CN116778276A - Safe production model training method, application method, device, equipment and medium - Google Patents

Abstract

The invention discloses a safety production model training method, application method, device, equipment and medium. The safety production model includes multiple edge detection models and center detection models. During the actual detection process, the present invention compares the detection results of the edge detection model with the detection results of the center detection model. When the two results are inconsistent, the actual safety of the site is obtained. Based on the actual safety results on site, the edge detection model or the center detection model is model updated and trained, and the incremental learning method is used to continuously collect data on the edge detection model and the center detection model's misjudgment, and use it for iterative update of the model. , thereby achieving high accuracy of the edge detection model and high versatility of the center detection model, effectively solving the problem that the amount of data at a single point is insufficient to train a model with better performance, and improving the accuracy and robustness of the model, making Monitoring is more accurate and reliable.

Description

Safety production model training methods, application methods, devices, equipment and media

Technical field

The invention relates to computer technology, and in particular to a safety production model training method, application method, device, equipment and medium.

Background technique

With the continuous development of artificial intelligence, computer vision technology is increasingly used in the field of safety production. Among them, the monitoring of fire, smoke and other hazards, the monitoring of dangerous behaviors of personnel, and the monitoring of hazardous chemicals are all inseparable from the support of computer vision technology.

In order to achieve real-time monitoring, it is often necessary to rely on cameras to collect images and perform frame-by-frame inference on the scene. However, this requires that the model must be lightweight and take into account the usage requirements of different scenarios, so specific models need to be used for different camera points. However, the amount of data collected at a single point is not enough to train a good performing model.

Contents of the invention

The present invention provides a safety production model training method, application method, device, equipment and medium to solve the problem that the amount of data at a single point is insufficient to train a model with good performance, improve the accuracy and robustness of the model, and make Monitoring is more accurate and reliable.

In a first aspect, the present invention provides a safety production model training method. The safety production model includes multiple edge detection models and center detection models. The method includes:

Obtain images of the production site collected by image acquisition equipment;

The image is input into the corresponding edge detection model for processing, and the edge model detection result is obtained;

When the edge model detection result indicates that there is a security risk, the image is input into the center detection model for processing to obtain the center model detection result;

When the central model detection result indicates that there is a safety hazard, a safety alarm prompt is issued;

When the central model detection result shows that there is no safety hazard, obtain the actual safety results on site;

When the actual safety result of the site indicates that there is a safety hazard, use the image as a training sample to update and train the central detection model;

When the actual safety result of the scene is that there is no safety hazard, the image is used as a training sample to update and train the edge detection model.

Optionally, safety production model training methods also include:

When the edge model detection result is that there is no security risk, obtain the edge model detection result within a preset time period before the current time node;

When the edge model detection results within the preset time period are all indicating that there is no security risk, the determination threshold of the edge detection model is lowered.

Optional, detection results that pose safety risks include at least one of the following:

The object to be identified has illegal appearance information, the object to be identified has illegal behavior, the type of illegal vehicle, the vehicle has entered a prohibited area, the production equipment has abnormal status, there are fireworks, and there are leftovers.

Optionally, use the image as a training sample to update and train the center detection model, including:

When the actual safety result of the scene is that there is a safety hazard, the image is stored in the first database;

Determine whether the number of images in the first database reaches a first threshold;

If so, use the images in the first database as a training set to train the center detection model, and update the model parameters of the center detection model.

Optionally, use the image as a training sample to update and train the edge detection model, including:

When the actual safety result of the scene is that there is no safety hazard, the image is stored in the second database;

Determine whether the number of images in the second database reaches a second threshold;

If so, use the images in the second database as a training set to train the edge detection model, and update the model parameters of the edge detection model.

Optionally, the model architecture of the edge detection model is smaller than that of the center detection model.

In a second aspect, the present invention also provides a production safety detection method, which is applied to the production safety model trained by the production safety model training method provided in the first aspect of the invention.

