Key point identification method and device based on security and protection monitoring

Abstract

The invention provides a key point identification method and device based on security and protection monitoring. The method comprises the following steps: converting an acquired video data source into image frame data; labeling the image frame data by using a labeling tool to obtain an image frame label; based on the image frame labels and the image frame data, training the plurality of constructed key point identification networks to obtain a plurality of trained key point identification networks; determining a target key point identification network from the plurality of trained key point identification networks; acquiring a to-be-detected video data source in real time, and converting the to-be-detected video data source into to-be-detected image frame data; and inputting the to-be-detected image frame data into the target key point identification network to obtain a monitoring key point. Through automatic labeling and multi-network training and selection, the problems of high labeling cost, poor model generalization ability and the like in the prior art are solved, and the analysis efficiency of security monitoring and the reliability of the monitoring process are enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field of security monitoring, and in particular to a key point identification method and device based on security monitoring. Background Art

[0002] With the increasing demand for industrial safety, the application of security monitoring systems is becoming more and more widespread. However, traditional monitoring systems rely on manual monitoring, which has problems such as limited monitoring range, high cost and low efficiency. The advancement of computer vision technology provides new solutions for security monitoring.

[0003] However, the current key point recognition model for security monitoring has poor generalization ability and may not be able to generalize well in different scenarios or data sets, resulting in unsatisfactory recognition results in different applications, which further reduces the reliability of monitoring. Summary of the invention

[0004] The present invention provides a key point identification method and device based on security monitoring, which are used to realize a more accurate security monitoring auxiliary labeling function, and enhance the analysis efficiency of security monitoring and the reliability of the monitoring process.

[0005] In a first aspect, the present invention provides a key point identification method based on security monitoring, comprising:

[0006] Convert the acquired video data source into image frame data;

[0007] Annotate the image frame data using an annotation tool to obtain an image frame label;

[0008] Based on the image frame labels and the image frame data, training the constructed multiple key point recognition networks to obtain multiple trained key point recognition networks;

[0009] Determining a target key point recognition network from the plurality of trained key point recognition networks;

[0010] Acquire the video data source to be tested in real time, and convert the video data source to be tested into image frame data to be tested;

[0011] The image frame data to be tested is input into the target key point recognition network to obtain monitoring key points.

[0012] Optionally, based on the image frame label and the image frame data, the constructed multiple key point recognition networks are trained to obtain multiple trained key point recognition networks, including:

[0013] Inputting the image frame data into the key point recognition network to obtain corresponding training prediction key points;

[0014] The trained key point recognition network is obtained according to the training predicted key points and the image frame labels corresponding to the image frame data.

[0015] Optionally, obtaining the trained key point recognition network according to the training predicted key points and the image frame labels corresponding to the image frame data includes:

[0016] Determining a training error according to the training prediction key points and the image frame labels corresponding to the image frame data;

[0017] Based on the training error, the network parameters are adjusted by a back propagation algorithm to obtain the optimal network parameters;

[0018] The target LSTM neural network is generated using the optimal network parameters.

[0019] Optionally, determining a target key point recognition network from a plurality of the trained key point recognition networks comprises:

[0020] Inputting the test image frame data into the trained key point recognition network to obtain the test prediction key points;

[0021] Determining an evaluation index of the trained key point recognition network based on the test predicted key points and the image frame labels corresponding to the test image frame data;

[0022] According to the evaluation index, the target key point recognition network is determined from the trained key point recognition networks.

[0023] Optionally, after training the constructed multiple key point recognition networks based on the image frame labels and the image frame data to obtain multiple trained key point recognition networks, the method further includes:

[0024] Generate a training process log for each of the trained key point recognition networks.

[0025] In a second aspect, the present invention provides a key point identification device based on security monitoring, comprising:

[0026] A first conversion module, used to convert the acquired video data source into image frame data;

[0027] A labeling module, used to label the image frame data using a labeling tool to obtain an image frame label;

[0028] A training module, used for training the constructed multiple key point recognition networks based on the image frame labels and the image frame data to obtain multiple trained key point recognition networks;

[0029] A recognition module, used to determine a target key point recognition network from a plurality of the trained key point recognition networks;

[0030] A second conversion module is used to obtain a video data source to be tested in real time, and convert the video data source to be tested into image frame data to be tested;

[0031] The prediction module is used to input the image frame data to be tested into the target key point recognition network to obtain monitoring key points.

