An ai intelligent identification method and identification system for industrial safety

Abstract

The application relates to an AI intelligent identification method and identification system for industrial safety, which comprises the following steps: identifying objects in images based on acquired image data, the objects including personnel, equipment and environmental factors; determining the states of the objects, the states including moving states and static states; simulating the moving tracks of the objects in the moving states to obtain predicted moving tracks; calculating the interference conditions of any two predicted moving tracks and the interference conditions between the predicted moving tracks and the objects in the static states; screening the interference conditions; and issuing a warning according to the screening. The AI intelligent identification method and identification system for industrial safety disclosed by the application can discover potential dangerous factors in a construction site and issue a warning in a manner of object classification and pre-simulation, so that the effectiveness of supervision means can be improved, and safety accidents can be avoided.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to an AI-powered intelligent identification method and system for industrial safety. Background Technology

[0002] Industrial safety involves multiple aspects and is directly related to personal and property safety. Current approaches often involve pre-job training and case studies to instill basic concepts, followed by necessary oversight during the work process. However, the limitations of these oversight methods must be considered, as they are influenced by factors such as the number of personnel, their qualifications, and the cost of supervision.

[0003] With the development of AI technology, it has become possible to achieve supervision through AI intelligent recognition. Taking construction sites as an example, the hazards caused by vehicle movement, personnel movement and site changes can be identified using AI intelligent recognition. However, how to identify and discover hazards or potential hazards still requires further research. Summary of the Invention

[0004] This application provides an AI-powered intelligent identification method and system for industrial safety. By classifying objects and conducting pre-simulation, it identifies potential hazards in construction sites and provides early warnings, thereby improving the effectiveness of supervision and preventing safety accidents.

[0005] The above-mentioned objective of this application is achieved through the following technical solution: Firstly, this application provides an AI-powered intelligent identification method for industrial safety, comprising: Based on the acquired image data, identify objects in the images, including people, equipment, and environmental factors; Determine the state of the object, which includes moving state and stationary state; Simulate the movement trajectory of an object in a moving state to obtain the predicted movement trajectory; Calculate the interference between any two predicted trajectories and the interference between the predicted trajectories and an object in a stationary state; Interference scenarios are screened and warnings are issued based on the screening results.

[0006] In one possible implementation of the first aspect, identifying objects in an image includes: Create multiple outlines in the image data based on color differences. The image data is decomposed using contours to obtain multiple potential objects; The suspected object is divided into multiple sub-suspected objects, and the sub-suspected objects are compared with the objects stored in the model library to determine the type of the sub-suspected objects; The type of the suspected object is determined based on the attribution type of all the suspected sub-objects, thus obtaining the object.

[0007] In one possible implementation of the first aspect, creating contours in image data based on color differences includes: Image data is processed by sending it into different color channels to obtain monochrome image data. The color channels include the red channel, green channel and blue channel. Generate the frequency curve for each row in the monochrome image data; Obtain the abrupt change regions of the frequency curve and mark the abrupt change regions on the image data; Create outlines using abruptly changing regions; Each abrupt change region is located on the created contour.

[0008] In one possible implementation of the first aspect, the abrupt change region of the frequency curve includes: Identify the amplitude abrupt change point on the frequency curve and construct an amplitude abrupt change reference region based on the amplitude abrupt change point; The frequency curve corresponding to the reference region with abrupt amplitude change is smoothed to obtain the reference curve; A moving screening wave is generated, which has a fixed frequency and a fixed attenuation rate; The moving screening wave is driven to move horizontally. The sudden change region of the frequency curve is obtained by superimposing the moving screening wave and the frequency curve. The length of the sudden change region is equal to the length of the moving screening wave.

