Building labor whole life cycle real-name management method and system based on multi-modal AI perception

Abstract

The application discloses a kind of based on multi-modal AI perception's building labor full life cycle real-name system management method and system, belong to building wisdom construction site technical field.The present application collects at least three biological characteristics of labor personnel, and generates digital identity and chain genesis notarization after feature fusion by living body detection;Dynamic non-inductive identity verification and attendance are realized through all-scene edge AI terminal;Using multi-modal behavior recognition model, identify and real-time early warning with the illegal behavior of post qualification binding;Based on multidimensional data, labor personnel digital ability portrait and credit rating are constructed, and post intelligent adaptation is realized;Relying on tamper-proof data of blockchain, complete salary automatic accounting, two-way confirmation and full-link notarization issue, form the whole life cycle closed-loop management from entry to exit.The present application has the advantages of identity security, control refinement, data credibility, intelligent efficiency, judicial traceability and the like.

Description

Technical Field

[0001] This invention relates to the field of smart construction site technology, specifically a method and system for real-name management of the entire lifecycle of construction labor based on multimodal AI perception. Background Technology

[0002] In the construction engineering field, real-name labor management is an important system and technical means to protect workers' legitimate rights and interests, regulate on-site employment practices, prevent and resolve labor disputes, and implement the main responsibility for safe production. Currently, traditional real-name labor management largely relies on manual registration and filing, IC card access, or simply uses single fingerprint or static facial recognition access control equipment, which has significant technical shortcomings in practical application.

[0003] First, the identity authentication methods are too simplistic and easily forged. Static facial recognition is vulnerable to attacks using photos, videos, and simulated masks, while fingerprint recognition can be cracked through fingerprint films, resulting in insufficient overall security. If multiple biometric authentication methods are used, there are problems such as cumbersome processes, low efficiency, and poor on-site experience, making it difficult to adapt to the high-frequency passage scenarios on construction sites.

[0004] Second, there are serious data silos and a disconnect in the management chain. Data such as personnel identity, attendance records, payroll calculation, safety behavior, and job qualifications are scattered across different systems and platforms, making it impossible to interconnect and share information. This hinders the formation of an integrated and coordinated management system that links "person-certificate-job-salary-behavior," resulting in fragmented management and inadequate supervision.

[0005] Third, the safety monitoring model is passive, and early warning and response are delayed. Video monitoring at construction sites mainly relies on manual patrols, which cannot identify work activities in real time, automatically, and accurately, nor can it quickly associate safety hazards with the identities of specific workers, making it difficult to achieve pre-event warnings and in-event intervention.

[0006] Fourth, the management process is fragmented and lacks automation. From worker registration, attendance tracking, and job duties to payroll calculation and departure clearance, each step relies heavily on manual operation and paper-based document processing. This not only leads to low efficiency and high costs but also makes it prone to errors, omissions, and data loss.

[0007] Existing improved technologies (such as the prior art document CN112001284A) still fail to address the core pain points of the industry: identity authentication relies solely on facial recognition, covering only site entrances and exits, and cannot achieve continuous and reliable verification across all scenarios such as work areas and special equipment zones, leaving the risk of forgery and impersonation still present; behavioral recognition capabilities are limited, only able to identify a few simple behaviors such as not wearing a safety helmet or smoking, and cannot be combined with personnel job qualifications and work permissions for refined management; core data is stored in a centralized database, which is susceptible to tampering, loss, and forgery, making it difficult to provide authoritative and tamper-proof electronic evidence in the event of labor disputes; management processes only cover fragmented stages of "entry attendance - salary calculation," failing to cover the entire lifecycle from entry registration, pre-job training, job matching, work control, salary payment to departure credit evaluation, and thus cannot fundamentally solve the problems of irregular, opaque, and untraceable construction labor management.

[0008] In summary, how to construct a real-name management technology solution for construction labor that integrates high-security multimodal identity verification, closed-loop management throughout the entire process, refined proactive early warning, trusted evidence storage across the entire chain, and intelligent optimization of labor has become an urgent technical problem to be solved in the field of smart construction sites. Summary of the Invention

[0009] This invention aims to address the shortcomings of existing technologies, such as insecure identity authentication, incomplete management processes, imprecise behavior control, and unreliable data. It provides a real-name management solution for the entire lifecycle of construction labor that integrates multimodal biometrics, AI behavior perception, edge computing, and blockchain notarization, thereby achieving credible identities, closed-loop processes, manageable behavior, traceable data, transparent salaries, and intelligent employment.

