Camera-based multi-person and multi-site visual identification process judgment method and system
By using camera visual recognition methods to determine the attribution relationship between action and facial data, action cache and temporary process cache are generated and merged into the final process, solving the problem of action attribution and process judgment in multi-person and multi-location scenarios, and realizing accurate calculation of hospital infection control business.
Patent Information
- Application Number
- CN202411573814.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-06
- Publication Date
- 2025-03-11
AI Technical Summary
Existing technologies cannot correctly attribute and judge actions in multi-person, multi-location scenarios, resulting in the inability to accurately determine the process and nature of the action and the inability to effectively calculate hospital infection control data.
By using a camera-based visual recognition method, the relationship between action and face data is determined, action cache and temporary process cache are generated, the temporary process is merged into the final process, and the nature and name of the process are determined.
It enables accurate and rapid judgment of action processes in multi-person, multi-location scenarios, ensuring the accuracy of action attribution and the efficiency of business processing.
Smart Images
Figure CN119600527A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the field of visual recognition, and specifically relates to a process judgment method and system based on visual recognition of multiple people and multiple venues using cameras. The application scenario is that multiple cameras are used in multiple venues, and there is one or more people doing business under each camera at a time. Background Art
[0003] Hand hygiene (hand washing, sanitary hand disinfection, surgical hand disinfection, etc.) in the process of hospital infection, isolation of multi-drug resistant bacteria, disinfection of surgical instruments, endoscopes, etc., are very suitable for monitoring and management using machine vision. Because machine vision has the characteristics of high accuracy and no intervention. Compared with other methods, it has the following advantages: 1. Non-contact measurement will not cause any damage to the observer or the observed, thus improving the reliability of the system.
[0004] 2. It has a wider spectral response range, such as using infrared measurements that are invisible to the human eye, which expands the visual range of the human eye.
[0005] 3. Long-term stable work. It is difficult for humans to observe the same object for a long time, but machine vision can perform measurement, analysis and identification tasks for a long time.
[0006] However, when using machine vision to manage hospital infection control services, the services based on hospital infection control are generally nursing services. Hospital nursing services are diverse, but they all require monitoring of multiple actions of the same service. We call multiple actions of the same service a process. Monitor the accuracy of the actions in this process, the duration of the actions, whether there are omissions in the actions, whether there are foul actions, whether the order of the actions is correct, etc.
[0007] The actions reported by AI machine vision recognition may be the same action, actions in different places, or actions by different people. The system needs to identify and judge these actions and classify them into the same process. Only then can the subsequent business calculations be carried out.
[0008] Among them, there is a scenario: including one or more venues in different places, one venue has one or more cameras, and one or more people work together under each camera. A complete work process of each person will be completed in multiple venues, and one or more actions may be performed in each venue.
[0009] In the above scenario, the following conditions exist: 1. The same person completes different actions of the same business in different venues or in the same venue; 2. The same person completes different actions for different businesses in different venues or in the same venue; 3. Different people complete different actions of the same business or the same action of the same business in different places or in the same place; 4. Different people complete different actions of different businesses or the same action of the same business in different venues or the same venue.
[0010] In the prior art, only the process judgment of visual recognition under the condition of one person in multiple venues and only one camera in one venue is solved. For the above scenarios, when multiple people appear in the same venue or different venues, and multiple people do different business and do not move, it is impossible to determine to which person the action belongs, and it is also impossible to correctly judge the process to which the action belongs, and the nature and name of the process, and ultimately the calculation of hospital infection business cannot be achieved.
[0011] In view of this, how to solve the problem in the prior art that actions in multiple-person and multi-venue scenarios cannot be correctly attributed and judged has become a topic to be studied and solved by the present invention. Summary of the invention
[0012] The purpose of the present invention is to provide a process judgment method and system based on multi-person and multi-site visual recognition of cameras, which is used to determine the process to which an action belongs in a multi-person and multi-site scenario, as well as the nature and name of the process, thereby providing business capabilities related to machine vision results.