[0009] In one possible implementation of the first aspect, generating the moving screening wave includes: The image data is processed into grayscale to obtain the outline; The reference frequency and reference attenuation rate of the moving screening wave are determined based on the corresponding position of the moving screening wave on the contour. Calculate the superposition of the moving screening wave at corresponding positions on the contour to obtain the superposition amount; Adjust the reference frequency and reference attenuation rate of the moving screening wave according to the superposition amount to maximize the superposition amount; The fixed frequency and fixed attenuation rate of the moving screening wave are determined based on the result of maximizing the superposition amount, and the moving screening wave is generated using the fixed frequency and fixed attenuation rate.

[0010] In one possible implementation of the first aspect, obtaining the predicted movement trajectory includes: Obtain the current movement trajectory of an object in the moving state; The existing movement trajectory is segmented to obtain the existing movement trajectory segments; Determine the direction vector of the end of the existing trajectory segment; By statistically analyzing the changing trend of the direction vector in a time series, the pointing direction or pointing region can be obtained. The direction or area pointed to is used as the predicted movement trajectory.

[0011] In one possible implementation of the first aspect, the changing trend of the statistical direction vector includes: Calculate the similarity between two adjacent directional vectors in a time series. The similarity includes horizontal similarity and vertical similarity. The similarity direction is determined based on the ratio of horizontal similarity to vertical similarity; When similar directions are the same, the pointing direction is obtained; when similar directions are different, the pointing region is obtained.

[0012] Secondly, this application provides an AI-powered intelligent identification device for industrial safety, comprising: The recognition unit is used to identify objects in an image based on the acquired image data. Objects include people, equipment, and environmental factors. The state determination unit is used to determine the state of an object, which includes moving state and stationary state; The simulation unit is used to simulate the movement trajectory of an object in a moving state to obtain the predicted movement trajectory. The calculation unit is used to calculate the interference between any two predicted moving trajectories and the interference between the predicted moving trajectory and an object in a stationary state. The early warning unit is used to screen interference situations and issue early warnings based on the screening results.

[0013] Thirdly, this application provides an AI-powered intelligent identification system for industrial safety, the system comprising: One or more memories for storing instructions; and One or more processors are configured to call and execute the instructions from the memory to perform the methods described in the first aspect and any possible implementation thereof.

[0014] Fourthly, this application provides a computer-readable storage medium, the computer-readable storage medium comprising: The program, when run by a processor, is executed as described in the first aspect and any possible implementation thereof.

[0015] Fifthly, this application provides a computer program product, including program instructions that, when run by a computing device, execute the method described in the first aspect and any possible implementation thereof.

[0016] Sixthly, this application provides a chip system including a processor for implementing the functions involved in the foregoing aspects, such as generating, receiving, transmitting, or processing the data and / or information involved in the foregoing methods.

[0017] This chip system can consist of chips or include chips and other discrete components.

[0018] In one possible design, the chip system also includes a memory for storing necessary program instructions and data. The processor and the memory can be decoupled and located on different devices, connected via wired or wireless means, or the processor and the memory can be coupled to the same device. Attached Figure Description

[0019] Figure 1 This is a flowchart illustrating the steps of an AI-powered intelligent recognition method for industrial safety provided in this application.

[0020] Figure 2 This is a schematic diagram of using contours to decompose image data, as provided in this application.

[0021] Figure 3 This is a schematic diagram of a suspected object being divided into multiple sub-suspected objects, as provided in this application.

[0022] Figure 4 This is a schematic diagram of a frequency curve provided in this application.

[0023] Figure 5 This is a schematic diagram of segmenting an existing movement trajectory, as provided in this application.

[0024] Figure 6 This is a schematic diagram showing the consistent trend of change in a direction vector provided in this application.

[0025] Figure 7 This is a schematic diagram illustrating the discrete nature of the changing trend of a direction vector provided in this application. Detailed Implementation

[0026] The technical solutions in this application will be further described in detail below with reference to the accompanying drawings.