[0010] The specific technical solution adopted by this invention is as follows:

[0011] A method for real-name management of the entire lifecycle of construction labor based on multimodal AI perception includes the following steps:

[0012] S1, Multimodal Biometrics and Full-Dimensional Qualification Registration and Database Entry: Responding to the registration request of workers entering the site, collecting workers' identity information, job qualifications, labor contracts, and pre-job training assessment results, while collecting at least three biometric features from face, fingerprint, voiceprint, and iris; generating a unique multimodal biometric template through liveness verification and feature fusion, binding it with personnel information to form a digital identity ID, encrypting and storing it, and uploading the core metadata to the blockchain for genesis evidence storage;

[0013] S2, Full-Scene Dynamic Seamless Identity Verification and On-Duty Attendance: Real-time tracking and multimodal feature collection of personnel throughout the construction area through edge AI sensing terminals, digital identity ID is confirmed through dual verification of 3D liveness detection and multimodal features, automatic generation of full-dimensional attendance records and on-chain storage of attendance hash values;

[0014] S3, AI perception of refined work behavior bound to job permissions: The edge AI perception terminal collects audio and video streams of the work scene in real time, and analyzes them through multimodal behavior recognition model to identify violations bound to job qualifications and work permissions. It generates violation records by associating the personnel's digital identity ID, provides real-time warnings, and puts the hash value of the violation record on the blockchain for evidence storage.

[0015] S4, Personnel Capability Profile and Intelligent Job Matching: Integrating training, qualification, attendance, violation, and work quality data, a digital capability profile and credit rating are constructed through a multi-dimensional evaluation algorithm to achieve intelligent job matching, and the profile and credit data are stored on the blockchain for evidence.

[0016] S5, Multi-source data fusion intelligent salary calculation: Based on attendance, violation, qualification, and work data stored on the blockchain, combined with salary rules and reward and punishment system, it automatically calculates basic salary, position allowance, performance bonus and reward and punishment amount, generates salary details and stores the calculation source data and rules on the blockchain for evidence;

[0017] S6, Two-way Confirmation and End-to-End Salary Payment Evidence Storage: Salary details are pushed to both employers and employees for two-way confirmation. If there is no objection, the bank will make the payment. Key data of the entire payment process will be packaged and stored on the consortium blockchain. Employee capability profiles and credit ratings will be updated to complete the management loop.

[0018] Preferably, the edge AI sensing terminal in step S2 is deployed at entrances and exits, work areas, living areas, and special equipment operation areas; the dual verification includes: first performing 3D deep liveness detection and 1:N rapid comparison of the face, and then completing auxiliary cross-verification through voiceprint, fingerprint or iris.

[0019] Preferably, the multimodal behavior recognition model in step S3 is constructed by fusing a spatiotemporal graph convolutional network with a VisionTransformer.

[0020] Preferably, the violations include: not wearing safety protective equipment, operating special equipment in violation of regulations, performing special operations without qualifications, entering dangerous areas beyond authorized limits, performing open flame operations in violation of regulations, and not wearing a safety belt when working at height.

[0021] Preferably, the digital capability profile in step S4 includes a quantitative score of professional skill level, safety and compliance awareness, and work performance capability; the credit rating is linked to job requirements and is used to recommend the optimal employment plan for the project.

[0022] Preferably, in step S5, the salary calculation automatically integrates the basic salary, position allowance, performance bonus, and reward / penalty amount to generate a traceable salary calculation report.

[0023] Preferably, in step S6, when there is a dispute between the employer and employee, the blockchain-stored evidence source data is automatically retrieved to generate a dispute verification report; key data for salary payment include employee ID, accounting cycle, amount due, amount actually paid, attendance hash value, violation record hash value, electronic signature, and payment voucher.

[0024] Preferably, the blockchain is a consortium blockchain jointly participated in by the housing and construction department, construction companies, banks, and labor service companies.

[0025] In addition, this invention also discloses a real-name management system for the entire lifecycle of construction labor services based on multimodal AI perception, including:

[0026] The cloud-based management and blockchain-based evidence storage platform is used to store personnel's full-dimensional information, biometric templates, attendance, behavior, salary, and ability profile data, providing data management, intelligent analysis, and consortium blockchain evidence storage.

[0027] The multimodal registration and training terminal is used to collect multimodal biometrics, identity and qualification information, conduct pre-job training and assessment, and complete personnel registration and database entry.