[0013] To achieve the above-mentioned purpose, the technical solution adopted in the first aspect of the present invention is: a process judgment method based on visual recognition of multiple people and multiple venues by cameras is proposed, and its innovation lies in that the process judgment method includes: Based on one or more cameras, the motion data and face data of one or more persons to be identified are obtained and reported; Determine the ownership of the recognition action data, determine the ownership of the recognition face data, and establish the corresponding ownership relationship between the face data and the action data of the same person to be recognized; According to the corresponding belonging relationship of the same person to be identified, the same action of the same person to be identified is classified and cached, and the action cache is generated and stored in the database cache; The multiple action caches of the same person to be identified are judged to determine whether the action data belongs to the same process and the same person to be identified. All action data belonging to the same process are collected into a set to form a temporary process, and a temporary process cache is established and reported. According to the temporary process cache reported by the cache, it is judged whether the current action data is the last action of the temporary process. If it is judged to be, multiple temporary processes of the same person are merged into a final process, the nature and name of the final process are judged, and finally saved in the database; In the step of merging multiple temporary processes of the same person into a final process, when it is determined that the current temporary process is the lower half of a whole process and the multiple temporary processes belong to the same face data, the multiple temporary processes of the same person are merged to generate the final process.
[0014] To achieve the above object, the technical solution adopted in the second aspect of the present invention is: a process judgment system based on visual recognition of multiple people and multiple venues by cameras is proposed, and its innovation lies in that the system includes: A camera, wherein the camera is one or more, and the one or more cameras are configured to obtain and report motion data and face data of one or more persons to be identified; An AI machine recognition module, which is used to recognize motion data and face data; A process judgment module, which is used for process judgment based on visual recognition of multiple people and multiple venues by cameras; A storage module stores a database, and the database is used for writing and reading action data, face data, action cache, temporary process cache, and final process.
[0015] The relevant contents of the present invention are explained as follows: 1. The above technical solution of the present invention includes an action attribution algorithm, a same-person same-action judgment program, a same-person temporary process action concentration algorithm, a same-person temporary process transfer algorithm, a temporary process merging judgment algorithm and a data storage program; the purpose is to provide a method for determining that an action recognized by machine vision belongs to a certain person, a same-person same-action judgment program, which merges multiple identical actions reported by machine vision into one action, and calculates the start and end time, and then uses the same-person temporary process action concentration algorithm to attribute multiple different actions of the same person to a temporary process, and finally merges the temporary processes into a final process through the same-person temporary process transfer algorithm, wherein the temporary process merging judgment algorithm determines how to merge temporary processes; the actions and processes here have acquired business significance through configuration; in this way, the action objects recognized by machine vision have also acquired business significance, and the target objects can also be processed for business purposes, and various reports and other businesses can be formed; finally, we can integrate machine vision into the business system. Through the implementation of the above content, a complete set of process judgment methods and systems based on visual recognition of multiple people and multiple venues by cameras can be used to judge the action process in the scene of one or more cameras and one or more persons to be identified, which can ensure that multiple different actions in a final process belong to the same person, and can correctly and quickly judge the nature and name of the process. The implementation of this method makes the retrieval of each data in the application of each step more reasonable and faster, and each step can be closely connected, thereby achieving the purpose of the present invention.
[0016] 2. In the technical solution described in the first aspect above, the ownership of facial data is determined by AI machine vision recognition, and the ownership of N actions is determined by the electronic fence corresponding to the camera to determine which faucet the action data belongs to, and to determine which action among the N action data belongs to which person to be identified, so as to establish a corresponding ownership relationship between the facial data and the action data of the same person to be identified.
[0017] 3. In the technical solution described in the first aspect above, in the step of classifying and caching the same action of the same person to be identified, the AI machine vision recognizes and reports the camera number and faucet number, generates an action cache Key for the action data of the same person to be identified, and checks whether the action cache Key exists in the action cache. If not, the action data is stored in the action cache; if the action cache Key exists, check whether the previous action data of the action cache Key and the action data are of the same action type. If so, update the end time of the action data; if they are not of the same action type, take out the previous action data from the action cache, and replace the action data of the action cache Key in the action cache with the current action.
[0018] 4. In the technical solution described in the first aspect above, in the step of establishing and reporting a temporary process cache, a keyword is generated based on the camera number and faucet number uploaded by the AI machine vision, and the keyword is used to query in the temporary process cache to see whether a temporary process with the keyword exists; if not, a new temporary process is generated, and the current action is added to the new temporary process; if it exists, it is determined whether the action data has face data, and if so, the face data of all action data in the temporary process is updated and saved in the temporary process; if not, the face data of the temporary process is obtained and updated to the action data, and at the same time, the action data is saved in the temporary process.
