Dangerous operation safety monitoring and early warning method and system

Abstract

The invention discloses a dangerous operation safety monitoring and early warning method and system, and relates to the field of AI intelligent identification, and the method comprises the steps: monitoring an operation site in real time through the autonomous learning of an employee, the installation of a network server and a mobile AI camera, and carrying out the data collection; screening and finely marking the collected data, carrying out model training, and improving the accuracy and recall rate through cross validation; through the application of the AI camera, a sound production field operation supervision method is established, violation behaviors are judged, warning and snapshot are carried out, and behaviors which are not corrected in time are sent to supervisors. Through real-time monitoring and intelligent analysis, the safety of a working place is ensured, and the accident occurrence risk is reduced. According to the system, violation behaviors are recognized in time through data acquisition and model training, the safety awareness of employees is enhanced, the supervision efficiency is improved, and therefore an efficient working environment is created.

Description

Technical Field

[0001] The present invention relates to the field of AI intelligent recognition, and particularly to a method and system for safety monitoring and early warning of dangerous operations. Background Art

[0002] With the rapid development of modern industry, especially in high-risk fields such as construction, mining, and chemical industries, the importance of work safety has become increasingly prominent. In the past few decades, many countries and regions have successively formulated and implemented strict safety management regulations, which have promoted the development of safety monitoring technologies. Traditional safety monitoring means such as manual inspections, regular inspections, and traditional monitoring systems, although providing certain guarantees for work safety, still lack in real-time performance, comprehensiveness, and intelligence. Specifically, traditional methods are difficult to achieve 24-hour uninterrupted monitoring, and there are delays in the judgment and response of violation behaviors, and potential safety risks cannot be warned in a timely manner. Therefore, there is an urgent need for improved and innovative monitoring means to improve the efficiency and accuracy of safety early warning, so as to ensure the lives of workers and the normal operation of enterprises.

[0003] At present, safety monitoring and early warning methods based on artificial intelligence technology show great potential. Although intelligent monitoring systems have been introduced in some fields, there are still some problems to be solved in the existing technologies. For example, many AI monitoring systems lack refinement for different working conditions and environmental conditions during data collection and processing, resulting in their inability to effectively identify and warn against violation behaviors in complex and changeable on-site environments. Moreover, existing technologies often rely on standard data sets in model training and evaluation, lack the ability of incremental learning for actual production environments, and are difficult to adapt to the rapidly changing industrial demands. Therefore, how to combine device self-learning and deep learning technologies, and further optimize the monitoring and early warning effects through high-quality data collection and accurate model training, has become the key to improving the level of work safety. Summary of the Invention

[0004] In view of the above existing problems, the present invention attempts to monitor the operation site with an AI camera. Timely capture and voice reminder of common violation behaviors in operations such as not wearing a safety helmet, illegal hot work, and smoking are carried out to strengthen the safety awareness of operators against "three violations", enhance the support of scientific and technological empowerment for work safety, and thus form a normalized safety management mechanism.

[0005] To solve the above technical problems, a method for safety monitoring and early warning of dangerous operations is proposed, including,

[0006] Through self-study by employees, network servers and mobile AI cameras are installed to monitor the work site in real time and collect data. The collected data is screened and finely labeled, and model training is carried out, with cross-validation to improve accuracy and recall. Through the application of AI cameras, a sound production site work supervision method is established to identify violations, issue warnings and take photos, and behaviors that are not corrected in time are sent to supervisors.

[0007] As a preferred solution of the dangerous operation safety monitoring and early warning method described in the present invention, the monitoring of the operation site includes installing a network server and a mobile AI camera to monitor the operation site and collect data through employee self-study.

[0008] As a preferred solution of the dangerous operation safety monitoring and early warning method described in the present invention, the data collection includes collecting light data, weather data and working condition data;

[0009] The collected data is screened, and the data finally retained includes different lighting data including day and night; different weather data including sunny days, cloudy days, and rainy days; different operating condition data including normal operating conditions and abnormal operating conditions.

