Pedestrian re-identification method and system

Abstract

The invention provides a pedestrian re-identification method and system. The method comprises the following steps: outputting a source domain text description corresponding to a source domain pedestrian image in real time through a first preset model; extracting an initial keyword and an initial confidence probability correspondingly contained in the source domain text description in real time, and generating a text feature expression template corresponding to the source domain pedestrian image in real time according to the initial keyword and the initial confidence probability; updating the initial keyword and the initial confidence probability in real time through a second preset model to generate a corresponding target keyword and a target confidence probability in real time, and training to generate a corresponding target learning network in real time; and correspondingly migrating the target learning network into the target domain to generate a target text description corresponding to the target pedestrian image in the target domain in real time through the target learning network, and completing pedestrian re-recognition processing of the target pedestrian image in real time according to the target text description. The working efficiency can be effectively improved.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a pedestrian re-identification method and system. Background Art

[0002] With the advancement of science and technology and the rapid development of productivity, video acquisition and image processing technology has also developed rapidly and has been widely used in many fields. It can be used to complete tasks that are difficult for people to complete, greatly facilitating people's lives.

[0003] Among them, pedestrian re-identification technology has also developed rapidly. Specifically, the existing pedestrian re-identification is a technology that matches the same pedestrian in real time in images taken by different cameras. It can be widely used in smart security for video detection and has high use value.

[0004] Furthermore, in the process of implementing pedestrian re-identification technology, most of the existing technologies directly use pre-trained models for text-to-image feature learning, and in the learning process, they heavily rely on pedestrian training sample pairs with sufficient annotations. However, in the actual application process, it is difficult to obtain sufficient pedestrian text-to-image annotations, resulting in certain usage limitations and correspondingly reducing work efficiency. Summary of the invention

[0005] Based on this, the purpose of the present invention is to provide a pedestrian re-identification method and system to solve the problem that the existing technology is difficult to obtain sufficient annotations of pedestrian text to image, resulting in certain usage limitations of the existing pedestrian re-identification technology.

[0006] The first aspect of the embodiment of the present invention proposes:

[0007] A pedestrian re-identification method, wherein the method comprises:

[0008] Acquire a source domain pedestrian image in a preset database, and output a source domain text description corresponding to the source domain pedestrian image in real time through a first preset model;

[0009] Extracting the initial keywords and initial confidence probabilities corresponding to the source domain text description in real time, and generating a text feature expression template corresponding to the source domain pedestrian image in real time according to the initial keywords and the initial confidence probabilities;

[0010] The initial keywords and the initial confidence probabilities are updated in real time by a second preset model to generate corresponding target keywords and target confidence probabilities in real time, and a preset learning network is trained in real time by the target keywords, the target confidence probabilities and the text feature expression template to generate corresponding target learning networks in real time;

[0011] The target learning network is correspondingly migrated to the target domain, so as to generate a target text description corresponding to the target pedestrian image in the target domain through the target learning network in real time, and complete the pedestrian re-identification processing of the target pedestrian image according to the target text description in real time.

[0012] The beneficial effect of the present invention is that by processing the source domain pedestrian images acquired in real time, the corresponding source domain text description can be obtained. Based on this, in order to facilitate subsequent recognition, the required initial keywords and initial confidence probabilities will be further extracted at this time. Based on this, a text feature expression template for subsequent recognition can be generated, and a target learning network for recognition can be further generated according to the text feature expression template. On this basis, pedestrian images existing in the target domain can be re-identified in real time through the target learning network, so that the pedestrian re-identification processing can be completed without using any representation, which eliminates the limitations of use and improves work efficiency.

[0013] Furthermore, the step of outputting a source domain text description corresponding to the source domain pedestrian image in real time through the first preset model includes:

[0014] When the source domain pedestrian image is acquired in real time, a first prompt word adapted to the source domain pedestrian image is matched in real time in a preset database;

[0015] Inputting the first prompt word into the first preset model to call out the first description algorithm correspondingly contained in the first preset model in real time;

[0016] The source domain pedestrian image is processed by the first description algorithm to generate the source domain text description accordingly.