[0019] 5. In the technical solution described in the first aspect above, in the step of centralizing all action data belonging to the same process into a set to form a temporary process, it includes judging and caching the start action data and the end action data, and generating continuous action data based on the start action and the end action.
[0020] 6. In the technical solution described in the first aspect above, the determination and caching of the start action data comprises the following steps: obtaining a keyword of the start action data, generating an action cache key of the start action data, querying whether the action cache key exists in a database, if the determination is yes, obtaining the action cache key of the start action data and returning it, if the determination is no, generating the action cache key of the start action data and caching it in the database; The judgment and caching of the end action data include the following steps: obtaining the keyword of the end action data, generating an action cache key for the end action data, querying whether the action cache key exists in the database, if the judgment is yes, obtaining the action cache key of the end action data and returning it, if the judgment is no, generating the action cache key of the end action data and caching it in the database.
[0021] 7. In the technical solution described in the first aspect above, in the step of merging multiple temporary processes of the same person, determine whether the current action data is the last action data of the temporary process. If so, merge the temporary processes to generate a final process, and save the current action data and the final process to the database; if it is not the last action data, query the database to see whether the temporary process of the action has been saved. If it has been saved, the current action data is stored in the database. If not, the current action data is stored in the database and the temporary process is saved.
[0022] 8. In the technical solution described in the first aspect above, when merging multiple temporary processes of the same person to generate a final process, it is determined whether the current temporary process is the lower half of a process, if not, then the process ends; if yes, then all temporary processes in the database cache are traversed to obtain the temporary processes in the traversal; it is determined whether the temporary process in the traversal and the current temporary process are the same temporary process, or the face data of the two are different, if yes, then the process ends; if not, then it is determined whether the difference between the first action time of the current temporary process and the last action time of the temporary process to be cycled is less than the time setting value, if no, then the process ends; if yes, then it is determined whether the time of the current temporary process is after the time of the temporary process to be cycled? If no, then the process ends; if yes, then the temporary processes are merged and the final process is generated.
[0023] 9. In the technical solution described in the second aspect above, the process judgment module is configured to execute the steps of the process judgment method described in the first aspect of the present invention.
[0024] Due to the application of the above scheme, the present invention has the following advantages and effects compared with the prior art: Through the implementation of the technical solution of the present invention, a complete set of process judgment methods and systems based on visual recognition of multiple people and multiple venues by cameras can be used to judge the action process in the scene of one or more cameras and one or more persons to be identified, which can ensure that multiple different actions in a final process belong to the same person, and can correctly and quickly judge the nature and name of the process. With the implementation of this method, the retrieval of each data in the application of each step is more reasonable, fast, efficient and accurate, and each step can be closely connected. Through the effective combination of each step, the purpose of the present invention is achieved. BRIEF DESCRIPTION OF THE DRAWINGS
[0025] Attached Figure 1 It is a flowchart of a process judgment method based on visual recognition of multiple people and multiple venues by cameras in Embodiment 1 of the present invention; Attached Figure 2 Schematic diagram of the flow of the action attribution algorithm in the first embodiment of the present invention; Attached Figure 3 It is a flowchart of the same person same action judgment program algorithm in the first embodiment of the present invention; Attached Figure 4 It is a flowchart of the algorithm for concentrating temporary process actions of the same person in the first embodiment of the present invention; Attached Figure 5 Schematic diagram of the process of determining and caching start action data in the first embodiment of the present invention; Attached Figure 6 Schematic diagram of the process of determining and caching the end action data in the first embodiment of the present invention; Attached Figure 7 It is a flowchart of the algorithm for transferring temporary process to process of the same person in the first embodiment of the present invention; Attached Figure 8 This is a flow chart of a temporary process merging judgment algorithm in Embodiment 1 of the present invention; Attached Fig. 9 It is a schematic diagram of an implementation of one application scenario of various embodiments of the present invention. DETAILED DESCRIPTION
[0026] The present invention will be further described below in conjunction with the accompanying drawings and embodiments: Embodiment: The present invention will be clearly described below with diagrams and detailed descriptions. After understanding the embodiments of the present invention, any person skilled in the art can make changes and modifications based on the techniques taught by the present invention without departing from the spirit and scope of the present invention.