[0027] This application discloses an AI-powered intelligent recognition method for industrial safety. For some examples, please refer to [link / reference needed]. Figure 1 This application discloses an AI-powered intelligent recognition method for industrial safety, comprising the following steps: S101, Based on the acquired image data, identify objects in the image, including people, equipment, and environmental factors; S102, Determine the state of the object, including moving state and stationary state; S103, Simulate the movement trajectory of an object in a moving state to obtain the predicted movement trajectory; S104, calculate the interference between any two predicted trajectories and the interference between the predicted trajectories and an object in a stationary state; S105, screen for interference situations and issue early warnings based on the screening.

[0028] The AI-powered intelligent recognition method for industrial safety disclosed in this application is applied to a server. The server acquires images through image acquisition terminals deployed at the construction site, analyzes the content in the images, and then provides the analysis results. When the analysis results show that there are dangerous factors, an early warning will be issued simultaneously.

[0029] Specifically, in step S101, objects in the image are first identified based on the acquired image data. These objects include people, equipment, and environmental factors.

[0030] People refer to the people in the image, equipment refers to the equipment in the image, and environmental factors refer to other objects in the image besides people and equipment that can directly have a negative impact on people and equipment.

[0031] For example, dents in the ground can cause injury to people or equipment falling in; similarly, collisions between people and vehicles can also result in injury. It should be specifically noted that the technical solution disclosed in this application is applied to intelligent ground recognition at construction sites, primarily ensuring the safety of personnel and equipment movement.

[0032] In step S102, the state of the object is determined. The state includes a moving state and a stationary state. The purpose of determining the state of the object is to understand whether the object will move. For example, for stationary people and equipment, it is only necessary to analyze whether other people or equipment will have a negative impact on it, rather than analyzing whether it will have a negative impact on other people or equipment.

[0033] In step S103, the movement trajectory of the object in the moving state is simulated to obtain the predicted movement trajectory. Then, the interference between any two predicted movement trajectories and the interference between the predicted movement trajectory and the object in the stationary state are calculated, which is the content of step S104.

[0034] Finally, in step S105, the interference situation is filtered and an early warning is issued based on the filtering. The interference situation includes two types: interference between any two predicted moving trajectories and interference between the predicted moving trajectory and an object in a stationary state.

[0035] The filtering method for any two predicted movement trajectories interfering with each other is to determine the types of the two objects involved. If they are two people, the interference is excluded. If the predicted movement trajectory interferes with an object that is stationary, all of them are retained.

[0036] Meanwhile, length is used to limit the predicted movement trajectory, for example, the length is limited to 5 meters. The premise of the length limit is to allow sufficient reaction time for personnel and equipment. Taking the vehicle in the equipment as an example, the driver's reaction time is specified as three seconds, then the length of the predicted movement trajectory is equal to the product of three seconds and the vehicle's speed.

[0037] For environmental factors, the general approach is to treat everything except the ground, people, and equipment as environmental factors. The identification method for environmental factors is the same as that for people and equipment, which will be described later.

[0038] In some cases, the specific way to identify objects in an image is as follows: S201, Create contours in the image data based on color differences, with multiple contours; S202, use contours to decompose the image data to obtain multiple suspected objects; S203, the suspected object is divided into multiple sub-suspected objects and the sub-suspected objects are compared with the objects stored in the model library to determine the type of the sub-suspected objects; S204, determine the type of the suspected object based on the belonging type of all the suspected sub-objects, and obtain the object.

[0039] Please see Figure 2 In steps S201 to S204, color difference is used to create contours in the image data, and then the image data is decomposed using the contours to obtain multiple suspected objects. At this point, it can be known which objects are included in the image data, but the type of the objects cannot be determined. When one suspected object is located inside another suspected object, identification is also required. If the type of the sub-suspected object can be determined, the two sub-suspected objects are processed separately. If the type of the sub-suspected object cannot be determined, the two sub-suspected objects are treated as a single suspected object.

[0040] Please see Figure 3 The method for determining this is to divide the suspected object into multiple sub-suspected objects and compare the sub-suspected objects with the objects stored in the model library to determine the type of the sub-suspected objects.