[0010] As a preferred solution of the dangerous operation safety monitoring and early warning method described in the present invention, the fine labeling includes: fine labeling of the screened data through a preset labeling method to ensure the high quality of the training set, setting the labeling accuracy threshold and the ratio threshold of difficult samples in the extraction process through n cross-validation, extracting the difficult samples generated in the labeling process, and performing secondary labeling confirmation by experts to generate a labeled data set.

[0011] As a preferred solution of the dangerous operation safety monitoring and early warning method described in the present invention, the model training includes using a labeled data set to train the model, training on a set machine, setting the duration of each training, and specifying the accuracy, recall rate, and mAP0.5 standard on the validation set.

[0012] As a preferred solution of the dangerous operation safety monitoring and early warning method described in the present invention, the model training also includes deploying the model in an actual industrial site for testing. During the test, the real-time video processing frame rate of the model is required to reach 15fps, the delay time is required to be less than 500ms, and the detection accuracy in actual operation is required to reach 95%.

[0013] As a preferred solution of a safety monitoring and early warning method for dangerous operations according to the present invention, wherein: the model training further includes continuously collecting new data through a self-learning algorithm for incremental training after the model is deployed, regularly updating the model monthly, and increasing the recognition accuracy threshold of the model.

[0014] Another object of the present invention is to provide a safety monitoring and early warning system for dangerous operations. Through real-time monitoring and intelligent analysis, the present invention ensures the safety of the workplace and reduces the risk of accidents. The system uses a network server and a mobile AI camera to monitor the operation site in real time, collect and process data, so as to establish a comprehensive understanding of the work process and environment. By continuously optimizing and improving the recognition accuracy through model training, it can judge violations in real time and issue warnings to ensure the timely handling of potential safety hazards. In addition, the system also aims to enhance employees' safety awareness, improve supervision efficiency, build a safer and more efficient working environment, protect the lives and health of employees, and ensure the smooth progress of production activities.

[0015] As a preferred solution of a safety monitoring and early warning system for dangerous operations according to the present invention, it is characterized by including a data processing module, a model training module, and a supervision module;

[0016] The data processing module, through the independent learning of employees, installs a network server and a mobile AI camera to monitor the operation site in real time and collect data;

[0017] The model training module screens and finely annotates the collected data, conducts model training, and improves the accuracy and recall rate through cross-validation;

[0018] The supervision module, through the application of the AI camera, establishes a sound supervision method for on-site production operations, judges violations, and issues warnings and captures images. Behaviors that are not corrected in time are sent to the supervisors.

[0019] A computer device includes a memory and a processor. The memory stores a computer program. It is characterized in that when the processor executes the computer program, the steps of the safety monitoring and early warning method for dangerous operations as described above are implemented.

[0020] A computer-readable storage medium stores a computer program thereon. It is characterized in that when the computer program is executed by a processor, the steps of the safety monitoring and early warning method for dangerous operations as described above are implemented.

[0021] Advantages of the present invention: By enabling employees to independently learn to install network servers and mobile AI cameras, the present invention realizes real-time monitoring and data collection of the operation site, comprehensively obtains data on light, weather, and working conditions, so as to establish a data foundation. Through fine screening and annotation, a high-quality training set is ensured, providing accurate and reliable data support for model training, and improving the efficiency and accuracy of identifying violation behaviors. The model is tested in an actual industrial site to ensure its effectiveness, and incremental training is carried out through a self-learning algorithm to continuously optimize the recognition ability of the model. At the same time, by integrating data processing, model training, and supervision modules, a comprehensive and automated monitoring and early warning system is formed.

[0022] After the implementation of the present invention, it is possible to promptly stop violation behaviors such as not wearing safety helmets, illegal hot work, smoking, and illegal entry at the operation site, improving the safety awareness of operators in "combating the three violations". By establishing and implementing the supervision method for on-site production operations, a normalized safety supervision mechanism for the operation site can be formed. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0024] Figure 1 It is the overall flowchart of a method for monitoring and early warning of dangerous operation safety provided by an embodiment of the present invention.

[0025] Figure 2 It is the safety helmet recognition diagram of a method for monitoring and early warning of dangerous operation safety provided by an embodiment of the present invention.

[0026] Figure 3 It is the open fire recognition diagram of a method for monitoring and early warning of dangerous operation safety provided by an embodiment of the present invention.

