Personnel violation operation determination method and system for identifying environment automatic monitoring station house based on depth image sequence

Abstract

The invention discloses a method and a system for judging illegal operation of personnel in an environment automatic monitoring station based on a depth image sequence. The system mainly comprises an image acquisition system, an image processing system, a calculation and analysis system, a violation operation alarm system and the like, and the system and an algorithm are mainly applied to an automatic atmospheric environment monitoring station to monitor violation or damage behaviors of corresponding personnel or animals to a station building, a sampling system, monitoring equipment and the like. According to the algorithm, features of an identification area, form identification features, face identification features, interface features of instrument operation and the like are integrated on the basis of human body posture features in the identification process, identification is more diversified, meanwhile, compared with multi-feature-point training, the identification accuracy can be improved, and misjudgment, false alarm and the like of a system are prevented.

Description

[0001] The present invention relates to the field of image analysis technology, and in particular to a method and system for determining illegal operations of personnel in an automatic monitoring station based on deep image sequence recognition environment.

[0002] This visual system and algorithm are mainly used in automatic atmospheric environment monitoring stations to monitor violations or damage to station buildings, sampling systems, monitoring equipment, etc. by relevant personnel or animals. The current traditional human posture recognition algorithms only make corresponding recognition and judgment on human posture and movements, such as fall algorithms, motion gesture recognition, human posture behavior recognition, etc. The accuracy of these recognition algorithms is not very high and is prone to false alarms or recognition errors.

[0003] During the recognition process, the algorithm incorporates the features of the recognition area, morphological recognition features, facial recognition features, interface features of instrument operation, etc. on the basis of human posture features, making the recognition more diversified. At the same time, the training of relatively more feature points can also improve the recognition accuracy and prevent the system from making misjudgments and false alarms.

[0004] On the one hand, the present application provides a method for determining illegal operations of personnel in an automatic monitoring station based on deep image sequence recognition environment; on the other hand, it provides a system for determining illegal operations of personnel in an automatic monitoring station based on deep image sequence recognition environment, so as to solve the problem that traditional human posture recognition algorithms only make corresponding recognition and judgment on human posture and movements, and the accuracy rate is not very high, which is prone to false alarms or recognition errors.

[0005] The present application provides a method for determining illegal operations of personnel in an automatic monitoring station based on deep image sequence recognition environment, the method comprising:

[0006] The movement trajectory and motion signals of the personnel in the automatic environmental monitoring station are collected, wherein the movement trajectory and the position signals of various parts of the body are collected and encoded through wearable positioning devices to form a time-ordered image sequence 1;

[0007] Furthermore, the wearable positioning device may preset location information;

[0008] Furthermore, the wearable positioning device has a communication system that can interact with other devices;

[0009] Furthermore, the wearable positioning device may compare the position information formed in real time with the preset position information;

[0010] Furthermore, the wearable positioning device may issue an alarm according to a threshold value or send an alarm signal to a control system;

[0011] Furthermore, the wearable positioning device has the function of realizing emergency alarm through sound or button.

[0012] The image signals of the actions of personnel in the automatic environment monitoring station are collected by the image acquisition system and form a real-time image sequence 2;

[0013] Further, the image acquisition system may acquire video;

[0014] Furthermore, the image acquisition system may be multiple groups;

[0015] Furthermore, the image acquisition system may preset multiple groups of images;

[0016] Further, the image acquisition system includes one or more image acquisition devices capable of acquiring images of specific locations;

[0017] Further, the image acquisition system includes one or more image acquisition devices capable of tracking a specific target to acquire a real-time image containing the specific target;

[0018] Furthermore, the image acquisition device capable of acquiring an image at a specific location and the image acquisition device capable of tracking a specific target to acquire a real-time image containing the specific target are the same device capable of realizing different functions;

[0019] Furthermore, the image acquisition system can transmit information with the image processing system;

[0020] Furthermore, the image acquisition system can track and acquire the target object based on the annotation information of the image processing system;

[0021] Furthermore, the image acquisition system includes an interactive operating interface and communicates with other devices.