[0041] Finally, the type of the suspected object is determined based on the attribution type of the sub-suspected objects. Specifically, the attribution type of the sub-suspected objects is counted. At this time, there may be one or more attribution types.

[0042] When there is only one attribution type, the type of the suspected object can be determined directly; when there are multiple attribution types, the objects are sorted according to the number of sub-suspected objects corresponding to each attribution type, and the number of sub-suspected objects in the sequential sequence tends to decrease.

[0043] Then, the first belonging type in the sequence is taken as the type of the suspected object.

[0044] In some examples, the specific way to create contours in image data based on color differences is as follows: S301, the image data is sent to different color channels for processing to obtain monochrome image data. The color channels include the red channel, green channel and blue channel. S302, Generate the frequency curve for each row in the monochrome image data; S303, obtain the abrupt change region of the frequency curve and mark the abrupt change region on the image data; S304, Use abrupt change regions to create contours; Each abrupt change region is located on the created contour.

[0045] In steps S301 to S304, different color channels are used to process the image data in order to obtain a more accurate contour. This is because in the application scenario, the contour is directly related to the accuracy of the subsequent determination of the suspected object type, and it can even be used to directly determine the type of the suspected object.

[0046] For a single point in image data, it is composed of red, green, and blue, with each color having a different proportion. If only grayscale is used for processing, it means that some color information will be lost.

[0047] For example, if two points in an image have similar gray values, it is difficult to distinguish the boundary. However, if the constituent colors (the proportions of red, green, and blue) of these two points are different, then using monochrome image data will allow us to identify the difference and obtain a more accurate outline.

[0048] For monochrome images, frequency curves for each row of the monochrome image data will be generated, such as... Figure 4 As shown, the horizontal axis of the coordinate system represents distance, indicating the distance between a point in the image data and the zero coordinate. The vertical axis represents the numerical value corresponding to the color of that point. Then, the abrupt change regions of the frequency curve are obtained and marked on the image data; these abrupt change regions form the contour.

[0049] The specific method for obtaining the abrupt change region of the frequency curve is as follows: S401, Determine the amplitude abrupt change point on the frequency curve and construct an amplitude abrupt change reference region based on the amplitude abrupt change point; S402, smooth the frequency curve corresponding to the reference region with a sudden change in amplitude to obtain the reference curve; S403 generates a moving screening wave with a fixed frequency and a fixed attenuation rate; S404 drives the moving screening wave to move horizontally. The sudden change region of the frequency curve is obtained by superimposing the moving screening wave and the frequency curve. The length of the sudden change region is equal to the length of the moving screening wave.

[0050] In steps S401 to S404, the amplitude abrupt change point on the frequency curve is first determined. The amplitude abrupt change point refers to the point on the frequency curve where the slope changes by a greater than the set allowable value, or the slope changes abruptly. Then, based on the amplitude abrupt change point, an amplitude abrupt change reference region is constructed. The length of the amplitude abrupt change region is generally three to five points.

[0051] Next, the frequency curve corresponding to the reference region with a sudden change in amplitude is smoothed to obtain the reference curve. The purpose of the smoothing is to initially remove the interference that occurred during the image acquisition process. The smoothing here is limited to the non-differentiable points on the reference curve, and the range of the smoothing is limited to the area of ​​the three to five pixels before and the three to five pixels after the non-differentiable points.

[0052] The amplitude change point and the change reference area are preliminary positioning, but a moving screening wave is still needed for precise positioning.

[0053] Then a moving screening wave is generated, which has a fixed frequency and a fixed attenuation rate.

[0054] After generating the moving screening wave, it is driven to move horizontally. The abrupt change region of the frequency curve is obtained by superimposing the moving screening wave and the frequency curve; the length of the abrupt change region is equal to the length of the moving screening wave. The function of the moving screening wave is to further determine the abrupt change region on the frequency curve.

[0055] The frequency of the moving screening wave is initially determined based on the frequency in the reference region of the amplitude abrupt change, and then adjusted by increasing and decreasing.