In a third aspect, the present invention also provides a safety production model training device. The safety production model includes multiple edge detection models and center detection models. The device includes:

The image acquisition module is used to acquire images of the production site collected by the image acquisition equipment;

The first detection module is used to input the image into the corresponding edge detection model for processing, and obtain the edge model detection result;

The second detection module is used to input the image into the center detection model for processing when the edge model detection result indicates that there is a security risk, and obtain the center model detection result;

An alarm prompt module is used to issue a safety alarm prompt when the central model detection result indicates that there is a safety hazard;

The actual result acquisition module is used to obtain the actual safety results on site when the central model detection result is that there is no safety hazard;

The first update training module is used to use the image as a training sample to update and train the central detection model when the actual safety result of the site is that there is a safety hazard;

The second update training module is used to update and train the edge detection model using the image as a training sample when the actual safety result of the site is that there is no safety hazard.

In a fourth aspect, the present invention also provides an electronic device, including:

one or more processors;

Memory, used to store one or more programs;

When the one or more programs are executed by the one or more processors, the one or more processors are caused to implement the safety production model training method provided by the first aspect of the present invention.

In a fifth aspect, the present invention also provides a computer-readable storage medium. Computer-executable instructions are stored in the computer-readable storage medium. When the computer-executable instructions are executed by a processor, they are used to implement the first aspect of the present invention. Provided safety production model training methods.

The invention provides a safety production model training method. The safety production model includes multiple edge detection models and center detection models. The method includes: acquiring images of the production site collected by image acquisition equipment, and inputting the images into the corresponding edge detection models for processing. The edge model detection result is obtained. When the edge model detection result indicates that there is a safety hazard, the image is input into the center detection model for processing, and the central model detection result is obtained. When the central model detection result indicates that there is a safety hazard, a safety alarm prompt is issued. When the central model detection result is that there is no safety hazard, the actual safety result on site is obtained. When the actual safety result on site is that there is safety hazard, the image is used as a training sample to update and train the central detection model. The actual safety result on site is that there is no safety hazard. When hidden dangers are detected, images are used as training samples to update and train the edge detection model. During the actual detection process, the present invention compares the detection results of the edge detection model with the detection results of the center detection model. When the two results are inconsistent, the actual safety results on site are obtained, and the edge detection model or the center detection model is compared based on the actual safety results on site. The detection model is trained for model update, and the incremental learning method is used to continuously collect data on error judgments of the edge detection model and the center detection model, and is used for iterative update of the model, thereby achieving high accuracy of the edge detection model and high accuracy of the center detection model. Highly versatile, it effectively solves the problem that the amount of data at a single point is insufficient to train a model with good performance, improves the accuracy and robustness of the model, and makes monitoring more accurate and reliable.

It should be understood that what is described in this section is not intended to identify key or important features of the embodiments of the invention, nor is it intended to limit the scope of the invention. Other features of the present invention will become easily understood from the following description.

Description of drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings needed to be used in the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings without exerting creative efforts.

Figure 1 is a flow chart of a safety production model training method provided by an embodiment of the present invention;

Figure 2 is a schematic structural diagram of a safety production model training device provided by an embodiment of the present invention;

FIG. 3 is a schematic structural diagram of an electronic device according to an embodiment of the present invention.

Through the above-mentioned drawings, clear embodiments of the present application have been shown, which will be described in more detail below. These drawings and text descriptions are not intended to limit the scope of the present application's concepts in any way, but are intended to illustrate the application's concepts for those skilled in the art with reference to specific embodiments.

Detailed ways

In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only These are some 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 should fall within the scope of protection of the present invention.

It should be noted that the terms "first", "second", etc. in the description and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It is to be understood that the data so used are interchangeable under appropriate circumstances so that the embodiments of the invention described herein are capable of being practiced in sequences other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, e.g., a process, method, system, product, or apparatus that encompasses a series of steps or units and need not be limited to those explicitly listed. Those steps or elements may instead include other steps or elements not expressly listed or inherent to the process, method, product or apparatus.

Figure 1 is a flow chart of a production safety model training method provided by an embodiment of the present invention. This embodiment can be applied to the situation of incremental training of a production safety model. This method can be trained by the production safety model provided by an embodiment of the present invention. The device can be implemented by software and/or hardware, and is usually configured in electronic equipment. As shown in Figure 1, the safety production model training method specifically includes the following steps:

S101. Obtain the images of the production site collected by the image acquisition equipment.

In the embodiment of the present invention, image collection equipment is deployed at the production site to collect images of the production site and upload them to electronic devices. The image collection device may be a camera, and the collected images may be pictures or videos, which are not limited in this embodiment of the present invention.