[0027] Figure 4 It is the person recognition diagram of a method for monitoring and early warning of dangerous operation safety provided by an embodiment of the present invention.

[0028] Figure 5 It is the recognition diagram of a method for monitoring and early warning of dangerous operation safety provided by an embodiment of the present invention.

[0029] Figure 6 It is the system scheme module diagram of a system for monitoring and early warning of dangerous operation safety provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0030] To make the above objects, features, and advantages of the present invention more apparent and understandable, the following provides a detailed description of the specific embodiments of the present invention in conjunction with the accompanying drawings of the specification. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the scope of protection of the present invention.

[0031] In the following description, many specific details are set forth to facilitate a thorough understanding of the present invention. However, the present invention may be implemented in other ways different from those described herein. Those skilled in the art can make similar generalizations without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0032] Secondly, the so-called "one embodiment" or "embodiment" herein refers to specific features, structures, or characteristics that may be included in at least one implementation manner of the present invention. The appearances of "in one embodiment" in different places in this specification do not all refer to the same embodiment, nor are they embodiments that are mutually exclusive with other embodiments individually or selectively.

[0033] The present invention is described in detail in conjunction with schematic diagrams. When detailing the embodiments of the present invention, for ease of explanation, the cross-sectional views showing the device structure will be enlarged locally in a non-general proportion, and the schematic diagrams are only examples and should not limit the scope of protection of the present invention here. In addition, in actual production, three-dimensional spatial dimensions including length, width, and depth should be included.

[0034] At the same time, in the description of the present invention, it should be noted that the orientation or positional relationships indicated by terms such as "upper, lower, inner, and outer" are based on the orientation or positional relationships shown in the drawings. It is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation to the present invention. In addition, the terms "first, second, or third" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0035] Unless otherwise clearly defined and limited in the present invention, the terms "installation, connection, and coupling" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can also be a mechanical connection, an electrical connection, or a direct connection, and can also be indirectly connected through an intermediate medium, or can be the communication inside two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific situations.

[0036] Example 1, referring to Figures 1 - 5 , which is the first embodiment of the present invention. This embodiment provides a method for safety monitoring and early warning of dangerous operations, including:

[0037] In the scenario of lightning optical monitoring, S1: Through self-study by employees, install a network server and a mobile AI camera to monitor the operation site in real time and collect data.

[0038] Furthermore, through self-study by employees, install a network server, an industrial control computer, a security dog, a 24-port gigabit switch, 3 Internet of Things cards, and a mobile AI camera to monitor the operation site and collect data; data collection is the first step in algorithm development, and its quality directly determines the effect of the model.

[0039] Specifically, collect illumination data, weather data, and working condition data;

[0040] Screen the collected data. Among the finally retained data, it includes different illumination data including day and night; different weather data including sunny, cloudy, and rainy; different working condition data including normal working conditions and abnormal operation conditions.

[0041] S2: Screen and finely annotate the collected data, perform model training, and improve the accuracy and recall rate through cross-validation.

[0042] Furthermore, finely annotate the screened data through a preset annotation method to ensure the high quality of the training set. Through n-fold cross-validation, set the annotation accuracy threshold to 98% and the proportion threshold of difficult samples in the extraction process to 5%. Extract the difficult samples generated during the annotation process, and have them re-annotated and confirmed by experts. When the accuracy reaches 98% after cross-validation, output the data to generate a well-annotated dataset.

[0043] Specifically, the preset annotation method can be key point annotation, bounding box annotation, semantic annotation, or negative sample annotation. In an embodiment of the present invention, the preset annotation method adopts a method combining bounding box annotation and key point annotation. The specific method is as follows:

[0044] Classify and organize the collected data, integrate the illumination, weather, and working condition data into a specified format, and perform a preliminary review of the data to identify different types of samples, so as to lay a foundation for subsequent annotation work.

[0045] The annotator will use a rectangular box to select relevant targets (such as equipment, personnel, etc.) in the collected video frames or images; when selecting, the annotator needs to ensure that the size of the box is appropriate, which can cover the outer contour of the target and should not surround too many interfering elements in the environment, quickly locate the target object, and provide its spatial position information.