[0017] Furthermore, the step of processing the source domain pedestrian image by using the first description algorithm to generate the source domain text description accordingly includes:

[0018] When the first description algorithm is acquired in real time, the source domain pedestrian image is analyzed and processed to detect the corresponding first image information in real time;

[0019] The first image information is converted in real time by using the first description algorithm to generate the source domain text description accordingly.

[0020] Furthermore, the step of updating the initial keywords and the initial confidence probabilities in real time by using the second preset model to generate corresponding target keywords and target confidence probabilities in real time includes:

[0021] When the initial keyword and the initial confidence probability are obtained respectively, a corresponding second prompt word is matched in real time in the preset database;

[0022] Inputting the second prompt word into the second preset model to extract the update algorithm contained in the second preset model in real time;

[0023] The initial keywords and the initial confidence probabilities are updated in real time by the updating algorithm to generate the target keywords and the target confidence probabilities in real time.

[0024] Furthermore, the step of updating the initial keywords and the initial confidence probabilities in real time by using the updating algorithm to generate the target keywords and the target confidence probabilities in real time includes:

[0025] When the update algorithm is obtained in real time, the initial keywords and the initial confidence probabilities are parsed in real time to extract corresponding original values ​​in real time;

[0026] The original value is updated in real time by the updating algorithm to generate the target keyword and the target confidence probability in real time.

[0027] Furthermore, the step of completing pedestrian re-identification processing of the target pedestrian image according to the target text description in real time includes:

[0028] When it is detected in real time that the pedestrian re-identification process has been completed, the corresponding processing result is obtained in real time, and a corresponding processing report is generated in real time according to the processing result;

[0029] The processing report is encrypted in real time to generate a corresponding encrypted processing report in real time, and the encrypted processing report is stored in a preset folder in real time.

[0030] Furthermore, the step of performing real-time encryption processing on the processing report to generate a corresponding encrypted processing report in real time includes:

[0031] When the processing report is obtained in real time, a number of numbers and letters corresponding to the processing report are detected in real time;

[0032] A number of the numbers and the letters are randomly arranged and combined to generate a number of serial numbers in real time, and a serial number is randomly selected as an encryption key of the processing report to generate the encrypted processing report in real time.

[0033] The second aspect of the embodiment of the present invention proposes:

[0034] A pedestrian re-identification system, wherein the system comprises:

[0035] An acquisition module, used to acquire a source domain pedestrian image from a preset database, and output a source domain text description corresponding to the source domain pedestrian image in real time through a first preset model;

[0036] An extraction module, used to extract the initial keywords and initial confidence probabilities corresponding to the source domain text description in real time, and generate a text feature expression template corresponding to the source domain pedestrian image in real time according to the initial keywords and the initial confidence probabilities;

[0037] A training module, used to update the initial keywords and the initial confidence probabilities in real time through a second preset model to generate corresponding target keywords and target confidence probabilities in real time, and to train a preset learning network through the target keywords, the target confidence probabilities and the text feature expression template in real time to generate corresponding target learning networks in real time;

[0038] A processing module is used to migrate the target learning network to the target domain, so as to generate a target text description corresponding to the target pedestrian image in the target domain through the target learning network in real time, and complete the pedestrian re-identification processing of the target pedestrian image according to the target text description in real time.

[0039] Furthermore, the acquisition module is specifically used for:

[0040] When the source domain pedestrian image is acquired in real time, a first prompt word adapted to the source domain pedestrian image is matched in real time in a preset database;

[0041] Inputting the first prompt word into the first preset model to call out the first description algorithm correspondingly contained in the first preset model in real time;

[0042] The source domain pedestrian image is processed by the first description algorithm to generate the source domain text description accordingly.

[0043] Furthermore, the acquisition module is specifically used for:

[0044] When the first description algorithm is acquired in real time, the source domain pedestrian image is analyzed and processed to detect the corresponding first image information in real time;

[0045] The first image information is converted in real time by using the first description algorithm to generate the source domain text description accordingly.