[0032] Optionally, the training module includes:

[0033] A training prediction input submodule, used to input the image frame data into the key point recognition network to obtain corresponding training prediction key points;

[0034] The training submodule is used to obtain the trained key point recognition network according to the training prediction key points and the image frame labels corresponding to the image frame data.

[0035] In a third aspect, the present invention provides an electronic device, comprising a processor and a memory, wherein the memory stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, the steps in the method provided in the first aspect are performed.

[0036] In a fourth aspect, the present invention provides a storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, runs the steps of the method provided in the first aspect.

[0037] In a fifth aspect, the present invention provides a computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the steps in the method provided in the first aspect are performed.

[0038] It can be seen from the above technical solutions that the present invention has the following advantages:

[0039] The present invention provides a key point recognition method and device based on security monitoring, the method comprising: converting the acquired video data source into image frame data; using a labeling tool to label the image frame data to obtain image frame labels; based on the image frame labels and the image frame data, training multiple key point recognition networks constructed to obtain multiple trained key point recognition networks; determining a target key point recognition network from multiple trained key point recognition networks; acquiring a video data source to be tested in real time, and converting the video data source to be tested into image frame data to be tested; inputting the image frame data to be tested into the target key point recognition network to obtain monitoring key points. By labeling the image frame data using a labeling tool, and then training the key point recognition network based on the labeled data, the workload and cost of manual labeling are significantly reduced. At the same time, by selecting multiple trained key point recognition networks, the accuracy and reliability of key point recognition can be improved, ensuring stable recognition effects under different conditions. In summary, the present invention overcomes the problems of high labeling cost and poor model generalization ability in the prior art through automatic labeling, as well as multi-network training and selection, and enhances the analysis efficiency of security monitoring and the reliability of the monitoring process. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0041] Figure 1 This is a flow chart of a key point identification method based on security monitoring according to Embodiment 1 of the present invention;

[0042] Figure 2 This is a flow chart of a second embodiment of a key point identification method based on security monitoring of the present invention;

[0043] Figure 3 It is a marked schematic diagram of a second embodiment of a key point identification method based on security monitoring of the present invention;

[0044] Figure 4 The present invention is a structural block diagram of a key point identification device based on security monitoring embodiment of the present invention. DETAILED DESCRIPTION

[0045] The embodiments of the present invention provide a key point identification method and device based on security monitoring, which are used to realize a more accurate security monitoring auxiliary labeling function, enhance the analysis efficiency of security monitoring and the reliability of the monitoring process.

[0046] In order to make the purpose, features and advantages of the present invention more obvious and easy to understand, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described below are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0047] For example, see Figure 1 , Figure 1 This is a flow chart of a first embodiment of a key point identification method based on security monitoring of the present invention, the method comprising:

[0048] Step S101, converting the acquired video data source into image frame data;

[0049] In an embodiment of the present application, a video data source is obtained from a surveillance camera or other video source, and then a video decoding library is used to decode the video data into continuous image frames, and image frame data is extracted from the video stream at a certain frame rate (such as 30 frames per second).

[0050] Step S102, annotating the image frame data using an annotation tool to obtain an image frame label;

[0051] In the embodiment of the present application, a suitable annotation tool is selected to manually annotate the image frame data to mark the key point positions (personnel features).

[0052] Step S103, training the constructed multiple key point recognition networks based on the image frame labels and the image frame data to obtain multiple trained key point recognition networks;

[0053] In the embodiment of the present application, a plurality of key point recognition networks with different structures are designed, and then the annotated image frame data and label data are divided into a training set, a validation set and a test set. After the image frame data and label data are loaded into the model training framework, each network is trained using the training set data, and the network parameters are optimized to minimize the loss function. Subsequently, the validation set data is used to evaluate the network performance during training, and the hyperparameters are adjusted to optimize the model to obtain a trained key point recognition network.

[0054] Step S104, determining a target key point recognition network from the plurality of trained key point recognition networks;

[0055] In the embodiment of the present application, the test set data is used to evaluate the performance of each network, and the evaluation index is calculated. Then, the evaluation index of each network is compared, and the network with the best performance is selected as the target key point recognition network.