[0027] The terms used herein are only for describing specific embodiments and are not intended to be limiting of the present invention. Singular forms such as "a", "this", "here", "this" and "the" as used herein also include plural forms.
[0028] The terms “include,” “including,” “have,” etc. used in this document are open-ended terms, meaning including but not limited to.
[0029] The terms used in this document generally have the ordinary meaning of each term used in this field, in the context of the case and in the specific context, unless otherwise noted. Certain terms used to describe the present invention will be discussed below or elsewhere in this specification to provide additional guidance to those skilled in the art in describing the present invention. Example
[0030] Embodiment 1 of the present invention proposes a process determination method based on visual recognition of multiple people and multiple venues by cameras, and the process determination method includes: Based on one or more cameras, the motion data and face data of one or more persons to be identified are obtained and reported; Determine the ownership of the recognition action data, determine the ownership of the recognition face data, and establish the corresponding ownership relationship between the face data and the action data of the same person to be recognized; According to the corresponding belonging relationship of the same person to be identified, the same action of the same person to be identified is classified and cached, and the action cache is generated and stored in the database cache; The multiple action caches of the same person to be identified are judged to determine whether the action data belongs to the same process and the same person to be identified. All action data belonging to the same process are collected into a set to form a temporary process, and a temporary process cache is established and reported. According to the temporary process cache reported by the cache, it is judged whether the current action data is the last action of the temporary process. If it is judged to be, multiple temporary processes of the same person are merged into a final process, the nature and name of the final process are judged, and finally saved in the database; In the step of merging multiple temporary processes of the same person into a final process, when it is determined that the current temporary process is the lower half of a whole process and the multiple temporary processes belong to the same face data, the multiple temporary processes of the same person are merged to generate the final process.
[0031] Reference Figure 1 In the process determination method of the first embodiment of the present invention, the algorithm content can be divided into the following: Action attribution algorithm: determine the attribution of the recognized action data, determine the attribution of the recognized face data, and establish the corresponding attribution relationship between the face data and action data of the same person to be identified.
[0032] The same person and same action judgment program algorithm: According to the corresponding belonging relationship of the same person to be identified, the same action of the same person to be identified is classified and cached, and the action cache is generated and stored in the database cache.
[0033] The algorithm for concentrating actions in temporary processes of the same person: multiple action caches of the same person to be identified are judged to determine whether the action data belongs to the same process and the same identified person, and all action data belonging to the same process are concentrated into a set to form a temporary process, and a temporary process cache is established and reported.
[0034] The algorithm for converting temporary processes of the same person to processes is as follows: the temporary process cache reported by the cache is used to determine whether the current action data is the last action of the temporary process. If so, multiple temporary processes of the same person are merged into a final process, the nature and name of the final process are determined, and finally saved in the database.
[0035] Temporary process merging judgment algorithm: In the step of merging multiple temporary processes of the same person into a final process, when it is judged that the current temporary process is the lower half of a whole process and multiple temporary processes belong to the same face data, multiple temporary processes of the same person are merged to generate a final process.
[0036] The above contents are explained in detail below.
[0037] Action attribution algorithm: The process of the action attribution algorithm can be found in the attached Figure 2, use AI machine vision recognition to determine the ownership of facial data, use the electronic fence corresponding to the camera to determine the ownership of N actions, determine which faucet the action data belongs to, and determine which action among the N action data belongs to which person to be identified, so as to establish the corresponding ownership relationship between the facial data and action data of the same person to be identified.
[0038] The same person and the same action judgment program algorithm: The algorithm flow of the same person and same action judgment program can be found in the attached Figure 3 , in the step of classifying and caching the same action of the same person to be identified: AI machine vision recognizes and reports the camera number and faucet number; Generate action cache key from the action data of the same person to be identified; Check whether the action cache key exists in the action cache; If it does not exist, the action data is stored in the action cache; If the action cache key exists, check whether the previous action data of the action cache key and the action data are of the same action type; If yes, update the end time of the action data; If they are not the same action type, the previous action data is taken from the action cache, and the action data of the action cache key in the action cache is replaced with the current action.