[0022] The image processing system indexes the position signals and image signals of the image sequence 1 and the image sequence 2 to form the image sequence 3 and the image sequence 4 including the time, position and specific indexing data, and recognizes the time-series images;

[0023] Furthermore, the indexing method can be manual indexing or machine indexing;

[0024] Furthermore, machine indexing can improve indexing efficiency and accuracy through machine learning;

[0025] Further, the machine learning preferably performs deep image sequence recognition on the indexing through a convolutional neural network;

[0026] Furthermore, the types of indexing include but are not limited to the type of object, the attribute of the object, the task of the object, etc.;

[0027] Furthermore, the indexed attribute values ​​include but are not limited to region identification features, morphology identification features, hand identification features, face identification features, etc.;

[0028] Furthermore, the recognition efficiency and accuracy are improved by reducing the dimension of indexed data and classifying indexed data;

[0029] Furthermore, after the indexing, the staff will manually review the indexed images and correct the indexing errors that may have occurred;

[0030] Furthermore, the indexed attributes need to include at least one target attribute that can represent the target object;

[0031] Further, the image processing system may selectively index the image containing the target object according to the instruction issued by the image acquisition system;

[0032] Furthermore, the image processing system includes a plurality of indexing devices, and the plurality of indexing devices can interact with each other or with other devices.

[0033] The computing and analysis system can compare and analyze the labeled image sequence 3 and image sequence 4, and judge whether the real-time behavior and actions of the station personnel are in violation of regulations, erroneous operation or dangerous according to the machine learning results;

[0034] Furthermore, in the process of comparing and analyzing the image sequence 3 and the image sequence 4, the learning efficiency can be improved by denoising;

[0035] Furthermore, denoising can be performed by a preset denoising procedure;

[0036] Furthermore, denoising can be performed through machine learning;

[0037] Furthermore, machine learning can be performed using a preset training data set;

[0038] Furthermore, the training data set can be preset in multiple groups to improve the efficiency and accuracy of machine learning;

[0039] Furthermore, different training data sets containing indices are selected according to different station house environment weights;

[0040] Furthermore, the machine learning algorithm is selected from one of common machine learning algorithms such as decision tree, random forest algorithm, logistic regression, SVM, naive Bayes, K nearest neighbor algorithm, K means algorithm, Adaboost algorithm, neural network algorithm, Markov algorithm, etc.;

[0041] Furthermore, the machine learning algorithm selects a neural network algorithm;

[0042] Further, a specific denoising factor is set within the area formed by the positioning position obtained by the wearable device;

[0043] Further, a specific denoising factor is set in the area formed by the positioning position obtained by the wearable device to be smaller than the denoising factors in other areas;

[0044] Furthermore, machine learning obtains a denoising prediction model through training to perform denoising to obtain a predicted image sequence;

[0045] Furthermore, the image sequence indexed by learning the deep image sequence recognition network is compared with the image sequence obtained by the denoising prediction model to obtain the indexing denoising deviation value;

[0046] Further, the denoising factor is manually corrected according to the above deviation value;

[0047] Further, the above deviation value and denoising factor are reversely input into the deep image sequence recognition network and the denoising prediction model, and convolution and current learning are performed to optimize the deep image sequence recognition network parameters and the denoising prediction model;

[0048] Furthermore, machine learning automatically monitors the location information of the wearable positioning system when the station personnel operate and the corresponding indexed target attribute values ​​in the image processing system according to the environment, and records the relevant data information when the operation is correct in combination with the interface features of the instrument operation;

[0049] Furthermore, the real-time image sequence and position information obtained by the image acquisition system, the image processing system and the wearable device are analyzed in real time, and the deviation value of the relevant data information during correct operation is output.

[0050] The illegal operation alarm system determines whether to issue an illegal operation or danger alarm based on the deviation value output by the calculation and analysis system;

[0051] Furthermore, different thresholds are set according to the working environment time of the environment automatic monitoring station, and an early warning message is issued when the set threshold is exceeded;

[0052] Furthermore, different thresholds are set according to the number of people entering the station obtained by the image acquisition system, image processing system and wearable devices, and an early warning message is issued when the set threshold is exceeded;

[0053] Furthermore, the early warning information can be sent by the sound and light equipment in the automatic environmental monitoring station or sent to the outside world through communication equipment.

