to the target face image is calculated by the transformation matrix to obtain a target projection column vector set.

3. The tool management method according to claim 2, characterized in that: The step of converting each target face image in the preset face database into a face column vector to obtain a face matrix also includes: Obtaining employee information of all target operators, the employee information including employee face images, names, and work numbers; The employee face image is preprocessed, and a preset face database is constructed based on the obtained target face image. The preprocessing operation includes histogram equalization, normalization, geometric correction and filtering.

4. The tool management method according to claim 2, characterized in that: The performing face recognition on the current operator based on the face feature vector and the difference threshold to obtain the operator recognition result includes: Compare the facial feature vector with each of the target projection column vectors in the target projection column vector set one by one to obtain a difference value; If the difference value is lower than the difference threshold, the operator corresponding to the target projection column vector is used as the operator identification result.

5. A tool management device, characterized in that: include: The first acquisition module is used to acquire the facial image of the current operator and the tool status of each current tool in the tool library, wherein the tool status includes a library status and an out-of-library status. The specific process is as follows: Use the camera to obtain the facial image of the current operator in real time; A small camera is installed at the storage location of each tool and set within the library, and the status of the tool is determined based on the acquired image; A feature extraction module, used to extract a facial feature vector corresponding to the facial image using a preset PCA algorithm; The target tracking module is used to track the current tool through a preset DCF algorithm and update the tool status in real time. The target tracking module is specifically used to: Extracting optimal state information of the next frame according to the state information of the previous frame of the current tool by a preset DCF algorithm, wherein the state information includes position information and image size; Determine the current tool state according to the optimal state information and the preset area range of the current tool, and update the tool state according to the determination result; A face recognition module, used to perform face recognition on the current operator based on the face feature vector and the difference threshold, and obtain an operator recognition result; The association building module is used to establish an association relationship between the operator identification result and the current tool whose tool status is the out-of-stock status, so as to obtain a tool borrowing record.

6. The tool management device according to claim 5, characterized in that: Also includes: A matrix construction module is used to convert each target face image in a preset face database into a face column vector to obtain a face matrix; A feature calculation module, used to perform preset covariance calculation according to the face matrix to obtain a transformation matrix; The projection calculation module is used to calculate the target projection column vector corresponding to the target face image through the transformation matrix to obtain a target projection column vector set.

7. The tool management device according to claim 6, characterized in that: Also includes: The second acquisition module is used to acquire the employee information of all target operators, wherein the employee information includes the employee's face image, name and work number; The preprocessing module is used to perform preprocessing operations on the employee's face image and build a preset face database based on the obtained target face image. The preprocessing operations include histogram equalization, normalization, geometric correction and filtering.

8. The tool management device according to claim 6, characterized in that: The face recognition module is specifically used for: Compare the facial feature vector with each of the target projection column vectors in the target projection column vector set one by one to obtain a difference value; If the difference value is lower than the difference threshold, the operator corresponding to the target projection column vector is used as the operator identification result.

Images