[0056] In this application, it is necessary to remove the low-frequency components in the image data in order to obtain a more defined outline.

[0057] A fixed attenuation rate refers to the continuous decrease in the amplitude of the moving screening wave. For the selection of the fixed attenuation rate, when initially driving the moving screening wave horizontally, it is only necessary to ensure that the fixed attenuation rate is less than one (randomly assigned). Simultaneously, a fixed value will be initially given for the amplitude of the moving screening wave, and then adjusted based on the superposition of the moving screening wave and the frequency curve.

[0058] The specific process of generating the moving screening wave is as follows: S501, perform grayscale processing on the image data to obtain the outline; S502, determine the reference frequency and reference attenuation rate of the moving screening wave based on the corresponding position of the moving screening wave on the contour; S503, calculate the superposition of the moving screening wave at the corresponding position on the contour to obtain the superposition amount; S504, adjust the reference frequency and reference attenuation rate of the moving screening wave according to the superposition amount to maximize the superposition amount; S505 determines the fixed frequency and fixed attenuation rate of the moving screening wave based on the result of maximizing the superposition amount, and generates the moving screening wave using the fixed frequency and fixed attenuation rate.

[0059] Specifically, the image data is first processed into grayscale to obtain a contour that can be used for reference. Then, the reference frequency and reference attenuation rate of the moving screening wave are determined according to the corresponding position of the moving screening wave on the contour. The purpose of this method is to quickly determine the frequency and attenuation value of the moving screening wave.

[0060] Next, the reference frequency and reference attenuation rate of the moving screening wave are adjusted according to the superposition amount to maximize the superposition amount. The superposition amount refers to the change in the frequency curve after the moving screening wave is assigned a frequency curve; the higher the similarity between the two, the greater the change in the frequency curve. Maximizing the superposition amount is an iterative process, which involves repeatedly calculating to determine the reference frequency and reference attenuation rate.

[0061] Finally, based on the result of maximizing the superposition, the fixed frequency and fixed attenuation rate of the moving screening wave are determined, and the moving screening wave is generated using the fixed frequency and fixed attenuation rate.

[0062] In some cases, the specific way to obtain the predicted movement trajectory is as follows: S601, obtain the existing movement trajectory of the object in the moving state; S602, the existing movement trajectory is segmented to obtain the existing movement trajectory segment; S603, determine the direction vector of the end of the existing moving trajectory segment; S604: Statistically analyze the changing trend of the direction vector in the time series to obtain the pointing direction or pointing region; S605 uses the direction of the pointing or the area corresponding to the pointing area as the predicted movement trajectory.

[0063] In steps S601 to S605, it is first necessary to obtain the existing movement trajectory of the object in the moving state, and then to segment the existing movement trajectory to obtain the existing movement trajectory segments. Figure 5 As shown, the existing movement trajectory segments are used to analyze the trend of the existing movement trajectory.

[0064] Next, the direction vector at the end of the existing trajectory segment is determined, and then the trend of the change of the direction vector is statistically analyzed on the time series. The trend of the change of the direction vector is consistent. Figure 6 (as shown) and discreteness ( Figure 7 As shown in the figure, there are two results: the direction of the direction can be obtained through the consistency result, and the region of the direction can be obtained through the discrete result.

[0065] Finally, the direction of the pointing or the area corresponding to the pointing region is used as the predicted movement trajectory.

[0066] The specific method for statistically analyzing the changing trend of the direction vector is as follows: Calculate the similarity between two adjacent directional vectors in a time series. The similarity includes horizontal similarity and vertical similarity. The similarity direction is determined based on the ratio of horizontal similarity to vertical similarity; When similar directions are the same, the pointing direction is obtained; when similar directions are different, the pointing region is obtained.