S102. Input the image into the edge detection model for processing, and obtain the edge model detection result.

In the embodiment of the present invention, the safety production model includes multiple edge detection models and a central detection model. The edge detection models are deployed in corresponding edge devices, and the central detection model is deployed in the central server. Each edge detection model corresponds to the usage requirements of different scenarios.

After obtaining the images of the production site collected by the image acquisition equipment, the images are input into the corresponding edge detection model for processing, and the edge model detection results are obtained. The image processing process of the edge detection model may include preprocessing, image segmentation, feature extraction, feature fusion, feature space mapping, etc., which are not limited in this embodiment of the present invention. The edge model detection results indicate whether there are safety hazards at the current production site. The detection results of safety hazards include at least one of the following: the object to be identified has illegal appearance information, the object to be identified has illegal behavior, the type of illegal vehicle, and the vehicle has entered a prohibited area. , Production equipment has abnormal conditions, there are fireworks and leftovers, etc.

S103. When the edge model detection result indicates that there is a security risk, input the image into the center detection model for processing to obtain the center model detection result.

When the edge model detection result indicates that there is a security risk, the image is input into the center detection model for processing, and further judgment is made to obtain the center model detection result.

The image processing process of the center detection model may include preprocessing, image segmentation, feature extraction, feature fusion, feature space mapping, etc., which are not limited in this embodiment of the present invention. The central model detection results indicate whether there are safety hazards at the current production site. The detection results of safety hazards include at least one of the following: the object to be identified has illegal appearance information, the object to be identified has illegal behavior, the type of illegal vehicle, and the vehicle has entered a prohibited area. , Production equipment has abnormal conditions, there are fireworks and leftovers, etc.

When the edge model detection result is that there is no security risk, the edge model detection results within the preset time period (for example, 2 days) before the current time node are obtained. The edge model detection results within the preset time period are all when there is no security risk. , indicating that the judgment threshold of the edge detection model is set too high, causing the edge detection model to be unable to detect structures with potential safety hazards. At this time, the judgment threshold of the edge detection model should be lowered to improve the sensitivity of the edge detection model.

In some embodiments of the present invention, since the computing processing capabilities of edge devices are usually weak, in embodiments of the present invention, the model architecture of the edge detection model is smaller than that of the center detection model, which reduces the data processing pressure of the edge detection model and improves Detection efficiency of edge devices.

S104. When the central model detection result indicates that there is a safety hazard, a safety alarm prompt is issued.

In the embodiment of the present invention, the edge model detection result is compared with the center model detection result. If the edge model detection result indicates that there is a safety hazard, and the center model detection result is that there is a safety hazard, it is considered that there is indeed a safety hazard, and the alarm device is controlled. Issue a safety alarm prompt.

S105. When the central model detection result shows that there is no safety hazard, obtain the actual safety results on site.

When the central model detection result shows that there are no safety hazards, the actual safety results on site are obtained. For example, the actual on-site safety results can be determined by on-site personnel in person on-site and the results can be input into electronic devices.

S106. When the actual safety result on site indicates that there is a safety hazard, use the image as a training sample to update and train the central detection model.

When the actual safety result on site is that there is a safety hazard, it means that the central detection model has missed detection, and the image is used as a training sample to update the central detection model to improve the detection accuracy of the central detection model.

Exemplarily, in some embodiments of the present invention, when the actual safety result on site is that there is a safety hazard, the images are stored in the first database, and then it is determined whether the number of images in the first database reaches the first threshold, and if so , then use the images in the first database as a training set to train the center detection model, and update the model parameters of the center detection model. Specifically, the method of calculating the loss value can be used to determine whether the center detection model has converged. When the loss value is less than the preset value, it is determined that the center detection model has converged.

S107. When the actual safety result on site is that there is no safety hazard, use the image as a training sample to update and train the edge detection model.

When the actual safety result on site is that there is no safety hazard, it means that the edge detection model has misjudged. The image is used as a training sample to update the edge detection model to improve the detection accuracy of the edge detection model.

For example, in some embodiments of the present invention, when the actual safety result on site is that there is no safety hazard, the images are stored in the second database, and then it is determined whether the number of images in the second database reaches the second threshold, If so, use the images in the second database as a training set to train the edge detection model, and update the model parameters of the edge detection model. Specifically, the method of calculating the loss value can be used to determine whether the center detection model has converged. When the loss value is less than the preset value, it is determined that the center detection model has converged.