[0046] After completing the annotation of the bounding box, the annotator will add marks to the key corners of the target object (such as the hands and feet of people, the operating buttons of equipment, etc.). After the annotation is completed, it will immediately enter the quality control, and first conduct random sampling verification through 20% of the samples to judge the accuracy of the preliminary annotation. The annotator ensures that the annotation reaches the set 98% accuracy rate and tracks the annotation of difficult samples. Understand whether there are mislabeling or missing labels in the annotation process, and monitor that suspicious samples should be recorded immediately. When the annotation results of some samples are controversial (that is, the annotator gives inconsistent annotations to the same object), the threshold for the proportion of difficult samples in the extraction process is 5%. The difficult samples generated during the extraction and annotation process are confirmed by experts for secondary annotation.

[0047] It should be noted that the model training is carried out using labeled data sets on a set machine, and the duration of each training is set to 24 hours. At the same time, the accuracy standard on the validation set is set to 0.95, the recall rate is 0.90, and the mAP0.5 standard is 0.91. When the accuracy, recall rate and mAP standards are met, the trained data is allowed to be output.

[0048] Specifically, the model is deployed in an actual industrial site for testing. During the test, the real-time video processing frame rate of the model is required to reach 15fps, the delay time is required to be less than 500ms, and the detection accuracy in actual operation is required to reach 95%.

[0049] Through the self-learning algorithm, new data is continuously collected for incremental training after the model is deployed, and the model is updated regularly every month to improve the recognition accuracy threshold of the model.

[0050] S3: Through the application of AI cameras, a sound production site supervision method is established to identify violations, issue warnings and take photos, and send any uncorrected behavior to supervisors; improve workers' safety awareness and good work behavior codes.

[0051] Embodiment 2, the second embodiment of the present invention, provides a dangerous operation safety monitoring and early warning method, comprising:

[0052] The collected data in S2 is screened and finely labeled, and model training is performed, and the accuracy and recall are improved through cross-validation.

[0053] The filtered data is finely annotated by a preset annotation method. Specifically, the preset annotation method can be key point annotation, bounding box annotation, semantic annotation, or negative sample annotation. In one embodiment of the present invention, the preset annotation method adopts a method combining semantic annotation and negative sample annotation. The specific method is as follows:

[0054] Clean and preprocess the data to ensure its usability. Filter out the brightness values in the light data that exceed the reasonable range (such as less than 0 or greater than 1000 lumens), and exclude the negative weather data (such as invalid weather states in the database); Group the data and process the light, weather, and working condition information separately. Each group of data will be verified through manual or automatic algorithm verification to ensure that each piece of data conforms to the real scenario;

[0055] In the image or data, the annotator needs to identify key objects such as workers and equipment, and combine the context information to label each object. For example, the actions of the workers (such as "welding", "loading and unloading") need to be clearly labeled and should be associated with other elements in the working environment (such as tools, machines);

[0056] At the same time, the status information of each element also needs to be marked. For example, whether the worker wears a safety helmet, gloves, etc., and the working status of the equipment (normal, faulty, etc.);

[0057] Record the spatial relationships between different elements. For example, the relative positions of the workers and mechanical equipment in response to the requirements of the safety distance.

[0058] For the samples that do not meet the standards, the annotator clearly differentiates them by comparing the known safe operations with the actual data, and identifies any samples that do not meet the safety standards or operating procedures, such as workers operating in dangerous areas, not wearing appropriate protective equipment, etc.; Mark the improper operations as negative samples to emphasize the safety risks and potential dangerous situations. After the preliminary annotation is completed, conduct a quality review. Automatically calculate the accuracy rate of the current annotation through an algorithm, depending on the feedback of the validation set, evaluate by comparing with the known annotations, classify the samples marked as difficult, and analyze the main reasons including unclear images, operation complexity, subjective biases of the annotators, etc. For difficult samples, form an expert group for secondary annotation confirmation.

[0059] Example 3, the third example of the present invention, which is different from the previous two examples:

[0060] If the above-mentioned functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs.

[0061] The logic and / or steps represented in the flowchart or described in other ways herein, for example, can be considered as a definite sequence list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other systems that can fetch instructions from the instruction execution system, apparatus, or device and execute the instructions), or in combination with these instruction execution systems, apparatuses, or devices. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transmit a program for use by or in combination with an instruction execution system, apparatus, or device.