[0028] The full-scene edge AI perception terminal is equipped with a built-in camera, audio pickup device and edge computing unit, and is used for seamless identity verification, refined behavior AI perception and real-time early warning in all scenarios;

[0029] The mobile interactive terminal is used by laborers and managers to confirm salaries, file objections, query information, and receive early warnings; the intelligent job matching and competency profiling module has a built-in multi-dimensional evaluation algorithm to build digital competency profiles, credit ratings, and intelligent job matching.

[0030] Furthermore, the present invention also includes a computer-readable storage medium storing a computer program that, when executed by a processor, implements the real-name management method for the entire lifecycle of construction labor based on multimodal AI perception as described above.

[0031] Compared with the prior art, the present invention has at least one of the following advantages or beneficial effects:

[0032] 1) Significant improvement in identity credibility: The identity authentication system adopts at least three types of multimodal biometric fusion + full-scenario dynamic cross-core, which completely solves the shortcomings of existing single face recognition technology that is easy to be forged and can only cover access control scenarios. It enables continuous and reliable verification of personnel identity throughout the construction area, and significantly improves security.

[0033] 2) Achieve a closed-loop management process throughout the entire lifecycle: Breaking through the limitations of existing technologies that only cover the fragmented process of "entry attendance - payroll calculation", we build a closed-loop management system that covers the entire lifecycle from registration and database entry, pre-job training, job matching, operation control, payroll distribution to departure credit evaluation, thereby fundamentally solving the problems of fragmented construction labor management processes and disconnected control.

[0034] 3) Significantly improved level of refinement and intelligence in behavior control: The multimodal behavior recognition model, which integrates spatiotemporal graph convolutional network and VisionTransformer, is adopted to realize multi-dimensional violation behavior recognition that is bound to personnel job qualifications and work permissions. This breaks through the limitation of existing technologies that can only identify two types of simple violations, transforming passive post-event tracing into proactive real-time early warning, and greatly improving the level of safety management at construction sites.

[0035] 4) Trustworthy and tamper-proof data across the entire chain: Based on consortium blockchain technology, the core data of personnel identity, attendance, violations, salary calculation, and voucher issuance are stored in an immutable and tamper-proof manner throughout the entire process, which completely solves the problems of easy data tampering in centralized databases and difficulty in providing evidence in labor disputes, and builds a solid foundation of trust between labor and management.

[0036] 5) Achieve intelligent upgrade of labor management: By building digital capability profiles and professional credit systems for laborers, intelligent matching of personnel skills with job requirements can be achieved, providing data support for construction companies to optimize labor allocation and improve construction efficiency, which has extremely high industry promotion value. Attached Figure Description

[0037] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this specification, illustrate exemplary embodiments of the invention and are used to explain the invention, but do not constitute an undue limitation of the invention.

[0038] Figure 1 This is a flowchart of a method for real-name management of the entire lifecycle of construction labor based on multimodal AI perception, as described in an embodiment of the present invention. Detailed Implementation

[0039] To make the objectives and advantages of this invention clearer, the invention will be specifically described below with reference to embodiments. It should be understood that the following text is merely used to describe one or more specific embodiments of the invention and does not strictly limit the scope of protection specifically claimed by the invention.

[0040] Example 1:

[0041] like Figure 1 As shown in the figure, this embodiment discloses a method for real-name management of the entire life cycle of construction labor based on multimodal AI perception. Specifically, the method includes the following six core steps:

[0042] S1, Steps for registering and storing multimodal biometrics and full-dimensional qualifications:

[0043] In response to the workers' registration requests, the system collects their identity information, job qualification certificates, labor contracts, and pre-job safety training assessment results; it also collects at least three of their multimodal biometric information from their face, fingerprints, voiceprints, and irises.

[0044] The collected multimodal biometric information is preprocessed, liveness verification is performed, and feature fusion extraction is carried out to generate a unique multimodal biometric fusion template.

[0045] The multimodal biometric fusion template is uniquely bound to personnel identity information, qualification information, contract information, and training and assessment results to generate a unique digital identity ID for each person. This ID is then encrypted and stored in a cloud database. The personnel digital identity ID and core metadata are simultaneously uploaded to the blockchain network for genesis and evidence preservation, generating an immutable personnel digital identity file.