[0046] Furthermore, the training module is specifically used for:

[0047] When the initial keyword and the initial confidence probability are obtained respectively, a corresponding second prompt word is matched in real time in the preset database;

[0048] Inputting the second prompt word into the second preset model to extract the update algorithm contained in the second preset model in real time;

[0049] The initial keywords and the initial confidence probabilities are updated in real time by the updating algorithm to generate the target keywords and the target confidence probabilities in real time.

[0050] Furthermore, the training module is specifically used for:

[0051] When the update algorithm is obtained in real time, the initial keywords and the initial confidence probabilities are parsed in real time to extract corresponding original values ​​in real time;

[0052] The original value is updated in real time by the updating algorithm to generate the target keyword and the target confidence probability in real time.

[0053] Furthermore, the processing module is specifically used for:

[0054] When it is detected in real time that the pedestrian re-identification process has been completed, the corresponding processing result is obtained in real time, and a corresponding processing report is generated in real time according to the processing result;

[0055] The processing report is encrypted in real time to generate a corresponding encrypted processing report in real time, and the encrypted processing report is stored in a preset folder in real time.

[0056] Furthermore, the processing module is specifically used for:

[0057] When the processing report is obtained in real time, a number of numbers and letters corresponding to the processing report are detected in real time;

[0058] A number of the numbers and the letters are randomly arranged and combined to generate a number of serial numbers in real time, and a serial number is randomly selected as an encryption key of the processing report to generate the encrypted processing report in real time.

[0059] The third aspect of the embodiment of the present invention proposes:

[0060] A computer comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned pedestrian re-identification method when executing the computer program.

[0061] The fourth aspect of the embodiments of the present invention proposes:

[0062] A readable storage medium stores a computer program, wherein the program, when executed by a processor, implements the pedestrian re-identification method as described above.

[0063] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 A flowchart of a pedestrian re-identification method provided by the first embodiment of the present invention;

[0065] Figure 2 This is a structural block diagram of a pedestrian re-identification system provided in the third embodiment of the present invention.

[0066] The following specific implementation manner will further illustrate the present invention in conjunction with the above-mentioned drawings. DETAILED DESCRIPTION

[0067] In order to facilitate the understanding of the present invention, the present invention will be described more fully below with reference to the relevant drawings. Several embodiments of the present invention are given in the drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. On the contrary, the purpose of providing these embodiments is to make the disclosure of the present invention more thorough and comprehensive.

[0068] It should be noted that when an element is referred to as being "fixed to" another element, it may be directly on the other element or there may be a central element. When an element is considered to be "connected to" another element, it may be directly connected to the other element or there may be a central element at the same time. The terms "vertical", "horizontal", "left", "right" and similar expressions used herein are for illustrative purposes only.

[0069] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used herein in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more related listed items.

[0070] See also Figure 1 , shown is a pedestrian re-identification method provided by the first embodiment of the present invention. The pedestrian re-identification method provided by this embodiment can quickly and effectively complete the pedestrian re-identification process without using identification, which eliminates the limitations of use and improves work efficiency.

[0071] Specifically, this embodiment provides:

[0072] A pedestrian re-identification method specifically comprises the following steps:

[0073] Step S10, obtaining a source domain pedestrian image in a preset database, and outputting a source domain text description corresponding to the source domain pedestrian image in real time through a first preset model;

[0074] Step S20, extracting the initial keywords and initial confidence probabilities corresponding to the source domain text description in real time, and generating a text feature expression template corresponding to the source domain pedestrian image in real time according to the initial keywords and the initial confidence probabilities;

[0075] Step S30, updating the initial keywords and the initial confidence probabilities in real time through a second preset model to generate corresponding target keywords and target confidence probabilities in real time, and training a preset learning network through the target keywords, the target confidence probabilities and the text feature expression template in real time to generate corresponding target learning networks in real time;

[0076] Step S40, migrating the target learning network to the target domain, so as to generate a target text description corresponding to the target pedestrian image in the target domain through the target learning network in real time, and completing the pedestrian re-identification processing of the target pedestrian image according to the target text description in real time.