[0067] In this approach, horizontal and vertical similarity are used to evaluate the similarity between two directional vectors. A coordinate system is needed for this description. A directional vector can be divided into two reference vectors with fixed directions (a first reference vector and a second reference vector). After normalization, the reference vectors have fixed lengths. Therefore, the directional vector can be expressed as: Direction vector = M * first reference vector + N * second reference vector; When evaluating the similarity between two directional vectors, M and N are used for evaluation. If M is always greater than N, it means that the similar directions are the same, and the pointing direction is obtained. If there is a part where M is greater than N and the rest where M is less than N, the pointing region is obtained.

[0068] The pointing region includes multiple pointing directions, and is the largest area covered by these multiple pointing directions. For the existing movement trajectory of an object, the time window length for its value is generally 3-5 seconds; for the predicted movement trajectory, the time window length for its value is also about 3-5 seconds.

[0069] This application also provides an AI-powered intelligent identification device for industrial safety, comprising: The recognition unit is used to identify objects in an image based on the acquired image data. Objects include people, equipment, and environmental factors. The state determination unit is used to determine the state of an object, which includes moving state and stationary state; The simulation unit is used to simulate the movement trajectory of an object in a moving state to obtain the predicted movement trajectory. The calculation unit is used to calculate the interference between any two predicted moving trajectories and the interference between the predicted moving trajectory and an object in a stationary state. The early warning unit is used to screen interference situations and issue early warnings based on the screening results.

[0070] Furthermore, it also includes: The first contour creation unit is used to create contours in the image data based on color differences; there are multiple contours. The first image decomposition unit is used to decompose the image data using contours to obtain multiple suspected objects; The judgment and processing unit is used to cut the suspected object into multiple sub-suspected objects and compare the sub-suspected objects with the objects stored in the model library to determine the type of the sub-suspected objects; The type determination unit is used to determine the type of the suspected object based on the belonging type of all the suspected sub-objects, and thus obtain the object.

[0071] Furthermore, it also includes: The image decomposition unit is used to send image data into different color channels for processing to obtain monochrome image data. The color channels include the red channel, green channel and blue channel. The frequency curve generation unit is used to generate the frequency curve for each row in the monochrome image data. The region determination unit is used to obtain the abrupt change regions of the frequency curve and mark the abrupt change regions on the image data; The second contour creation unit is used to create contours using abruptly changing regions. Each abrupt change region is located on the created contour.

[0072] Furthermore, it also includes: The position determination unit is used to determine the amplitude abrupt change point on the frequency curve and construct the amplitude abrupt change reference region based on the amplitude abrupt change point; The smoothing unit is used to smooth the frequency curve corresponding to the reference region with a sudden change in amplitude to obtain the reference curve; A generation unit is used to generate a moving screening wave, which has a fixed frequency and a fixed attenuation rate. The moving processing unit is used to drive the moving screening wave to move in the horizontal direction, and obtain the abrupt change region of the frequency curve based on the superposition of the moving screening wave and the frequency curve. The length of the abrupt change region is equal to the length of the moving screening wave.

[0073] Furthermore, it also includes: The grayscale processing unit is used to process image data in grayscale to obtain contours; The first parameter determination unit is used to determine the reference frequency and reference attenuation rate of the moving screening wave based on the corresponding position of the moving screening wave on the contour. The first calculation unit is used to calculate the superposition of the moving screening wave at the corresponding position on the contour to obtain the superposition amount; The second calculation unit is used to adjust the reference frequency and reference attenuation rate of the moving screening wave according to the superposition amount, so as to maximize the superposition amount. The second parameter determination unit is used to determine the fixed frequency and fixed attenuation rate of the moving screening wave based on the result of maximizing the superposition amount, and to generate the moving screening wave using the fixed frequency and fixed attenuation rate.

[0074] Furthermore, it also includes: The movement trajectory acquisition unit is used to obtain the existing movement trajectory of an object in a moving state; The movement trajectory processing unit segments the existing movement trajectory to obtain existing movement trajectory segments; The end-processing unit determines the direction vector of the end of the existing movement trajectory segment; The trend analysis unit is used to statistically analyze the changing trend of direction vectors over a time series to obtain the pointing direction or pointing area. The result unit is used to take the area corresponding to the pointing direction or the pointing area as the predicted movement trajectory.