[0054] The present application also provides a system for automatically monitoring the station room for personnel illegal operation judgment based on the deep image sequence recognition environment. The system includes software, hardware, firmware or a combination thereof that can implement the above judgment method.

[0055] When software is used for implementation, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from a website site, a computer, a server or a data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode to another website site, computer, server or data center. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or a data center that includes one or more available media integration. The available medium can be a magnetic medium, (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state drive (SSD)), etc.

[0056] The various illustrative logic units and circuits described in this application can be implemented or operated by a general-purpose processor, a digital signal processor, an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or any combination of the above. The general-purpose processor can be a microprocessor, and optionally, the general-purpose processor can also be any conventional processor, controller, microcontroller or state machine. The processor can also be implemented by a combination of computing devices, such as a digital signal processor and a microprocessor, a plurality of microprocessors, one or more microprocessors in conjunction with a digital signal processor core, or any other similar configuration.

[0057] The steps of the method or algorithm described in this application can be directly embedded in hardware, software units executed by a processor, or a combination of the two. The software unit can be stored in RAM memory, flash memory, ROM memory, EPROM memory, EEPROM memory, register, hard disk, removable disk, CD-ROM or any other form of storage medium in the art. Exemplarily, the storage medium can be connected to the processor so that the processor can read information from the storage medium and write information to the storage medium. Optionally, the storage medium can also be integrated into the processor. The processor and the storage medium can be arranged in an ASIC, and the ASIC can be arranged in a terminal. Optionally, the processor and the storage medium can also be arranged in different components in the terminal. These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are performed on the computer or other programmable device to produce a computer-implemented process, so that the instructions executed on the computer or other programmable device provide for implementing the process in the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0058] The present invention incorporates the features of the recognition area, morphological recognition features, facial recognition features, interface features of instrument operation, etc. on the basis of the human body posture features during the recognition process, so that the recognition is more diversified. At the same time, the training of relatively multiple feature points can also improve the accuracy of recognition and prevent the system from making misjudgments and false alarms. Especially in the specific working environment of the automatic environmental monitoring station, the coding information of the wearable device and the operation information of the instrument interface are used to effectively supplement the image sequence recognition. Compared with the simple image sequence, the efficiency and accuracy of the image sequence recognition and information denoising are greatly improved, the accuracy of recognition is greatly improved, and the system is effectively prevented from making misjudgments and false alarms. At the same time, through the introduction and combination of the coding information of the wearable device, overfitting is avoided in the convolutional neural network learning process, and the loss of details of the target attributes in the denoising process is also avoided, so as to achieve a balance between efficiency and detail retention. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 The figure is a schematic diagram of the overall process of the method for determining illegal operations of personnel in an automatic monitoring station based on deep image sequence recognition environment.

[0060] Figure 2 Schematic diagram of the indexing process for an image processing system that performs machine learning on image sequence indexing by combining convolutional neural networks and manual indexing.

[0061] Figure 3 Schematic diagram of the denoising process of the computational analysis system using a neural network algorithm to denoise the indexed image sequence. DETAILED DESCRIPTION

[0062] The technical solution of the present invention will be described clearly and completely below. Obviously, the recorded embodiments are not all embodiments of the present invention. Based on the embodiments of the present invention, all embodiments obtained by those skilled in the art without making any creative work belong to the scope of protection of the present invention.

[0063] It is further explained that the step numbering of the embodiments of the present invention is for explaining the steps of the specific embodiments and does not represent the sequence or logical order of the corresponding steps. The execution of the relevant steps can be replaced according to the settings of the relevant programs or the connection sequence of the devices, and the execution order of the corresponding steps can be adjusted without affecting the implementation.

[0064] Embodiment 1

[0065] Step 1: Build a model, which includes a locatable wearable device, an image acquisition system, an image processing system, a computing and analysis system, and an illegal operation alarm system;

[0066] Step 2: The image processing system performs machine learning on the image sequence labeling by combining convolutional neural networks and manual labeling;

[0067] Step 3: The computing and analysis system uses a neural network algorithm to denoise the indexed image sequence;

[0068] Step 4: The personnel in the automatic environmental monitoring station wear positionable wearable devices, start the image acquisition system, image processing system, calculation and analysis system and illegal operation alarm system, and enter the station;

[0069] Step 5: The image acquisition system collects images in the environment automatic monitoring station in real time to form a real-time time-series image sequence;

[0070] Step 6: The calculation and analysis system uses the position information of the wearable positioning system of the personnel in the automatic environmental monitoring station and the corresponding indexed target attribute value in the image processing system, and records the relevant data information when the operation is correct in combination with the interface features of the instrument operation;

[0071] Step 7: Dynamically adjust the threshold according to the working hours, and send a warning message through the alarm system when the real-time data deviation exceeds the threshold.