[0077] Specifically, in this embodiment, it should be noted that in order to quickly and effectively complete the pedestrian re-identification process, it is necessary to obtain information related to pedestrians in real time. Preferably, in order to accurately obtain information related to pedestrians, the present invention will take pedestrian images as processing objects. It should be pointed out that after the existing cameras or cameras capture various images in real time, they will be stored in a pre-set database accordingly. Based on this, a database of historical data will be formed, that is, the required historical database will be generated in real time. Based on this, the pedestrian re-identification method provided by the present invention can obtain several source domain pedestrian images for subsequent training in the existing historical database in real time, that is, pedestrian images for subsequent training. Based on this, the present invention will further call out the required multimodal model from the existing model database. Specifically, the multimodal model can be obtained from the prior art. At the same time, the present invention will split the current multimodal model and split the required first preset model, i.e., the basic model, and the second preset model, i.e., the feedback model. Based on this, the present invention will further extract the initial keywords and initial confidence probabilities corresponding to the current several source domain pedestrian images in real time through the above-mentioned first preset model to facilitate subsequent processing.

[0078] Furthermore, the present invention will further enable the above-mentioned second preset model. Specifically, in order to improve the accuracy of model training, it is necessary to obtain the latest training parameters accordingly. Based on this, the present invention will further update the current initial keywords and initial confidence probabilities through the current second preset model, and can further generate the required target keywords and target confidence probabilities. At the same time, it can further generate the required text image representation features according to the current target keywords and target confidence probabilities. Based on this, the subsequent model training can be further completed, and the required target learning network can be finally trained. On this basis, it is necessary to further transfer the current target learning network to the corresponding target domain, that is, the scenario where it needs to be applied, and be able to perform real-time pedestrian re-identification processing on the target domain pedestrian images in the current target domain, so that the re-identification processing can be completed quickly and effectively without using identification, which eliminates the usage limitations and improves work efficiency.

[0079] Second embodiment

[0080] Furthermore, the step of outputting a source domain text description corresponding to the source domain pedestrian image in real time through the first preset model includes:

[0081] When the source domain pedestrian image is acquired in real time, a first prompt word adapted to the source domain pedestrian image is matched in real time in a preset database;

[0082] Inputting the first prompt word into the first preset model to call out the first description algorithm correspondingly contained in the first preset model in real time;

[0083] The source domain pedestrian image is processed by the first description algorithm to generate the source domain text description accordingly.

[0084] Furthermore, the step of processing the source domain pedestrian image by using the first description algorithm to generate the source domain text description accordingly includes:

[0085] When the first description algorithm is acquired in real time, the source domain pedestrian image is analyzed and processed to detect the corresponding first image information in real time;

[0086] The first image information is converted in real time by using the first description algorithm to generate the source domain text description accordingly.

[0087] Furthermore, the step of updating the initial keywords and the initial confidence probabilities in real time by using the second preset model to generate corresponding target keywords and target confidence probabilities in real time includes:

[0088] When the initial keyword and the initial confidence probability are obtained respectively, a corresponding second prompt word is matched in real time in the preset database;

[0089] Inputting the second prompt word into the second preset model to extract the update algorithm contained in the second preset model in real time;

[0090] The initial keywords and the initial confidence probabilities are updated in real time by the updating algorithm to generate the target keywords and the target confidence probabilities in real time.

[0091] Furthermore, the step of updating the initial keywords and the initial confidence probabilities in real time by using the updating algorithm to generate the target keywords and the target confidence probabilities in real time includes:

[0092] When the update algorithm is obtained in real time, the initial keywords and the initial confidence probabilities are parsed in real time to extract corresponding original values ​​in real time;

[0093] The original value is updated in real time by the updating algorithm to generate the target keyword and the target confidence probability in real time.