[0056] Step S105, acquiring a video data source to be tested in real time, and converting the video data source to be tested into image frame data to be tested;

[0057] In the embodiment of the present application, the video data source to be tested is decoded and the image frame data to be tested is extracted.

[0058] Step S106, inputting the image frame data to be tested into the target key point recognition network to obtain monitoring key points.

[0059] The embodiment of the present invention provides a key point recognition method based on security monitoring, including: converting the acquired video data source into image frame data; using a labeling tool to label the image frame data to obtain an image frame label; based on the image frame label and the image frame data, training the constructed multiple key point recognition networks to obtain multiple trained key point recognition networks; determining the target key point recognition network from the multiple trained key point recognition networks; acquiring the video data source to be tested in real time, and converting the video data source to be tested into the image frame data to be tested; inputting the image frame data to be tested into the target key point recognition network to obtain the monitoring key point. By labeling the image frame data using a labeling tool, and then training the key point recognition network based on the labeled data, the workload and cost of manual labeling are significantly reduced. At the same time, by selecting multiple trained key point recognition networks, the accuracy and reliability of key point recognition can be improved, ensuring stable recognition effects under different conditions. In summary, the embodiment of the present application overcomes the problems of high labeling cost and poor model generalization ability in the prior art through automatic labeling, as well as multi-network training and selection, and enhances the analysis efficiency of security monitoring and the reliability of the monitoring process.

[0060] For example 2, please refer to Figure 2 , Figure 2 This is a flow chart of a second embodiment of a key point identification method based on security monitoring of the present invention, the steps include:

[0061] Step S201, converting the acquired video data source into image frame data;

[0062] The embodiment of the present application supports the import of various video data sources, such as MP4, AVI, MKV, MOV, WMV, FLV, etc., covering different types of security monitoring equipment data sources, and then sets the frame rate to convert the video data source into image data frames at fixed intervals.

[0063] Step S202, annotating the image frame data using an annotation tool to obtain an image frame label;

[0064] The embodiment of the present application is based on the existing algorithm model, performs algorithm reasoning on the image frame data before annotation, and outputs the detection frame and category recognized by the existing model, such as Figure 3 As shown, it assists in subsequent manual labeling, and the output format can be selected based on the existing algorithm model type, such as COCO, Pascal VOC, YOLO, etc.

[0065] In the specific implementation, multiple annotation types are supported, such as target box annotation, key point annotation, and multi-category annotation.

[0066] Step S203, inputting the image frame data into the key point recognition network to obtain corresponding training prediction key points;

[0067] Step S204, obtaining the trained key point recognition network according to the training prediction key points corresponding to the image frame data and the image frame labels;

[0068] Step S205, inputting the test image frame data into the trained key point recognition network to obtain the test prediction key points;

[0069] In the embodiment of the present application, parameters including learning rate, batch size, number of training rounds and data enhancement method can be configured through the parameter setting interface before training. The training process can automatically query, apply for and allocate computing resources according to the video memory required for the training task, prevent video memory overflow and optimize resource utilization. In addition, the training process can also be displayed in real time, including training progress, loss function curve and model accuracy changes.

[0070] Step S206, generating a training process log of each trained key point recognition network;

[0071] In an embodiment of the present application, the training process log records in detail all information during the training process, including training parameters, warning and error messages during the training process, training results, etc., to facilitate users to perform subsequent analysis and tuning.

[0072] Step S207, determining an evaluation index of the trained key point recognition network based on the test prediction key points and the image frame labels corresponding to the test image frame data;

[0073] In the embodiment of the present application, the evaluation indicators include accuracy, recall rate, F1-score and ROC curve.

[0074] Step S208, determining the target key point recognition network from the trained key point recognition networks according to the evaluation index;

[0075] Step S209, acquiring the video data source to be tested in real time, and converting the video data source to be tested into image frame data to be tested;

[0076] Step S210: input the image frame data to be tested into the target key point recognition network to obtain monitoring key points.