In the safety production model training method provided by the embodiment of the present invention, the safety production model includes multiple edge detection models and center detection models. The method includes: acquiring images of the production site collected by the image acquisition device, and inputting the images into the corresponding edge detection models. Process to obtain the edge model detection result. When the edge model detection result indicates that there is a safety hazard, the image is input into the center detection model for processing, and the central model detection result is obtained. When the center model detection result indicates that there is a safety hazard, a security alarm prompt is issued. , when the central model detection result is that there is no safety hazard, the actual safety result on site is obtained. When the actual safety result on site is that there is safety hazard, the image is used as a training sample to update and train the central detection model. When the actual safety result on site is no When there is a security risk, the image is used as a training sample to update the edge detection model. During the actual detection process, the present invention compares the detection results of the edge detection model with the detection results of the center detection model. When the two results are inconsistent, the actual safety results on site are obtained, and the edge detection model or the center detection model is compared based on the actual safety results on site. The detection model is trained for model update, and the incremental learning method is used to continuously collect data on error judgments of the edge detection model and the center detection model, and is used for iterative update of the model, thereby achieving high accuracy of the edge detection model and high accuracy of the center detection model. Highly versatile, it effectively solves the problem that the amount of data at a single point is insufficient to train a model with good performance, improves the accuracy and robustness of the model, and makes monitoring more accurate and reliable.

Embodiments of the present invention also provide a production safety detection method, which method is applied to the production safety model trained by the production safety model training method provided by any of the foregoing embodiments of the present invention.

Figure 2 is a schematic structural diagram of a production safety model training device provided by an embodiment of the present invention. The production safety model includes multiple edge detection models and center detection models. As shown in Figure 2, the production safety model training device includes:

The image acquisition module 201 is used to acquire images of the production site collected by the image acquisition device;

The first detection module 202 is used to input the image into the corresponding edge detection model for processing, and obtain the edge model detection result;

The second detection module 203 is used to input the image into the center detection model for processing when the edge model detection result indicates that there is a security risk, and obtain the center model detection result;

The alarm prompt module 204 is used to issue a safety alarm prompt when the central model detection result indicates that there is a safety hazard;

The actual result acquisition module 205 is used to obtain the actual safety results on site when the central model detection result is that there is no safety hazard;

The first update training module 206 is used to update and train the central detection model using the image as a training sample when the actual safety result of the site is that there is a safety hazard;

The second update training module 207 is used to update and train the edge detection model using the image as a training sample when the actual safety result of the site is that there is no safety hazard.

In some embodiments of the present invention, the safety production model training device further includes:

A detection result acquisition module, configured to acquire the edge model detection results within a preset time period before the current time node when the edge model detection result is that there is no security risk;

The threshold adjustment module is used to lower the judgment threshold of the edge detection model when the edge model detection results within the preset time period are all indicating that there is no safety hazard.

In some embodiments of the present invention, the detection results of potential safety hazards include at least one of the following:

The object to be identified has illegal appearance information, the object to be identified has illegal behavior, the type of illegal vehicle, the vehicle has entered a prohibited area, the production equipment has abnormal status, there are fireworks, and there are leftovers.

In some embodiments of the invention, the first update training module 206 includes:

A first storage unit configured to store the image in a first database when the actual safety result of the site is that there is a safety hazard;

A first judgment unit, used to judge whether the number of images in the first database reaches a first threshold;

A first update training unit, configured to use the images in the first database as a training set when the number of images in the first database reaches a first threshold, train the center detection model, and update the center Detect the model parameters of the model.

In some embodiments of the invention, the second update training module 207 includes:

A second storage unit, configured to store the image in a second database when the actual safety result of the site is that there is no safety hazard;

A second judgment unit, used to judge whether the number of images in the second database reaches a second threshold;

A second update training unit, configured to use the images in the second database as a training set when the number of images in the second database reaches a second threshold, train the edge detection model, and update the edge Detect the model parameters of the model.

Optionally, the model architecture of the edge detection model is smaller than that of the center detection model.

The above-mentioned safety production model training device can execute the safety production model training method provided by the previous embodiments of the present invention, and has corresponding functional modules and beneficial effects for executing the safety production model training method.