[0072] Embodiment 2

[0073] Step 1: Build a model, which includes a locatable wearable device, an image acquisition system, an image processing system, a computing and analysis system, and an illegal operation alarm system;

[0074] Step 2.1: The image processing system performs machine learning on the image sequence labeling by combining convolutional neural networks and manual labeling;

[0075] Step 2.2: The indexing type includes but is not limited to the type of object, the attribute of the object, the task of the object, etc., and the indexing attribute value includes but is not limited to the region recognition feature, the shape recognition feature, the hand recognition feature, the face recognition feature, etc.;

[0076] Step 3.1: The computing and analysis system denoises the indexed image sequence using a neural network algorithm;

[0077] Step 3.2: Setting a specific denoising factor within the region formed by the positioning position obtained by the wearable device during the denoising process;

[0078] Step 4: The personnel in the automatic environmental monitoring station wear positionable wearable devices, start the image acquisition system, image processing system, calculation and analysis system and illegal operation alarm system, and enter the station;

[0079] Step 5: The image acquisition system acquires images in the automatic environment monitoring station in real time, wherein at least one image acquisition device tracks and acquires images of personnel in the automatic environment monitoring station, and the image acquisition system acquires images of multiple targets and forms a time-series image sequence;

[0080] Step 6: The calculation and analysis system uses the position information of the wearable positioning system of the personnel in the automatic environmental monitoring station and the corresponding indexed target attribute value in the image processing system, and records the relevant data information when the operation is correct in combination with the interface features of the instrument operation;

[0081] Step 7: Adjust the threshold according to the working hours and the number of wearable devices entering the station room. When the real-time data deviation exceeds the threshold, send an early warning message through the alarm system.

[0082] Embodiment 3

[0083] Step 1: Build a model, which includes a locatable wearable device, an image acquisition system, an image processing system, a computing and analysis system, and an illegal operation alarm system;

[0084] Step 2.1: The image processing system performs machine learning on the image sequence labeling by combining convolutional neural networks and manual labeling;

[0085] Step 2.2: The indexing type includes but is not limited to the type of object, the attribute of the object, the task of the object, etc., and the indexing attribute value includes but is not limited to the region recognition feature, the shape recognition feature, the hand recognition feature, the face recognition feature, etc.;

[0086] Step 3.1: The computing and analysis system denoises the indexed image sequence using a neural network algorithm;

[0087] Step 3.2: Setting a specific denoising factor within the region formed by the positioning position obtained by the wearable device during the denoising process;

[0088] Step 4: The personnel in the automatic environmental monitoring station wear positionable wearable devices, start the image acquisition system, image processing system, calculation and analysis system and illegal operation alarm system, and enter the station;

[0089] Step 5.1: The image acquisition system acquires images in the automatic environment monitoring station in real time, wherein at least one image acquisition device tracks and acquires images of personnel in the automatic environment monitoring station, and the image acquisition system acquires images of multiple targets and forms a time-series image sequence;

[0090] Step 5.2: The image acquisition system acquires images in the automatic environment monitoring station in real time, wherein at least one image acquisition device acquires images of a specific area of ​​the automatic environment monitoring station in real time and forms a time-series image sequence;

[0091] Step 6.1: The calculation and analysis system uses the position information of the wearable positioning system of the personnel in the automatic environmental monitoring station and the corresponding indexed target attribute value in the image processing system, and records the relevant data information when the operation is correct in combination with the interface features of the instrument operation;

[0092] Step 6.2: Analyze the real-time image sequence and position information obtained by the image acquisition system, image processing system and wearable device in real time, and output the deviation value of the relevant data information during correct operation.