[0039] The algorithm for the temporary process action concentration of the same person: The process of the algorithm for the temporary process action concentration of the same person can be found in the attached Figure 4 , in the steps of creating a temporary process cache and reporting: Generate keywords based on the camera number and faucet number uploaded by AI machine vision; Use the keyword to query in the temporary procedure cache to see whether the temporary procedure with the keyword exists; If it does not exist, a new temporary process is generated and the current action is added to the new temporary process; If it exists, determine whether the action data contains face data; If yes, update the face data of all motion data in the temporary process and save the motion data to the temporary process; If not, the face data of the temporary process is obtained and updated to the motion data, and at the same time, the motion data is saved in the temporary process.
[0040] The step of collecting all action data belonging to the same process into a set to form a temporary process includes judging and caching the start action data and the end action data, and generating continuous action data according to the start action and the end action.
[0041] Reference Figure 5 , the judgment and caching of the start action data includes the following steps: obtaining a keyword of the start action data, generating an action cache key of the start action data, querying whether the action cache key exists in the database, if the judgment is yes, obtaining the action cache key of the start action data and returning it, if the judgment is no, generating the action cache key of the start action data and caching it in the database; Reference Figure 6 , the judgment and caching of the end action data include the following steps: obtaining the keyword of the end action data, generating the action cache key of the end action data, querying whether the action cache key exists in the database, if the judgment is yes, obtaining the action cache key of the end action data and returning it, if the judgment is no, generating the action cache key of the end action data and caching it in the database.
[0042] The algorithm for transferring temporary processes to processes by the same person: The algorithm for transferring temporary processes to processes by the same person can be found in the attached Figure 7 , in the step of merging multiple temporary processes of the same person: Determine whether the current action data is the last action data of the temporary process; If yes, merge the temporary process to generate a final process, and save the current action data and the final process to the database; If it is not the last action data, query the database to see whether the temporary process of the action has been saved; If it has been saved, the current action data will be stored in the database; If not saved, the current action data is stored in the database and the temporary process is saved.
[0043] Temporary process merge judgment algorithm: The temporary process merging judgment algorithm can be found in the attached Figure 8 , when merging multiple temporary processes of the same person into a final process: Determine whether the current temporary process is the lower half of a process; If no, then end; If yes, then traverse all temporary procedures in the database cache and obtain the temporary procedures in the traversal; Determine whether the temporary process in the traversal and the current temporary process are the same temporary process, or whether the face data of the two are different; If yes, then end; If not, determine whether the difference between the first action time of the current temporary process and the last action time of the temporary process to be cycled is less than the time setting value, which can be set to 3 minutes in one embodiment; If no, then end; If so, determine whether the time of the current temporary process is after the time of the temporary process that is looped to? If no, then end; If so, merge the temporary processes and produce the final process, then end. Example
[0044] Embodiment 2 of the present invention proposes a process judgment system based on visual recognition of multiple people and multiple venues by cameras, and the system includes: A camera, wherein the camera is one or more, and the one or more cameras are configured to obtain and report motion data and face data of one or more persons to be identified; An AI machine recognition module, which is used to recognize motion data and face data; A process judgment module, which is used for process judgment based on visual recognition of multiple people and multiple venues by cameras; A storage module stores a database, and the database is used for writing and reading action data, face data, action cache, temporary process cache, and final process.
[0045] In the second embodiment of the present invention, the process determination module is configured to perform the following steps: Based on one or more cameras, the motion data and face data of one or more persons to be identified are obtained and reported; Determine the ownership of the recognition action data, determine the ownership of the recognition face data, and establish the corresponding ownership relationship between the face data and the action data of the same person to be recognized; According to the corresponding belonging relationship of the same person to be identified, the same action of the same person to be identified is classified and cached, and the action cache is generated and stored in the database cache; The multiple action caches of the same person to be identified are judged to determine whether the action data belongs to the same process and the same person to be identified. All action data belonging to the same process are collected into a set to form a temporary process, and a temporary process cache is established and reported. According to the temporary process cache reported by the cache, it is judged whether the current action data is the last action of the temporary process. If it is judged to be, multiple temporary processes of the same person are merged into a final process, the nature and name of the final process are judged, and finally saved in the database; In the step of merging multiple temporary processes of the same person, when it is determined that the current temporary process is the lower half of a whole process and the multiple temporary processes belong to the same face data, the multiple temporary processes of the same person are merged to generate a final process.