3 is a schematic structural diagram of an electronic device provided by an embodiment of the present invention. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workbenches, personal digital assistants, servers, and blades. servers, mainframe computers, and other suitable computers. Electronic devices may also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smart phones, wearable devices (eg, helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are examples only and are not intended to limit the implementation of the invention described and/or claimed herein.

As shown in Figure 3, the electronic device includes at least one processor 11, and a memory communicatively connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., wherein the memory stores A computer program executable by at least one processor. The processor 11 may execute according to a computer program stored in a read-only memory (ROM) 12 or loaded from a storage unit 18 into a random access memory (RAM) 13 Various appropriate actions and treatments. In the RAM 13, various programs and data required for the operation of the electronic device can also be stored. The processor 11, the ROM 12 and the RAM 13 are connected to each other via the bus 14. An input/output (I/O) interface 15 is also connected to bus 14 .

Multiple components in the electronic device are connected to the I/O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as magnetic disk, optical disk, etc. ; And communication unit 19, such as network card, modem, wireless communication transceiver, etc. The communication unit 19 allows the electronic device to exchange information/data with other devices through computer networks such as the Internet and/or various telecommunications networks.

Processor 11 may be a variety of general and/or special purpose processing components having processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, digital signal processing processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 11 executes various methods and processes described above, such as the safety production model training method.

In some embodiments, the safety production model training method may be implemented as a computer program, which is tangibly included in a computer-readable storage medium, such as the storage unit 18 . In some embodiments, part or all of the computer program may be loaded and/or installed onto the electronic device via the ROM 12 and/or the communication unit 19 . When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the safety production model training method described above may be performed. Alternatively, in other embodiments, the processor 11 may be configured to perform the safe production model training method in any other suitable manner (eg, by means of firmware).

Various implementations of the systems and techniques described above may be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on a chip implemented in a system (SOC), load programmable logic device (CPLD), computer hardware, firmware, software, and/or a combination thereof. These various embodiments may include implementation in one or more computer programs executable and/or interpreted on a programmable system including at least one programmable processor, the programmable processor The processor, which may be a special purpose or general purpose programmable processor, may receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device. An output device.

Computer programs for implementing the methods of the invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that the computer program, when executed by the processor, causes the functions/operations specified in the flowcharts and/or block diagrams to be implemented. A computer program may execute entirely on the machine, partly on the machine, as a stand-alone software package, partly on the machine and partly on a remote machine or entirely on the remote machine or server.

In the context of this invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in connection with an instruction execution system, apparatus, or device. Computer-readable storage media may include, but are not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or devices, or any suitable combination of the foregoing. Alternatively, the computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media would include one or more wire-based electrical connections, laptop disks, hard drives, random access memory (RAM), read only memory (ROM), erasable programmable read only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination of the above.

To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having a display device (eg, a CRT (cathode ray tube) or LCD (liquid crystal display)) for displaying information to the user monitor); and a keyboard and pointing device (e.g., a mouse or a trackball) through which a user can provide input to the electronic device. Other kinds of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and may be provided in any form, including Acoustic input, voice input or tactile input) to receive input from the user.

The systems and techniques described herein may be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., A user's computer having a graphical user interface or web browser through which the user can interact with implementations of the systems and technologies described herein), or including such backend components, middleware components, or any combination of front-end components in a computing system. The components of the system may be interconnected by any form or medium of digital data communication (eg, a communications network). Examples of communication networks include: local area network (LAN), wide area network (WAN), blockchain network, and the Internet.

Computing systems may include clients and servers. Clients and servers are generally remote from each other and typically interact over a communications network. The relationship of client and server is created by computer programs running on corresponding computers and having a client-server relationship with each other. The server can be a cloud server, also known as cloud computing server or cloud host. It is a host product in the cloud computing service system to solve the problems of difficult management and weak business scalability in traditional physical hosts and VPS services. defect.

Embodiments of the present invention also provide a computer program product, including a computer program that, when executed by a processor, implements the safety production model training method provided by any embodiment of the present application.

During the implementation of the computer program product, computer program code for performing the operations of the present invention can be written in one or more programming languages or a combination thereof. Programming languages include object-oriented programming languages, such as Java, Smalltalk , C++, and also includes conventional procedural programming languages, such as the "C" language or similar programming languages. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In situations involving remote computers, the remote computer can be connected to the user's computer through any kind of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computer (such as an Internet service provider through Internet connection).