[0094] Furthermore, the step of completing pedestrian re-identification processing of the target pedestrian image according to the target text description in real time includes:

[0095] When it is detected in real time that the pedestrian re-identification process has been completed, the corresponding processing result is obtained in real time, and a corresponding processing report is generated in real time according to the processing result;

[0096] The processing report is encrypted in real time to generate a corresponding encrypted processing report in real time, and the encrypted processing report is stored in a preset folder in real time.

[0097] Furthermore, the step of performing real-time encryption processing on the processing report to generate a corresponding encrypted processing report in real time includes:

[0098] When the processing report is obtained in real time, a number of numbers and letters corresponding to the processing report are detected in real time;

[0099] A number of the numbers and the letters are randomly arranged and combined to generate a number of serial numbers in real time, and a serial number is randomly selected as an encryption key of the processing report to generate the encrypted processing report in real time.

[0100] In addition, in the present embodiment, it is also necessary to explain that after the required source domain pedestrian image is acquired in real time through the above steps, in order to further extract the source domain text description corresponding to the current source domain pedestrian image, it is necessary to further parse the current source domain pedestrian image. Preferably, the present invention will enable the above-mentioned first preset model. Based on this, the present invention will further match the first prompt word corresponding to the current source domain pedestrian image in real time. It should be noted that in order to shorten the model training time, the present invention applies the prompt word. Specifically, the prompt word can be used to extract the first description algorithm corresponding to the current first preset model in real time, so that the first image information of the current source domain pedestrian image can be directly converted and processed by the first description algorithm, and can be further converted into the required source domain text description. Based on this, the initial keywords and initial confidence probabilities contained in the internal corresponding parts of the current source domain text description can be further extracted in real time. Based on this, in order to obtain the latest training parameters, the above-mentioned second preset model will be further enabled at this time. Similarly, in order to further shorten the model training time, the present invention will further match the update algorithm corresponding to the current initial keywords and initial confidence probabilities. Based on this, the current initial keywords and initial confidence probabilities can be directly updated through the update algorithm, and the target keywords and target confidence probabilities for final training can be further obtained to facilitate subsequent processing.

[0101] Furthermore, after obtaining the required target keywords and target confidence probabilities in real time through the above steps, the current pre-set learning network can be further trained by the target keywords and target confidence probabilities, and the target learning network for subsequent identification can be finally trained. Based on this, it is necessary to further apply the current target learning network, that is, transfer it to the above target domain to complete the subsequent pedestrian re-identification processing in real time. Based on this, after the real-time detection to complete the identification processing, in order to enable the staff to intuitively obtain the identification results, the present invention will further generate a corresponding processing report. At the same time, in order to further improve the confidentiality of the processing report to prevent data leakage, the current processing report will be further extracted. At the same time, the current numbers and letters are immediately randomly arranged and combined, and a number of serial numbers can be generated accordingly. Based on this, a serial number can be randomly selected as the encryption key of the current processing report, so that the current processing report can be encrypted in real time, and then the pedestrian re-identification processing can be completed quickly and effectively without using an identifier, which eliminates the usage limitations and improves work efficiency.

[0102] See also Figure 2, the third embodiment of the present invention provides:

[0103] A pedestrian re-identification system, wherein the system comprises:

[0104] An acquisition module, used to acquire a source domain pedestrian image from a preset database, and output a source domain text description corresponding to the source domain pedestrian image in real time through a first preset model;

[0105] An extraction module, used to extract the initial keywords and initial confidence probabilities corresponding to the source domain text description in real time, and generate a text feature expression template corresponding to the source domain pedestrian image in real time according to the initial keywords and the initial confidence probabilities;

[0106] A training module, used to update the initial keywords and the initial confidence probabilities in real time through a second preset model to generate corresponding target keywords and target confidence probabilities in real time, and to train a preset learning network through the target keywords, the target confidence probabilities and the text feature expression template in real time to generate corresponding target learning networks in real time;

[0107] A processing module is used to migrate the target learning network to the target domain, so as to generate a target text description corresponding to the target pedestrian image in the target domain through the target learning network in real time, and complete the pedestrian re-identification processing of the target pedestrian image according to the target text description in real time.