[0046] One of the scenarios when implementing the embodiments of the present invention can be referred to in the attached Fig. 9 As shown, there are 3 cameras and 3 persons to be identified, the cameras are camera 1, camera 2, and camera 3, the persons to be identified are A, B, and C, A, B, and C perform hospital infection control business under camera 1, camera 2, and camera 3 respectively, and the single-frame actions and continuous actions of A, B, and C are shown in the attached figure. Fig. 9 The identification indicates that the camera obtains the motion data and face data of one or more persons to be identified, and can establish the corresponding attribution relationship between the face data and the motion data of the same person to be identified. It can classify and cache the single-frame A action of A, the single-frame B action of A, the single-frame A action of B, the single-frame B action of B, and the single-frame A action of C, classify and cache the same action (single-frame action) of the same person to be identified, generate the action cache and store it in the database, judge the attribution of multiple action caches of the same person to be identified, and collect all the action data belonging to the same process into a set to form a temporary process (single-frame action is converted into continuous action), and finally merge multiple temporary processes of the same person into a final process.
[0047] Through the implementation of the above embodiments, its contents include an action attribution algorithm, a same-person same-action judgment program, a same-person temporary process action concentration algorithm, a same-person temporary process transfer algorithm, a temporary process merging judgment algorithm and a data saving program; the purpose is to provide a method for determining that an action recognized by machine vision belongs to a certain person, a same-person same-action judgment program, which merges multiple identical actions reported by machine vision into one action, and calculates the start and end time, and then uses the same-person temporary process action concentration algorithm to attribute multiple different actions of the same person to a temporary process, and finally merges the temporary processes into a final process through the same-person temporary process transfer algorithm, among which the temporary process merging judgment algorithm determines how to merge temporary processes; the actions and processes here have acquired business significance through configuration; in this way, the action objects recognized by machine vision have also acquired business significance, and the target objects can also be processed for business purposes, and various reports and other businesses can be formed; finally, we can integrate machine vision into the business system. Through the implementation of the above content, a complete set of process judgment methods and systems based on visual recognition of multiple people and multiple venues by cameras can be used to judge the action process in the scene of one or more cameras and one or more persons to be identified, which can ensure that multiple different actions in a final process belong to the same person, and can correctly and quickly judge the nature and name of the process. The implementation of this method makes the retrieval of each data in the application of each step more reasonable and faster, and each step can be closely connected, thereby achieving the purpose of the present invention.
[0048] Those skilled in the art can understand that the operation of the process judgment module in the above-mentioned embodiment can be realized by instructing the relevant hardware (camera) and database (stored in the storage module) through a program, and the program is stored in a storage medium (storage module), including several instructions to enable a device (which can be a single-chip microcomputer, chip, etc.) or a processor (processor) to execute the operation of the process judgment module in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program code, database data, face data, and action data.
[0049] The above embodiments are only for illustrating the technical concept and features of the present invention, and their purpose is to enable people familiar with the technology to understand the content of the present invention and implement it accordingly, and they cannot be used to limit the protection scope of the present invention. Any equivalent changes or modifications made according to the spirit of the present invention should be included in the protection scope of the present invention.
Claims
1. A process judgment method based on visual recognition of multiple people and multiple venues by cameras, characterized in that: The process judgment method comprises: Based on one or more cameras, the motion data and face data of one or more persons to be identified are obtained and reported; Determine the ownership of the recognition action data, determine the ownership of the recognition face data, and establish the corresponding ownership relationship between the face data and the action data of the same person to be recognized; According to the corresponding belonging relationship of the same person to be identified, the same action of the same person to be identified is classified and cached, and the action cache is generated and stored in the database cache; The multiple action caches of the same person to be identified are judged to determine whether the action data belongs to the same process and the same person to be identified. All action data belonging to the same process are collected into a set to form a temporary process, and a temporary process cache is established and reported. According to the temporary process cache reported by the cache, it is judged whether the current action data is the last action of the temporary process. If it is judged to be, multiple temporary processes of the same person are merged into a final process, the nature and name of the final process are judged, and finally saved in the database; In the step of merging multiple temporary processes of the same person, when it is determined that the current temporary process is the lower half of a whole process and the multiple temporary processes belong to the same face data, the multiple temporary processes of the same person are merged to generate a final process.