[0077] The embodiment of the present invention discloses a key point recognition method based on security monitoring, including: converting the acquired video data source into image frame data; using a labeling tool to label the image frame data to obtain image frame labels; based on the image frame labels and the image frame data, training the constructed multiple key point recognition networks to obtain multiple trained key point recognition networks; determining the target key point recognition network from the multiple trained key point recognition networks; acquiring the video data source to be tested in real time, and converting the video data source to be tested into the image frame data to be tested; inputting the image frame data to be tested into the target key point recognition network to obtain monitoring key points. Through automated labeling, as well as multi-network training and selection, the problems of high labeling cost and poor model generalization ability in the prior art are overcome, and the analysis efficiency of security monitoring and the reliability of the monitoring process are enhanced. In addition, through the systematic training and evaluation process and detailed training process log records, the training efficiency, accuracy and stability of the model are improved, and the traceability of the training process and the support capability of model debugging and optimization are also enhanced.

[0078] For example 3, please refer to Figure 4 , Figure 4 This is a structural block diagram of an embodiment of a key point identification device based on security monitoring of the present invention, the device includes:

[0079] A first conversion module 301, used to convert the acquired video data source into image frame data;

[0080] A labeling module 302 is used to label the image frame data using a labeling tool to obtain an image frame label;

[0081] A training module 303 is used to train the constructed multiple key point recognition networks based on the image frame labels and the image frame data to obtain multiple trained key point recognition networks;

[0082] An identification module 304, used to determine a target key point recognition network from the plurality of trained key point recognition networks;

[0083] The second conversion module 305 is used to obtain the data source of the video data to be tested in real time, and convert the data source of the video data to be tested into the image frame data to be tested;

[0084] The prediction module 306 is used to input the image frame data to be tested into the target key point recognition network to obtain monitoring key points.

[0085] In an optional embodiment, the training module 303 includes:

[0086] A training prediction input submodule, used to input the image frame data into the key point recognition network to obtain corresponding training prediction key points;

[0087] The training submodule is used to obtain the trained key point recognition network according to the training prediction key points and the image frame labels corresponding to the image frame data.

[0088] In an optional embodiment, the training submodule includes:

[0089] A training error determination unit, configured to determine a training error based on the training prediction key points and the image frame labels corresponding to the image frame data;

[0090] A parameter adjustment unit, configured to adjust the network parameters through a back propagation algorithm based on the training error to obtain optimal network parameters;

[0091] The target network generation unit is used to generate the target LSTM neural network by adopting the optimal network parameters.

[0092] In an optional embodiment, the identification module 304 includes:

[0093] A test submodule, used to input the test image frame data into the trained key point recognition network to obtain the test prediction key points;

[0094] An evaluation index determination submodule, used to determine the evaluation index of the trained key point recognition network based on the test prediction key points and the image frame labels corresponding to the test image frame data;

[0095] The recognition submodule is used to determine the target key point recognition network from the trained key point recognition networks according to the evaluation index.

[0096] In an optional embodiment, it also includes:

[0097] The log generation module is used to generate a training process log of each of the trained key point recognition networks.

[0098] Embodiment 4, the embodiment of the present invention further provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of a key point identification method based on security monitoring in any embodiment, including:

[0099] Convert the acquired video data source into image frame data;

[0100] Annotate the image frame data using an annotation tool to obtain an image frame label;

[0101] Based on the image frame labels and the image frame data, training the constructed multiple key point recognition networks to obtain multiple trained key point recognition networks;

[0102] Determining a target key point recognition network from the plurality of trained key point recognition networks;

[0103] Acquire the video data source to be tested in real time, and convert the video data source to be tested into image frame data to be tested;

[0104] The image frame data to be tested is input into the target key point recognition network to obtain monitoring key points.

[0105] In an optional embodiment, based on the image frame label and the image frame data, the constructed multiple key point recognition networks are trained to obtain multiple trained key point recognition networks, including:

[0106] Inputting the image frame data into the key point recognition network to obtain corresponding training prediction key points;

[0107] The trained key point recognition network is obtained according to the training predicted key points and the image frame labels corresponding to the image frame data.

[0108] In an optional embodiment, obtaining the trained key point recognition network according to the training prediction key points corresponding to the image frame data and the image frame labels includes:

[0109] Determining a training error according to the training prediction key points and the image frame labels corresponding to the image frame data;

[0110] Based on the training error, the network parameters are adjusted by a back propagation algorithm to obtain the optimal network parameters;

[0111] The target LSTM neural network is generated using the optimal network parameters.