[0108] Furthermore, the acquisition module is specifically used for:

[0109] When the source domain pedestrian image is acquired in real time, a first prompt word adapted to the source domain pedestrian image is matched in real time in a preset database;

[0110] Inputting the first prompt word into the first preset model to call out the first description algorithm correspondingly contained in the first preset model in real time;

[0111] The source domain pedestrian image is processed by the first description algorithm to generate the source domain text description accordingly.

[0112] Furthermore, the acquisition module is specifically used for:

[0113] When the first description algorithm is acquired in real time, the source domain pedestrian image is analyzed and processed to detect the corresponding first image information in real time;

[0114] The first image information is converted in real time by using the first description algorithm to generate the source domain text description accordingly.

[0115] Furthermore, the training module is specifically used for:

[0116] When the initial keyword and the initial confidence probability are obtained respectively, a corresponding second prompt word is matched in real time in the preset database;

[0117] Inputting the second prompt word into the second preset model to extract the update algorithm contained in the second preset model in real time;

[0118] The initial keywords and the initial confidence probabilities are updated in real time by the updating algorithm to generate the target keywords and the target confidence probabilities in real time.

[0119] Furthermore, the training module is specifically used for:

[0120] When the update algorithm is obtained in real time, the initial keywords and the initial confidence probabilities are parsed in real time to extract corresponding original values ​​in real time;

[0121] The original value is updated in real time by the updating algorithm to generate the target keyword and the target confidence probability in real time.

[0122] Furthermore, the processing module is specifically used for:

[0123] When it is detected in real time that the pedestrian re-identification process has been completed, the corresponding processing result is obtained in real time, and a corresponding processing report is generated in real time according to the processing result;

[0124] The processing report is encrypted in real time to generate a corresponding encrypted processing report in real time, and the encrypted processing report is stored in a preset folder in real time.

[0125] Furthermore, the processing module is specifically used for:

[0126] When the processing report is obtained in real time, a number of numbers and letters corresponding to the processing report are detected in real time;

[0127] A number of the numbers and the letters are randomly arranged and combined to generate a number of serial numbers in real time, and a serial number is randomly selected as an encryption key of the processing report to generate the encrypted processing report in real time.

[0128] A fourth embodiment of the present invention provides a computer, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the pedestrian re-identification method as described above when executing the computer program.

[0129] A fifth embodiment of the present invention provides a readable storage medium having a computer program stored thereon, wherein the program, when executed by a processor, implements the pedestrian re-identification method as described above.

[0130] In summary, the pedestrian re-identification method provided by the above-mentioned embodiment of the present invention can quickly and effectively complete the pedestrian re-identification process without using identification, thereby eliminating the limitations of use and improving work efficiency.

[0131] It should be noted that the above modules can be functional modules or program modules, and can be implemented by software or hardware. For modules implemented by hardware, the above modules can be located in the same processor; or the above modules can be located in different processors in any combination.

[0132] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0133] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0134] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0135] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0136] The above-described embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the present invention. It should be pointed out that, for a person of ordinary skill in the art, several modifications and improvements can be made without departing from the concept of the present invention, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the attached claims.

Claims

1. A pedestrian re-identification method, characterized in that: The method comprises: Acquire a source domain pedestrian image in a preset database, and output a source domain text description corresponding to the source domain pedestrian image in real time through a first preset model; Extracting the initial keywords and initial confidence probabilities corresponding to the source domain text description in real time, and generating a text feature expression template corresponding to the source domain pedestrian image in real time according to the initial keywords and the initial confidence probabilities; The initial keywords and the initial confidence probabilities are updated in real time by a second preset model to generate corresponding target keywords and target confidence probabilities in real time, and a preset learning network is trained in real time by the target keywords, the target confidence probabilities and the text feature expression template to generate corresponding target learning networks in real time; The target learning network is correspondingly migrated to the target domain, so as to generate a target text description corresponding to the target pedestrian image in the target domain through the target learning network in real time, and complete the pedestrian re-identification processing of the target pedestrian image according to the target text description in real time.