[0062] More specific examples (non-exhaustive list) of computer-readable media include the following: electrical connection parts with one or more wirings (electronic devices), portable computer disk cartridges (magnetic devices), random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memories), fiber optic devices, and portable compact disc read-only memories (CDROM). Additionally, the computer-readable medium can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other media, then editing, interpreting, or otherwise processing it as appropriate, and then storing it in a computer memory.

[0063] 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.

[0064] Example 4, reference Figure 6 , which is the fourth embodiment of the present invention, and this embodiment provides a dangerous operation safety monitoring and early warning system, including a data processing module 100, a model training module 200, and a supervision module 300;

[0065] Data processing module 100, through employee self-learning, installs network servers and mobile AI cameras to monitor the work site in real time and collect data;

[0066] The model training module 200 screens and finely annotates the collected data, performs model training, and improves accuracy and recall through cross-validation;

[0067] The supervision module 300, through the application of AI cameras, establishes a sound supervision method for production site operations, identifies violations, issues warnings and takes snapshots, and sends any behaviors that are not corrected in time to supervisors.

[0068] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A dangerous operation safety monitoring and early warning method, characterized in that: include, Through employee self-learning, network servers and mobile AI cameras were installed to monitor the work site in real time and collect data; The collected data is screened and finely labeled, and model training is performed to improve accuracy and recall through cross-validation; Through the application of AI cameras, a sound production site operation supervision method is established to identify violations, issue warnings and take photos, and any behavior that is not corrected in time will be sent to supervisors.

2. A dangerous operation safety monitoring and early warning method as claimed in claim 1, characterized in that: The monitoring of the work site includes, through self-learning by employees, installing network servers and mobile AI cameras to monitor the work site and collect data.

3. A dangerous operation safety monitoring and early warning method as claimed in claim 2, characterized in that: The data collection includes collecting illumination data, weather data and working condition data; The collected data is screened, and the data finally retained includes different lighting data including day and night; different weather data including sunny days, cloudy days, and rainy days; different operating condition data including normal operating conditions and abnormal operating conditions.

4. A dangerous operation safety monitoring and early warning method as claimed in claim 3, characterized in that: The fine labeling includes fine labeling of the screened data using a preset labeling method to ensure the high quality of the training set, setting a labeling accuracy threshold and a threshold for the proportion of difficult samples in the extraction process through n cross-validations, extracting difficult samples generated in the labeling process, and performing secondary labeling confirmation by experts to generate a labeled data set.

5. A dangerous operation safety monitoring and early warning method as claimed in claim 4, characterized in that: The model training includes using a labeled data set to train the model, training on a set machine, setting the duration of each training, and specifying the accuracy, recall rate, and mAP0.5 standard on the validation set.

6. A dangerous operation safety monitoring and early warning method as claimed in claim 5, characterized in that: The model training also includes deploying the model in an actual industrial site for testing. During the test, the model's real-time video processing frame rate is required to reach 15fps, the delay time is required to be less than 500ms, and the accuracy of detection in actual operation is required to reach 95%.

7. A dangerous operation safety monitoring and early warning method as claimed in claim 6, characterized in that: The model training also includes, through a self-learning algorithm, continuously collecting new data for incremental training after the model is deployed, regularly updating the model every month and improving the recognition accuracy threshold of the model.

8. A system using a dangerous operation safety monitoring and early warning method as claimed in any one of claims 1 to 7, characterized in that: Includes data processing module, model training module, and supervision module; The data processing module, through the self-learning of employees, installs a network server and a mobile AI camera to monitor the work site in real time and collect data; The model training module screens and finely annotates the collected data, performs model training, and improves accuracy and recall through cross-validation; The supervision module, through the application of AI cameras, establishes a sound production site operation supervision method, determines illegal behaviors, and issues warnings and captures them. Behaviors that are not corrected in time are sent to supervisors.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of a dangerous operation safety monitoring and early warning method described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of a dangerous operation safety monitoring and early warning method described in any one of claims 1 to 7 are implemented.

Images