[0112] In an optional embodiment, determining a target key point recognition network from a plurality of the trained key point recognition networks comprises:

[0113] Inputting the test image frame data into the trained key point recognition network to obtain the test prediction key points;

[0114] Determining an evaluation index of the trained key point recognition network based on the test predicted key points and the image frame labels corresponding to the test image frame data;

[0115] According to the evaluation index, the target key point recognition network is determined from the trained key point recognition networks.

[0116] In an optional embodiment, after training the constructed multiple key point recognition networks based on the image frame labels and the image frame data to obtain multiple trained key point recognition networks, the method further includes:

[0117] Generate a training process log for each of the trained key point recognition networks.

[0118] Embodiment 5, the embodiment of the present invention further provides a computer storage medium, on which a computer program is stored, and when the computer program is executed by the processor, the steps of implementing a key point identification method based on security monitoring in any embodiment include:

[0119] Convert the acquired video data source into image frame data;

[0120] Annotate the image frame data using an annotation tool to obtain an image frame label;

[0121] Based on the image frame labels and the image frame data, training the constructed multiple key point recognition networks to obtain multiple trained key point recognition networks;

[0122] Determining a target key point recognition network from the plurality of trained key point recognition networks;

[0123] Acquire the video data source to be tested in real time, and convert the video data source to be tested into image frame data to be tested;

[0124] The image frame data to be tested is input into the target key point recognition network to obtain monitoring key points.

[0125] In an optional embodiment, based on the image frame label and the image frame data, the constructed multiple key point recognition networks are trained to obtain multiple trained key point recognition networks, including:

[0126] Inputting the image frame data into the key point recognition network to obtain corresponding training prediction key points;

[0127] The trained key point recognition network is obtained according to the training predicted key points and the image frame labels corresponding to the image frame data.

[0128] In an optional embodiment, obtaining the trained key point recognition network according to the training prediction key points corresponding to the image frame data and the image frame labels includes:

[0129] Determining a training error according to the training prediction key points and the image frame labels corresponding to the image frame data;

[0130] Based on the training error, the network parameters are adjusted by a back propagation algorithm to obtain the optimal network parameters;

[0131] The target LSTM neural network is generated using the optimal network parameters.

[0132] In an optional embodiment, determining a target key point recognition network from a plurality of the trained key point recognition networks comprises:

[0133] Inputting the test image frame data into the trained key point recognition network to obtain the test prediction key points;

[0134] Determining an evaluation index of the trained key point recognition network based on the test predicted key points and the image frame labels corresponding to the test image frame data;

[0135] According to the evaluation index, the target key point recognition network is determined from the trained key point recognition networks.

[0136] In an optional embodiment, after training the constructed multiple key point recognition networks based on the image frame labels and the image frame data to obtain multiple trained key point recognition networks, the method further includes:

[0137] Generate a training process log for each of the trained key point recognition networks.

[0138] Embodiment 6. The embodiment of the present invention further provides a computer program product, on which a computer program is stored. When the computer program is executed by the processor, the steps of implementing a key point identification method based on security monitoring in any embodiment include:

[0139] Convert the acquired video data source into image frame data;

[0140] Annotate the image frame data using an annotation tool to obtain an image frame label;

[0141] Based on the image frame labels and the image frame data, training the constructed multiple key point recognition networks to obtain multiple trained key point recognition networks;

[0142] Determining a target key point recognition network from the plurality of trained key point recognition networks;

[0143] Acquire the video data source to be tested in real time, and convert the video data source to be tested into image frame data to be tested;

[0144] The image frame data to be tested is input into the target key point recognition network to obtain monitoring key points.

[0145] In an optional embodiment, based on the image frame label and the image frame data, the constructed multiple key point recognition networks are trained to obtain multiple trained key point recognition networks, including:

[0146] Inputting the image frame data into the key point recognition network to obtain corresponding training prediction key points;

[0147] The trained key point recognition network is obtained according to the training predicted key points and the image frame labels corresponding to the image frame data.

[0148] In an optional embodiment, obtaining the trained key point recognition network according to the training prediction key points corresponding to the image frame data and the image frame labels includes:

[0149] Determining a training error according to the training prediction key points and the image frame labels corresponding to the image frame data;

[0150] Based on the training error, the network parameters are adjusted by a back propagation algorithm to obtain the optimal network parameters;

[0151] The target LSTM neural network is generated using the optimal network parameters.