[0093] Step 7: Adjust the threshold according to the working hours and the number of wearable devices entering the station room. When the real-time data deviation exceeds the threshold, send an early warning message through the alarm system.

[0094] Embodiment 4

[0095] Step 1: Build a model, which includes a locatable wearable device, an image acquisition system, an image processing system, a computing and analysis system, and an illegal operation alarm system;

[0096] Step 2.1: The image processing system performs machine learning on the image sequence labeling by combining convolutional neural networks and manual labeling;

[0097] Step 2.2: The indexing type includes but is not limited to the type of object, the attribute of the object, the task of the object, etc., and the indexing attribute value includes but is not limited to the region recognition feature, the shape recognition feature, the hand recognition feature, the face recognition feature, etc.;

[0098] Step 3.1: The computing and analysis system denoises the indexed image sequence using a neural network algorithm;

[0099] Step 3.2: Setting a specific denoising factor within the area formed by the positioning position obtained by the wearable device to be smaller than the denoising factors in other areas;

[0100] Step 4: The personnel in the automatic environmental monitoring station wear positionable wearable devices, start the image acquisition system, image processing system, calculation and analysis system and illegal operation alarm system, and enter the station;

[0101] Step 5.1: The image acquisition system acquires images in the automatic environment monitoring station in real time, wherein at least one image acquisition device tracks and acquires images of personnel in the automatic environment monitoring station, and the image acquisition system acquires images of multiple targets and forms a time-series image sequence;

[0102] Step 5.2: The image acquisition system acquires images in the automatic environment monitoring station in real time, wherein at least one image acquisition device acquires images of a specific area of ​​the automatic environment monitoring station in real time and forms a time-series image sequence;

[0103] Step 6.1: The calculation and analysis system uses the position information of the wearable positioning system of the personnel in the automatic environmental monitoring station and the corresponding indexed target attribute value in the image processing system, and records the relevant data information when the operation is correct in combination with the interface features of the instrument operation;

[0104] Step 6.2: Analyze the real-time image sequence and position information obtained by the image acquisition system, image processing system and wearable device in real time, and output the deviation value of the relevant data information during correct operation.

[0105] Step 7: Adjust the threshold according to the working hours and the number of wearable devices entering the station room. When the real-time data deviation exceeds the threshold, send an early warning message through the alarm system.

[0106] Through the present invention, in the specific working environment of the automatic environmental monitoring station, the coding information of the wearable device and the operation information of the instrument interface are used to effectively supplement the image sequence recognition. Compared with the simple image sequence, the efficiency and accuracy of the image sequence recognition and information denoising are greatly improved, the accuracy of the recognition is greatly improved, and the system is effectively prevented from making misjudgments, false alarms, etc. At the same time, compared with other embodiments, the fourth embodiment avoids overfitting in the convolutional neural network learning process and also avoids the loss of details of the target attribute in the denoising process through the introduction and combination of the coding information of the wearable device, thereby achieving a balance between efficiency and detail retention. Compared with the first and second embodiments, the third embodiment verifies that setting a specific denoising factor in the area formed by the positioning position obtained by the wearable device can effectively improve the accuracy of the recognition process and improve the learning efficiency of machine learning; compared with the third embodiment, the fourth embodiment verifies that setting a specific denoising factor in the area formed by the positioning position obtained by the wearable device in the denoising process is smaller than the denoising factor in other areas, effectively avoiding the loss of details of the target attribute, achieving a balance between denoising and detail retention, and also effectively increasing the efficiency of machine learning.

[0107] The foregoing description of certain embodiments is merely exemplary in nature and is in no way intended to limit the scope of the present disclosure or its application or use. In this detailed description of embodiments of the apparatus, system, and method of the present invention, reference is made to the accompanying drawings, which form a part of the present invention and illustrate in a graphical manner specific embodiments of the apparatus, system, and method in which the description can be practiced. These embodiments are described in sufficient detail to enable those skilled in the art to practice the apparatus, system, and method of the present invention, and it should be understood that other embodiments may be utilized, and structural and logical changes may be made without departing from the spirit and scope of the present disclosure. In addition, for the sake of clarity, when certain features are clear to those skilled in the art, their detailed description will no longer be discussed to avoid confusing the description of the embodiments of the present disclosure. Therefore, the discussion herein should not be understood in a restrictive sense, and the scope of the present disclosure is defined by the appended claims. As used herein, the term "device" may refer to a circuit, a device, a system, a component, or a combination thereof. For example, a device may be a computing device, a processor, a memory, a memory device, a mobile device, an edge device, a server, and / or a cloud computing system.