[0046] S2, Full-Scenario Dynamic Seamless Identity Verification and On-Duty Attendance Steps:

[0047] By deploying edge AI sensing terminals throughout the construction area (entrances and exits, work areas, living areas, and special equipment operation areas), real-time continuous tracking and multimodal biometric collection are carried out on personnel entering the sensing range.

[0048] Simultaneously, the edge AI sensing terminal performs the following: 3D depth liveness detection on the collected facial images to eliminate forgery attacks such as photos, videos, and masks; extracts the multimodal biometrics of the current person and performs a 1:N fast comparison with the locally cached feature template to complete the preliminary identity verification.

[0049] After the initial identity verification is passed, during the set contactless verification window period, the person's voiceprint is collected through environmental sound pickup equipment, or fingerprint / iris features are collected through the touch terminal on the work surface to complete the auxiliary cross-verification. After the dual verification is passed, the person's unique digital identity ID is confirmed.

[0050] Based on the location trajectory and verification records of personnel throughout the entire time, a full-dimensional attendance record containing personnel ID, timestamp, geographical location, verification node, and on-duty duration is automatically generated and uploaded to the cloud database in real time. The hash value of the attendance record is also synchronously uploaded to the blockchain network for evidence storage.

[0051] All liveness detection, feature extraction and comparison, and behavior recognition model inference of the aforementioned edge AI sensing terminal are completed at the edge, without relying entirely on the cloud. It is adapted to weak network / no network environment on construction sites, and the latency is controlled to ≤100ms. The aforementioned blockchain network is a consortium blockchain jointly participated in by housing and construction departments, construction companies, banks, and labor service companies. The stored data has judicial validity and can be directly used as evidence in labor disputes.

[0052] S3, Refined AI perception and association steps for job-related behaviors bound to job permissions:

[0053] By deploying edge AI sensing cameras in various work areas, video and audio streams of the work scene are collected in real time.

[0054] Using a pre-trained multimodal behavior recognition model based on the fusion of spatiotemporal graph convolutional networks and Vision Transformer, real-time inference analysis is performed on video and audio streams at the edge to identify preset violation events. These violation events include: not wearing safety protective equipment, operating special equipment in violation of regulations, carrying out special operations without qualifications, entering hazardous control areas beyond authorized permissions, performing hot work in violation of regulations, and not wearing safety belts while working at heights, etc., which are strongly tied to personnel's job qualifications and work permissions.

[0055] When a violation is detected, the system immediately captures the time, location, video clips, and audio clips of the incident, and extracts the facial image of the violator.

[0056] The facial image of the violator is compared with the multimodal biometric template in the cloud database, linked to the person's unique digital identity ID, and a violation record bound to that person's ID is generated. This record is then pushed to the on-site broadcast terminal for real-time voice warnings, and the hash value of the violation record is uploaded to the blockchain network for evidence storage.

[0057] S4, Steps for building a personnel competency profile that links pre-job training, job matching, and work performance:

[0058] Retrieve personnel's pre-job training assessment results, job qualification certificate levels, historical work attendance records, violation records, and work task completion quality data from the cloud database;

[0059] By using a pre-set multi-dimensional evaluation algorithm model, the professional skills, safety and compliance awareness, and work performance capabilities of personnel are quantitatively scored, and a digital capability profile and credit rating for each worker is constructed.

[0060] Based on the digital capability profiles and credit ratings of personnel and the job requirements of the project, intelligent matching and adaptation are carried out to recommend the best employment plan for project managers.

[0061] The digital capability profiles and credit rating updates of personnel are simultaneously uploaded to the blockchain network for storage, forming an unalterable professional credit file for laborers.

[0062] S5, the steps of intelligent salary calculation across all dimensions through multi-source data fusion:

[0063] Obtain immutable attendance records, violation records, job qualification levels, task completion status, digital ability profile scoring data of target personnel within a specified accounting period from cloud databases and blockchain storage nodes, as well as corresponding labor contract salary rules and project reward and punishment management systems;

[0064] Based on attendance records and job qualification levels, calculate the basic salary and job allowance for employees;

[0065] Performance bonuses for employees are calculated based on their task completion status and competency profile scores.

[0066] Based on records of violations, the amount of reward or penalty is calculated according to predefined reward and penalty rules;

[0067] By integrating basic salary, position allowance, performance bonus, and reward / penalty amounts, the system automatically generates a detailed payroll statement and salary calculation report for the employee. Simultaneously, it uploads the full-dimensional source data and calculation rules of the salary calculation to the blockchain network for evidence storage.