It should be understood that various forms of the process shown above may be used, with steps reordered, added or deleted. For example, each step described in the present invention can be executed in parallel, sequentially, or in different orders. As long as the desired results of the technical solution of the present invention can be achieved, there is no limitation here.

The above-mentioned specific embodiments do not constitute a limitation on the scope of the present invention. It will be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions are possible depending on design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention shall be included in the protection scope of the present invention.

Claims (10)

1. A method of training a safe production model, the safe production model comprising a plurality of edge detection models and a center detection model, the method comprising:

acquiring an image of a production site acquired by image acquisition equipment;

inputting the image into a corresponding edge detection model for processing to obtain an edge model detection result;

when the edge model detection result is that potential safety hazards exist, inputting the image into a center detection model for processing to obtain a center model detection result;

when the center model detection result shows that potential safety hazards exist, a safety alarm prompt is sent out;

when the center model detection result is that no potential safety hazard exists, acquiring an on-site actual safety result;

when the actual safety result of the site is that potential safety hazards exist, the image is used as a training sample to update and train the center detection model;

and when the actual safety result on the site is that no potential safety hazard exists, updating and training the edge detection model by taking the image as a training sample.

2. The safe production model training method of claim 1, further comprising:

when the edge model detection result is that no potential safety hazard exists, acquiring an edge model detection result in a preset time period before the current time node;

and when the edge model detection results within the preset time period are all potential safety hazards, reducing the judgment threshold value of the edge detection model.

3. The method of claim 1, wherein the detection of the presence of a potential safety hazard comprises at least one of:

the method comprises the steps of enabling an object to be identified to have illegal appearance information, enabling the object to be identified to have illegal behaviors, enabling the type of an illegal vehicle to be changed, enabling the vehicle to drive into a forbidden area, enabling production equipment to have abnormal states, enabling smoke and fire to be present and enabling the object to be identified to have legacy.

4. A method of training a safe production model according to any one of claims 1 to 3, wherein updating the central detection model using the image as a training sample comprises:

when the actual safety result of the site is that potential safety hazards exist, storing the image into a first database;

judging whether the number of the images in the first database reaches a first threshold value or not;

if yes, training the center detection model by taking the images in the first database as a training set, and updating model parameters of the center detection model.

5. A method of training a safe production model according to any one of claims 1 to 3, wherein updating the edge detection model using the image as a training sample comprises:

when the actual safety result of the site is that no potential safety hazard exists, storing the image into a second database;

judging whether the number of images in the second database reaches a second threshold value or not;

if yes, the image in the second database is used as a training set, the edge detection model is trained, and model parameters of the edge detection model are updated.

6. A method of training a safety production model according to any of claims 1-3, wherein the model architecture of the edge detection model is smaller than the center detection model.

7. A safety production detection method, characterized by being applied to a safety production model trained by the safety production model training method according to any one of claims 1 to 6.

8. A safety production model training apparatus, wherein the safety production model comprises a plurality of edge detection models and a center detection model, the apparatus comprising:

the image acquisition module is used for acquiring the image of the production site acquired by the image acquisition equipment;

the first detection module is used for inputting the image into a corresponding edge detection model for processing to obtain an edge model detection result;

the second detection module is used for inputting the image into a center detection model for processing when the edge model detection result is that the potential safety hazard exists, so as to obtain a center model detection result;

the alarm prompt module is used for sending out a safety alarm prompt when the detection result of the central model is that the potential safety hazard exists;

the actual result acquisition module is used for acquiring an on-site actual safety result when the center model detection result is that the potential safety hazard does not exist;

the first updating training module is used for updating and training the center detection model by taking the image as a training sample when the on-site actual safety result is that potential safety hazards exist;

and the second updating training module is used for updating and training the edge detection model by taking the image as a training sample when the on-site actual safety result is that no potential safety hazard exists.

9. An electronic device, comprising:

one or more processors;

a memory for storing one or more programs;

the one or more programs, when executed by the one or more processors, cause the one or more processors to implement the safe production model training method of any of claims 1-6.

10. A computer readable storage medium having stored therein computer executable instructions which when executed by a processor are for implementing the safe production model training method of any of claims 1-6.