[0108] Of course, it should be understood that any of the examples, embodiments, or processes described herein may be combined or separated with one or more other examples, embodiments, and / or processes and / or performed in a separate device or device portion according to the apparatus, systems, devices, and methods of the present invention.

[0109] Finally, the above discussion is intended to be illustrative only and should not be construed as limiting the appended claims to any particular embodiment or group of embodiments. Therefore, although various embodiments of the present disclosure have been described in particular detail, it should also be understood that a number of modifications and alternative embodiments may be devised by one of ordinary skill in the art without departing from the broader and intended spirit and scope of the present disclosure as set forth in the appended claims. Therefore, the description and drawings should be viewed in an illustrative manner and are not intended to limit the scope of the appended claims.

Claims

1. A method for determining illegal operations of personnel in an automatic monitoring station based on deep image sequence recognition environment. The method comprises the following steps: Step 1: Build a model, which includes a locatable wearable device, an image acquisition system, an image processing system, a computing and analysis system, and an illegal operation alarm system; Step 2: The image processing system performs machine learning on the image sequence labeling by combining convolutional neural networks and manual labeling; Step 3: The computing and analysis system uses a neural network algorithm to denoise the indexed image sequence; Step 4: The personnel in the automatic environmental monitoring station wear positionable wearable devices, start the image acquisition system, image processing system, calculation and analysis system and illegal operation alarm system, and enter the station; Step 5: The image acquisition system collects images in the environment automatic monitoring station in real time to form a real-time time-series image sequence; Step 6: The calculation and analysis system uses the position information of the wearable positioning system when the personnel of the automatic environmental monitoring station are operating and the corresponding indexed target attribute values ​​in the image processing system, and records the relevant data information when the operation is correct in combination with the interface features of the instrument operation; Step 7: Adjust the threshold according to the working hours and the number of wearable devices entering the station room. When the real-time data deviation exceeds the threshold, send an early warning message through the alarm system.

2. The method for determining illegal operation of personnel in an automatic monitoring station based on deep image sequence recognition environment as claimed in claim 1, characterized in that: The types of indexing include but are not limited to the type of object, the attribute of the object, the task of the object, etc., and the attribute values ​​of indexing include but are not limited to regional recognition features, morphological recognition features, hand recognition features, and facial recognition features.

3. The method for determining illegal operation of personnel in an automatic monitoring station based on deep image sequence recognition environment as claimed in claim 1 or 2, characterized in that: During the denoising process, a specific denoising factor is set within the area formed by the positioning position obtained by the wearable device.

4. The method for determining illegal operation of personnel in an automatic monitoring station based on deep image sequence recognition environment as claimed in claim 3, characterized in that: In step 3, a specific denoising factor is set in the area formed by the positioning position obtained by the wearable device to be smaller than the denoising factors in other areas.

5. The method for determining illegal operation of personnel in an automatic monitoring station based on deep image sequence recognition environment as claimed in claim 4, characterized in that: The image acquisition system acquires images in the automatic environment monitoring station in real time, wherein at least one image acquisition device tracks and acquires images of personnel in the automatic environment monitoring station, and the image acquisition system acquires images of multiple targets and forms a time-series image sequence.

6. The method for determining illegal operation of personnel in an automatic monitoring station based on deep image sequence recognition environment as claimed in claim 5, characterized in that: In step 6, the real-time image sequence and position information obtained by the image acquisition system, the image processing system and the wearable device are analyzed in real time, and the deviation value of the relevant data information during the correct operation is output.

7. The method for determining illegal operation of personnel in an automatic monitoring station based on deep image sequence recognition environment as claimed in claim 6, characterized in that: When the number of wearable devices entering the station room is one set, the threshold value is set higher than when the number of wearable devices entering the station room is multiple sets.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute the method for determining personnel illegal operations in an automatic monitoring station based on deep image sequence recognition environment as described in any one of claims 1 to 7.

Images