[0068] S6, the closed-loop process of salary payment with two-way confirmation and blockchain full-chain evidence storage:

[0069] The generated payroll details and payroll calculation report are pushed to the mobile terminal bound to the workers and the enterprise's labor management terminal for two-way confirmation between labor and personnel.

[0070] Receive confirmation feedback or objection / appeal information from workers and enterprises; for objection / appeal, automatically retrieve the evidence source data on the blockchain, generate an objection verification report, and provide it to both workers and employees for verification and handling.

[0071] After both employers and employees confirm that there are no errors, the bank's payroll interface is triggered to complete the payroll disbursement. The key data of the entire payroll disbursement process, including employee ID, accounting cycle, amount due, amount actually paid, attendance hash value, violation record hash value, electronic signatures of both parties, and disbursement voucher, are generated into an immutable data block and uploaded to the consortium blockchain network for full-chain evidence storage.

[0072] Based on the comprehensive performance of personnel during this accounting period, update the personnel's digital capability profiles and credit ratings to complete this management loop.

[0073] In a preferred embodiment of the present invention, the method specifically includes:

[0074] Before new workers enter the site, they complete the entry of their identity information, scaffolder special operation qualification certificate, and labor contract information through a multimodal registration and training terminal at the project site. They also complete pre-job safety training and pass the assessment. At the same time, the system collects three types of biometric information of the workers: face, fingerprint, and voiceprint. The system completes liveness detection, feature fusion extraction, and generates a unique multimodal biometric fusion template, which is bound to the worker's identity, qualification, and training information to generate a unique digital identity ID. This ID is encrypted and stored on the cloud platform, and the worker's core identity metadata is uploaded to the consortium blockchain for genesis and evidence storage, generating an unalterable digital identity file.

[0075] After workers enter the construction site, edge AI sensing terminals deployed at entrances, exits, and work floors capture video streams in real time. First, 3D liveness detection is performed on detected faces to rule out photo and video forgery attacks. Facial features are extracted and compared 1:N with a locally cached feature library to complete preliminary identity verification. Simultaneously, the terminal's voiceprint is collected to complete auxiliary cross-verification. After both verifications are successful, the worker's digital identity ID is confirmed. Based on the worker's location trajectory and verification nodes throughout the day, the system automatically generates a comprehensive attendance record including on-duty time and work area, uploads it to the cloud in real time, and synchronizes the hash value of the attendance record to the blockchain for evidence storage.

[0076] While workers were operating on the 12th-floor scaffolding, edge AI sensing terminals deployed on the work surface captured real-time video streams. Using a built-in multimodal behavior recognition model combining spatiotemporal graph convolution and Transformer fusion, the system performed real-time inference and analysis, identifying that the worker was not wearing a safety harness. The system also identified the worker as a scaffolder, qualified for working at heights, but engaging in a violation. The system immediately captured the violation video frame, extracted facial recognition data to confirm identity, generated a violation record linked to the worker's ID, issued a real-time voice warning via on-site broadcast, pushed a warning message to the safety officer's mobile terminal, and stored the hash value of the violation record on the blockchain for evidence.

[0077] At the end of the month, the system retrieves the worker's training and assessment results, qualification level, 26 perfect attendance records, 2 violations of not wearing a safety belt, and the completion status of the scaffolding erection task acceptance. Using a multi-dimensional evaluation algorithm model, it generates a professional skills score of 85, a safety compliance score of 72, and a performance capability score of 90, constructing a digital capability profile and assigning a credit rating of B. When a new scaffolding work task is initiated for the project, the system automatically matches and recommends this worker as a suitable candidate.

[0078] At the end of the month, during payroll calculation, the system automatically retrieves the worker's immutable attendance records, violation records, and work completion data from the blockchain's evidence storage node. Combining this with the labor contract's stipulated daily wage of 350 yuan, a daily allowance of 50 yuan for scaffolding work, performance bonus for external scaffolding operations, and a penalty of 100 yuan for each safety violation, the system automatically calculates: Basic salary = 26 × 350 = 9100 yuan, allowance = 26 × 50 = 1300 yuan, performance bonus = 1200 yuan, penalty for violation = 2 × 100 = 200 yuan, final payable = 9100 + 1300 + 1200 - 200 = 11400 yuan. The system generates a payroll detail and calculation report, and simultaneously stores all source data and calculation rules on the blockchain for evidence storage.