2. The pedestrian re-identification method according to claim 1, characterized in that: The step of outputting a source domain text description corresponding to the source domain pedestrian image in real time through a first preset model comprises: When the source domain pedestrian image is acquired in real time, a first prompt word adapted to the source domain pedestrian image is matched in real time in a preset database; Inputting the first prompt word into the first preset model to call out the first description algorithm correspondingly contained in the first preset model in real time; The source domain pedestrian image is processed by the first description algorithm to generate the source domain text description accordingly.

3. The pedestrian re-identification method according to claim 2, characterized in that: The step of processing the source domain pedestrian image by using the first description algorithm to generate the source domain text description accordingly includes: When the first description algorithm is acquired in real time, the source domain pedestrian image is analyzed and processed to detect the corresponding first image information in real time; The first image information is converted in real time by using the first description algorithm to generate the source domain text description accordingly.

4. The pedestrian re-identification method according to claim 3, characterized in that: The step of updating the initial keywords and the initial confidence probabilities in real time by using the second preset model to generate corresponding target keywords and target confidence probabilities in real time includes: When the initial keyword and the initial confidence probability are obtained respectively, a corresponding second prompt word is matched in real time in the preset database; Inputting the second prompt word into the second preset model to extract the update algorithm contained in the second preset model in real time; The initial keywords and the initial confidence probabilities are updated in real time by the updating algorithm to generate the target keywords and the target confidence probabilities in real time.

5. The pedestrian re-identification method according to claim 4, characterized in that: The step of updating the initial keywords and the initial confidence probabilities in real time by using the updating algorithm to generate the target keywords and the target confidence probabilities in real time includes: When the update algorithm is obtained in real time, the initial keywords and the initial confidence probabilities are parsed in real time to extract corresponding original values ​​in real time; The original value is updated in real time by the updating algorithm to generate the target keyword and the target confidence probability in real time.

6. The pedestrian re-identification method according to claim 5, characterized in that: The step of completing pedestrian re-identification processing of the target pedestrian image in real time according to the target text description comprises: When it is detected in real time that the pedestrian re-identification process has been completed, the corresponding processing result is obtained in real time, and a corresponding processing report is generated in real time according to the processing result; The processing report is encrypted in real time to generate a corresponding encrypted processing report in real time, and the encrypted processing report is stored in a preset folder in real time.

7. The pedestrian re-identification method according to claim 6, characterized in that: The step of performing real-time encryption processing on the processing report to generate a corresponding encrypted processing report in real time includes: When the processing report is obtained in real time, a number of numbers and letters corresponding to the processing report are detected in real time; A number of the numbers and the letters are randomly arranged and combined to generate a number of serial numbers in real time, and a serial number is randomly selected as an encryption key of the processing report to generate the encrypted processing report in real time.

8. A pedestrian re-identification system, characterized in that: The system comprises: An acquisition module, used to acquire a source domain pedestrian image from a preset database, and output a source domain text description corresponding to the source domain pedestrian image in real time through a first preset model; An extraction module, used to extract the initial keywords and initial confidence probabilities corresponding to the source domain text description in real time, and generate a text feature expression template corresponding to the source domain pedestrian image in real time according to the initial keywords and the initial confidence probabilities; A training module, used to update the initial keywords and the initial confidence probabilities in real time through a second preset model to generate corresponding target keywords and target confidence probabilities in real time, and to train a preset learning network through the target keywords, the target confidence probabilities and the text feature expression template in real time to generate corresponding target learning networks in real time; A processing module is used to migrate the target learning network to the target domain, so as to generate a target text description corresponding to the target pedestrian image in the target domain through the target learning network in real time, and complete the pedestrian re-identification processing of the target pedestrian image according to the target text description in real time.

9. A computer comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the pedestrian re-identification method according to any one of claims 1 to 7 is implemented.

10. A readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the pedestrian re-identification method as described in any one of claims 1 to 7 is implemented.

Images