[0075] Furthermore, it also includes: The third calculation unit is used to calculate the similarity between two adjacent directional vectors in the time series. The similarity includes horizontal similarity and vertical similarity. The similarity direction determination unit is used to determine the similarity direction based on the ratio of horizontal similarity to vertical similarity. When similar directions are the same, the pointing direction is obtained; when similar directions are different, the pointing region is obtained.

[0076] In one example, the unit in any of the above devices may be one or more integrated circuits configured to implement the above methods, such as one or more application-specific integrated circuits (ASICs), or one or more digital signal processors (DSPs), or one or more field-programmable gate arrays (FPGAs), or a combination of at least two of these integrated circuit forms.

[0077] For example, when the units in the device can be implemented through a processing element scheduler, the processing element can be a general-purpose processor, such as a central processing unit (CPU) or other processor capable of calling programs. Alternatively, these units can be integrated together to form a system-on-a-chip (SOC).

[0078] In this application, various objects such as messages / information / devices / network elements / systems / apparatus / actions / operations / processes / concepts may be named. It is understood that these specific names do not constitute a limitation on the relevant objects. The names may be changed depending on the scenario, context, or usage habits. The understanding of the technical meaning of the technical terms in this application should be mainly determined from their functions and technical effects embodied / performed in the technical solution.

[0079] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0080] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0081] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0082] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0083] It should also be understood that in the various embodiments of this application, the terms "first," "second," etc., are merely to indicate that multiple objects are different. For example, a first time window and a second time window are only to indicate different time windows. They should not have any effect on the time windows themselves, and the aforementioned terms "first," "second," etc., should not impose any limitations on the embodiments of this application.

[0084] It should also be understood that, in the various embodiments of this application, unless otherwise specified or in case of logical conflict, the terms and / or descriptions between different embodiments are consistent and can be referenced by each other, and the technical features in different embodiments can be combined to form new embodiments according to their inherent logical relationships.

[0085] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a computer-readable storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned computer-readable storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0086] This application also provides an AI-powered intelligent identification system for industrial safety, the system comprising: One or more memories for storing instructions; and One or more processors are configured to retrieve and execute the instructions from the memory, performing the methods described above.

[0087] This application also provides a computer program product including instructions that, when executed, cause the AI ​​intelligent recognition system to perform operations corresponding to the above-described method.

[0088] This application also provides a chip system including a processor for implementing the functions involved in the above description, such as generating, receiving, transmitting, or processing the data and / or information involved in the above methods.

[0089] This chip system can consist of chips or include chips and other discrete components.

[0090] The processor mentioned above can be a CPU, a microprocessor, an ASIC, or one or more integrated circuits that execute a program to control the method of transmitting the feedback information described above.

[0091] In one possible design, the chip system also includes a memory for storing necessary program instructions and data. The processor and the memory can be decoupled and located on different devices, connected via wired or wireless means to support the chip system in implementing the various functions described in the above embodiments. Alternatively, the processor and the memory can also be coupled to the same device.

[0092] Optionally, the computer instructions are stored in memory.

[0093] Optionally, the memory can be a storage unit within the chip, such as a register or cache. Alternatively, the memory can be a storage unit located outside the chip within the terminal, such as a ROM or other types of static storage devices that can store static information and instructions, such as RAM.

[0094] It is understood that the memory in this application may be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory.

[0095] Non-volatile memory can be ROM, programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory.

[0096] Volatile memory can be RAM, which is used as an external cache. There are many different types of RAM, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM), and direct memory bus random access memory.

[0097] The embodiments described in this specific implementation are preferred embodiments of this application and are not intended to limit the scope of protection of this application. Therefore, all equivalent changes made in accordance with the structure, shape and principle of this application should be covered within the scope of protection of this application.