[0079] The system synchronously pushes payroll details and accounting reports to the workers' mobile apps and the project department's labor relations management terminal. After both parties verify the information, they confirm it electronically. The system then triggers the bank's direct deposit interface to complete the payroll disbursement. It packages all key data from the entire payroll disbursement process, including employee ID, accounting period, payable / paid amount, attendance hash value, violation record hash value, electronic signatures from both parties, and bank disbursement voucher, into a data block and uploads it to the consortium blockchain network for end-to-end notarization, creating an immutable disbursement archive. Simultaneously, based on the worker's performance this month, the system updates the worker's digital competency profile and credit rating, completing the closed loop of this full lifecycle management.

[0080] Example 2:

[0081] This embodiment discloses a real-name management system for the entire lifecycle of construction labor based on multimodal AI perception. The system includes: a cloud management and consortium blockchain evidence storage platform deployed at the company headquarters; multimodal registration and training terminals deployed at each project department; full-scene edge AI perception terminals deployed in the entire construction area (entrances and exits, each working floor, tower crane operation area, hot work area, and dangerous areas near the edge); mobile interactive terminals (mobile APP) held by workers and managers; and a built-in intelligent job adaptation and capability profiling module.

[0082] The cloud management and blockchain evidence storage platform is used to store personnel's full-dimensional information, multimodal biometric templates, attendance, behavior, salary, and ability profile data. It provides data management, intelligent analysis, and interaction interfaces, and also has a built-in blockchain evidence storage module for on-chain evidence storage of core data throughout the entire process.

[0083] The multimodal registration and training terminal is deployed at the project site to collect multimodal biometric information, identity and qualification information of workers, conduct pre-job safety training and assessment, and complete personnel registration and database entry.

[0084] The full-scene edge AI perception terminal is deployed throughout the construction area (entrances and exits, work areas, living areas, and special equipment areas). It is equipped with high-definition cameras, audio pickup devices, and edge computing units to perform dynamic and seamless identity verification, refined AI perception of work behavior, and real-time early warning.

[0085] The mobile interactive terminal is a mobile APP / mini-program for laborers and managers, used for laborers and managers to confirm salary details, appeal objections, query information, and receive early warning messages;

[0086] The intelligent job matching and competency profiling module incorporates a multi-dimensional evaluation algorithm model to construct digital competency profiles and credit ratings for workers, enabling intelligent job matching and recommendations.

[0087] It is not difficult to see that this embodiment is a system embodiment corresponding to the above-described embodiment of the construction labor lifecycle real-name management method based on multimodal AI perception. This embodiment can be implemented in conjunction with the above-described embodiment of the construction labor lifecycle real-name management method based on multimodal AI perception. The relevant technical details mentioned in the above-described embodiment of the construction labor lifecycle real-name management method based on multimodal AI perception are still valid in this embodiment, and will not be repeated here to reduce repetition. Accordingly, the relevant technical details mentioned in this embodiment can also be applied to the above-described embodiment of the construction labor lifecycle real-name management method based on multimodal AI perception.

[0088] Example 3:

[0089] This embodiment discloses a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the management method described in Embodiment 1 above.

[0090] The above are merely preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention. Structures, devices, and operating methods not specifically described or explained in this invention are implemented according to conventional methods in the art unless otherwise specified or limited.