[0152] In an optional embodiment, determining a target key point recognition network from a plurality of the trained key point recognition networks comprises:

[0153] Inputting the test image frame data into the trained key point recognition network to obtain the test prediction key points;

[0154] Determining an evaluation index of the trained key point recognition network based on the test predicted key points and the image frame labels corresponding to the test image frame data;

[0155] According to the evaluation index, the target key point recognition network is determined from the trained key point recognition networks.

[0156] In an optional embodiment, after training the constructed multiple key point recognition networks based on the image frame labels and the image frame data to obtain multiple trained key point recognition networks, the method further includes:

[0157] Generate a training process log for each of the trained key point recognition networks.

[0158] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0159] In the several embodiments provided in the present application, it should be understood that the methods, devices, electronic devices and storage media disclosed in the present invention can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0160] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0161] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0162] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a readable storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned readable storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0163] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A key point identification method based on security monitoring, characterized in that: include: Convert the acquired video data source into image frame data; Annotate the image frame data using an annotation tool to obtain an image frame label; Based on the image frame labels and the image frame data, training the constructed multiple key point recognition networks to obtain multiple trained key point recognition networks; Determining a target key point recognition network from the plurality of trained key point recognition networks; Acquire the video data source to be tested in real time, and convert the video data source to be tested into image frame data to be tested; The image frame data to be tested is input into the target key point recognition network to obtain monitoring key points.

2. The key point identification method based on security monitoring according to claim 1 is characterized in that: Based on the image frame label and the image frame data, the constructed multiple key point recognition networks are trained to obtain multiple trained key point recognition networks, including: Inputting the image frame data into the key point recognition network to obtain corresponding training prediction key points; The trained key point recognition network is obtained according to the training predicted key points and the image frame labels corresponding to the image frame data.

3. The key point identification method based on security monitoring according to claim 2 is characterized in that: Obtaining the trained key point recognition network according to the training prediction key points corresponding to the image frame data and the image frame labels, including: Determining a training error according to the training prediction key points and the image frame labels corresponding to the image frame data; Based on the training error, the network parameters are adjusted by a back propagation algorithm to obtain the optimal network parameters; The target LSTM neural network is generated using the optimal network parameters.

4. The key point identification method based on security monitoring according to claim 1 is characterized in that: Determining a target key point recognition network from the plurality of trained key point recognition networks comprises: Inputting the test image frame data into the trained key point recognition network to obtain the test prediction key points; Determining an evaluation index of the trained key point recognition network based on the test predicted key points and the image frame labels corresponding to the test image frame data; According to the evaluation index, the target key point recognition network is determined from the trained key point recognition networks.

5. The key point identification method based on security monitoring according to claim 1 is characterized in that: Based on the image frame label and the image frame data, the constructed multiple key point recognition networks are trained to obtain multiple trained key point recognition networks, and further comprising: Generate a training process log for each of the trained key point recognition networks.

6. A key point identification device based on security monitoring, characterized in that: include: A first conversion module, used to convert the acquired video data source into image frame data; A labeling module, used to label the image frame data using a labeling tool to obtain an image frame label; A training module, used for training the constructed multiple key point recognition networks based on the image frame labels and the image frame data to obtain multiple trained key point recognition networks; A recognition module, used to determine a target key point recognition network from a plurality of the trained key point recognition networks; A second conversion module is used to obtain a video data source to be tested in real time, and convert the video data source to be tested into image frame data to be tested; The prediction module is used to input the image frame data to be tested into the target key point recognition network to obtain monitoring key points.

7. The key point identification device based on security monitoring according to claim 6 is characterized in that: The training module includes: A training prediction input submodule, used to input the image frame data into the key point recognition network to obtain corresponding training prediction key points; The training submodule is used to obtain the trained key point recognition network according to the training prediction key points and the image frame labels corresponding to the image frame data.

8. An electronic device, characterized in that: The method comprises a processor and a memory, wherein the memory stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, the method according to any one of claims 1 to 5 is executed.

9. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 5 is performed.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method according to any one of claims 1 to 5 is implemented.

Images