Claims

1. An AI-powered intelligent recognition method for industrial safety, characterized in that, include: Based on the acquired image data, identify objects in the images, including people, equipment, and environmental factors; Determine the state of the object, which includes moving state and stationary state; Simulate the movement trajectory of an object in a moving state to obtain the predicted movement trajectory; Calculate the interference between any two predicted trajectories and the interference between the predicted trajectories and an object in a stationary state; Interference scenarios are screened and warnings are issued based on the screening results; Objects identified in the image include: Create multiple outlines in the image data based on color differences. The image data is decomposed using contours to obtain multiple potential objects; The suspected object is divided into multiple sub-suspected objects, and the sub-suspected objects are compared with the objects stored in the model library to determine the type of the sub-suspected objects; The type of the suspected object is determined based on the attribution type of all suspected sub-objects, thus obtaining the object; Creating contours from image data based on color differences includes: Image data is processed by sending it into different color channels to obtain monochrome image data. The color channels include the red channel, green channel and blue channel. Generate the frequency curve for each row in the monochrome image data; Obtain the abrupt change regions of the frequency curve and mark the abrupt change regions on the image data; Create outlines using abruptly changing regions; Each abrupt change region is located on the created contour; The abrupt change regions of the frequency curve include: Identify the amplitude abrupt change point on the frequency curve and construct an amplitude abrupt change reference region based on the amplitude abrupt change point; The frequency curve corresponding to the reference region with abrupt amplitude change is smoothed to obtain the reference curve; A moving screening wave is generated, which has a fixed frequency and a fixed attenuation rate; The moving screening wave is driven to move horizontally. The sudden change region of the frequency curve is obtained by superimposing the moving screening wave and the frequency curve. The length of the sudden change region is equal to the length of the moving screening wave. Generating a moving filter wave includes: The image data is processed into grayscale to obtain the outline; The reference frequency and reference attenuation rate of the moving screening wave are determined based on the corresponding position of the moving screening wave on the contour. Calculate the superposition of the moving screening wave at corresponding positions on the contour to obtain the superposition amount; Adjust the reference frequency and reference attenuation rate of the moving screening wave according to the superposition amount to maximize the superposition amount; The fixed frequency and fixed attenuation rate of the moving screening wave are determined based on the result of maximizing the superposition amount, and the moving screening wave is generated using the fixed frequency and fixed attenuation rate. The predicted trajectory includes: Obtain the current movement trajectory of an object in the moving state; The existing movement trajectory is segmented to obtain the existing movement trajectory segments; Determine the direction vector of the end of the existing trajectory segment; By statistically analyzing the changing trend of the direction vector in a time series, the pointing direction or pointing region can be obtained. Use the direction of the pointing or the area corresponding to the pointing region as the predicted movement trajectory; The trends in the statistical direction vector include: Calculate the similarity between two adjacent directional vectors in a time series. The similarity includes horizontal similarity and vertical similarity. The similarity direction is determined based on the ratio of horizontal similarity to vertical similarity; When similar directions are the same, the pointing direction is obtained; when similar directions are different, the pointing region is obtained.

2. An AI-powered intelligent identification device for industrial safety, using the AI-powered intelligent identification method for industrial safety as described in claim 1, characterized in that, include: The recognition unit is used to identify objects in an image based on the acquired image data. Objects include people, equipment, and environmental factors. The state determination unit is used to determine the state of an object, which includes moving state and stationary state; The simulation unit is used to simulate the movement trajectory of an object in a moving state to obtain the predicted movement trajectory. The calculation unit is used to calculate the interference between any two predicted moving trajectories and the interference between the predicted moving trajectory and an object in a stationary state. The early warning unit is used to screen interference situations and issue early warnings based on the screening results.

3. An AI-powered intelligent recognition system for industrial safety, characterized in that, The system includes: One or more memories for storing instructions; and One or more processors are configured to retrieve and execute the instructions from the memory, performing the method as described in claim 1.

4. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes: The program, when run by the processor, executes the method as described in claim 1.

Images