2. The process judgment method based on visual recognition of multiple people and multiple venues by cameras according to claim 1 is characterized in that: The ownership of facial data is determined through AI machine vision recognition, and the ownership of N actions is determined through the electronic fence corresponding to the camera. It is determined which faucet the action data belongs to, and which action among the N action data belongs to which person to be identified, so as to establish the corresponding ownership relationship between the facial data and action data of the same person to be identified.
3. The process judgment method based on visual recognition of multiple people and multiple places by cameras according to claim 1 is characterized in that: In the step of classifying and caching the same action of the same person to be identified, the AI machine vision recognizes and reports the camera number and the faucet number, generates an action cache key from the action data of the same person to be identified, and checks whether the action cache key exists in the action cache. If not, the action data is stored in the action cache; If the action cache key exists, check whether the previous action data of the action cache key and the action data are of the same action type. If so, update the end time of the action data. If they are not of the same action type, take out the previous action data from the action cache and replace the action data of the action cache key in the action cache with the current action.
4. The process judgment method based on visual recognition of multiple people and multiple places by cameras according to claim 3 is characterized in that: In the step of establishing a temporary process cache and reporting, a keyword is generated according to the camera number and faucet number uploaded by the AI machine vision, and the keyword is used to query in the temporary process cache to see whether a temporary process with the keyword exists; if not, a new temporary process is generated, and the current action is added to the new temporary process; if it exists, it is determined whether the action data has face data, and if so, the face data of all action data in the temporary process is updated and saved in the temporary process; if not, the face data of the temporary process is obtained and updated to the action data, and the action data is saved in the temporary process.
5. The process judgment method based on visual recognition of multiple people and multiple places by cameras according to claim 4 is characterized in that: The step of collecting all action data belonging to the same process into a set to form a temporary process includes judging and caching the start action data and the end action data, and generating continuous action data according to the start action and the end action.
6. The process judgment method based on visual recognition of multiple people and multiple places by cameras according to claim 5 is characterized in that: The determination and caching of the start action data includes the following steps: obtaining a keyword of the start action data, generating an action cache key of the start action data, querying whether the action cache key exists in a database, if the determination is yes, obtaining the action cache key of the start action data and returning it, if the determination is no, generating the action cache key of the start action data and caching it in the database; The judgment and caching of the end action data include the following steps: obtaining the keyword of the end action data, generating an action cache key for the end action data, querying whether the action cache key exists in the database, if the judgment is yes, obtaining the action cache key of the end action data and returning it, if the judgment is no, generating the action cache key of the end action data and caching it in the database.
7. The process judgment method based on visual recognition of multiple people and multiple places by cameras according to claim 4 is characterized in that: In the step of merging multiple temporary processes of the same person, determine whether the current action data is the last action data of the temporary process. If so, merge the temporary processes to generate a final process, and save the current action data and the final process to the database; if it is not the last action data, query the database to see whether the temporary process of the action has been saved. If it has been saved, the current action data is stored in the database; if not, the current action data is stored in the database and the temporary process is saved.
8. The process judgment method based on visual recognition of multiple people and multiple places by cameras according to claim 7 is characterized in that: When merging multiple temporary processes of the same person to generate a final process, determine whether the current temporary process is the lower half of a process, if not, end; If yes, then traverse all temporary procedures in the database cache and obtain the temporary procedures in the traversal; Determine whether the temporary process in the traversal is the same as the current temporary process, or whether the face data of the two are different. If yes, then end. If not, then determine whether the difference between the first action time of the current temporary process and the last action time of the temporary process in the loop is less than the time setting value. If not, then end. If yes, determine whether the time of the current temporary process is after the time of the temporary process to be cycled, if not, end, if yes, merge the temporary processes and produce the final process.
9. A process judgment system based on visual recognition of multiple people and multiple venues by cameras, characterized in that: The system comprises: A camera, wherein the camera is one or more, and the one or more cameras are configured to obtain and report motion data and face data of one or more persons to be identified; An AI machine recognition module, which is used to recognize motion data and face data; A process judgment module, which is used for process judgment based on visual recognition of multiple people and multiple venues by cameras; A storage module stores a database, and the database is used for writing and reading action data, face data, action cache, temporary process cache, and final process.
10. The process judgment system based on visual recognition of multiple people and multiple places by cameras according to claim 9 is characterized in that: The process judgment module is configured to execute the steps of the process judgment method according to any one of claims 1 to 8.