Claims

1. A method for real-name management of the entire lifecycle of construction labor based on multimodal AI perception, characterized in that, Includes the following steps: S1, Multimodal Biometrics and Full-Dimensional Qualification Registration and Database Entry: Responding to the registration request of workers entering the site, collecting workers' identity information, job qualifications, labor contracts, and pre-job training assessment results, while collecting at least three biometric features from face, fingerprint, voiceprint, and iris; A unique multimodal biometric template is generated through liveness verification and feature fusion, which is then bound to personnel information to form a digital identity ID. This ID is encrypted, stored, and the core metadata is uploaded to the blockchain for genesis evidence preservation. S2, Full-Scene Dynamic Seamless Identity Verification and On-Duty Attendance: Real-time tracking and multimodal feature collection of personnel throughout the construction area through edge AI sensing terminals, digital identity ID is confirmed through dual verification of 3D liveness detection and multimodal features, automatic generation of full-dimensional attendance records and on-chain storage of attendance hash values; S3, AI perception of refined work behavior bound to job permissions: The edge AI perception terminal collects audio and video streams of the work scene in real time, and analyzes them through multimodal behavior recognition model to identify violations bound to job qualifications and work permissions. It generates violation records by associating the personnel's digital identity ID, provides real-time warnings, and puts the hash value of the violation record on the blockchain for evidence storage. S4, Personnel Capability Profile and Intelligent Job Matching: Integrating training, qualification, attendance, violation, and work quality data, a digital capability profile and credit rating are constructed through a multi-dimensional evaluation algorithm to achieve intelligent job matching, and the profile and credit data are stored on the blockchain for evidence. S5, Multi-source data fusion intelligent salary calculation: Based on attendance, violation, qualification, and work data stored on the blockchain, combined with salary rules and reward and punishment system, it automatically calculates basic salary, position allowance, performance bonus and reward and punishment amount, generates salary details and stores the calculation source data and rules on the blockchain for evidence; S6, Two-way Confirmation and End-to-End Salary Payment Evidence Storage: Salary details are pushed to both employers and employees for two-way confirmation. If there is no objection, the bank will make the payment. Key data of the entire payment process will be packaged and stored on the consortium blockchain. Employee capability profiles and credit ratings will be updated to complete the management loop.

2. The method for real-name management of the entire lifecycle of construction labor based on multimodal AI perception as described in claim 1, characterized in that, The edge AI sensing terminal mentioned in step S2 is deployed at entrances and exits, work areas, living areas, and special equipment operation areas; the dual verification includes: first performing 3D deep liveness detection and 1:N rapid comparison of the face, and then completing auxiliary cross-verification through voiceprint, fingerprint or iris.

3. The method for real-name management of the entire lifecycle of construction labor based on multimodal AI perception as described in claim 1, characterized in that, The multimodal behavior recognition model described in step S3 is constructed by fusing a spatiotemporal graph convolutional network with a VisionTransformer.

4. The method for real-name management of the entire lifecycle of construction labor based on multimodal AI perception as described in claim 1, characterized in that, The violations include: not wearing safety protective equipment, operating special equipment in violation of regulations, performing special operations without qualifications, entering dangerous areas beyond authorized limits, performing open flame operations in violation of regulations, and not wearing a safety belt when working at heights.

5. The method for real-name management of the entire lifecycle of construction labor based on multimodal AI perception as described in claim 1, characterized in that, The digital capability profile mentioned in step S4 includes a quantitative score of professional skill level, safety and compliance awareness, and work performance capability; the credit rating is linked to job requirements and is used to recommend the optimal employment plan for the project.

6. The method for real-name management of the entire lifecycle of construction labor based on multimodal AI perception as described in claim 1, characterized in that, In step S5, the payroll calculation automatically integrates basic salary, position allowance, performance bonus, and reward / penalty amount to generate a traceable payroll calculation report.

7. The method for real-name management of the entire lifecycle of construction labor based on multimodal AI perception as described in claim 1, characterized in that, In step S6, when there is a dispute between the employer and employee, the blockchain-stored evidence source data is automatically retrieved to generate a dispute verification report; key data for salary payment include employee ID, accounting cycle, amount due, amount actually paid, attendance hash value, violation record hash value, electronic signature, and payment voucher.

8. The method for real-name management of the entire lifecycle of construction labor based on multimodal AI perception as described in claim 1, characterized in that, The blockchain is a consortium blockchain jointly participated in by the housing and construction department, construction companies, banks, and labor service companies.

9. A real-name management system for the entire lifecycle of construction labor services based on multimodal AI perception, characterized in that: include: The cloud-based management and blockchain-based evidence storage platform is used to store personnel's full-dimensional information, biometric templates, attendance, behavior, salary, and ability profile data, providing data management, intelligent analysis, and consortium blockchain evidence storage. The multimodal registration and training terminal is used to collect multimodal biometrics, identity and qualification information, conduct pre-job training and assessment, and complete personnel registration and database entry. The full-scene edge AI perception terminal is equipped with a built-in camera, audio pickup device and edge computing unit, and is used for seamless identity verification, refined behavior AI perception and real-time early warning in all scenarios; The mobile interactive terminal is used by laborers and managers to confirm salaries, file objections, query information, and receive early warnings; the intelligent job matching and competency profiling module has a built-in multi-dimensional evaluation algorithm to build digital competency profiles, credit ratings, and intelligent job matching.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the real-name management method for the entire lifecycle of construction labor based on multimodal AI perception as described in any one of claims 1-8.

Images