Test management auxiliary method and system based on gesture recognition

Abstract

The invention provides a test management auxiliary method and system based on gesture recognition, and the method comprises the steps: collecting video information through cameras disposed in all working places in a test field, and obtaining the working state information of all working equipment in the test field, identifying the test progress and the working state of the test site based on the video information and the working state information; whether a worker gesture exists in the video information or not is recognized, the worker gesture is recognized, gesture instruction information is obtained, and the worker gesture comprises at least one of a static gesture and a dynamic gesture; and determining the current test progress and working state of the test site according to the gesture instruction information, and recording and prompting test management personnel of the current test progress and working state of the test site. According to the invention, the test management personnel can inquire the test progress through gestures, so that the use of a traditional interphone communication mode to transmit information is avoided, and the test working efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of test site management systems, and particularly to a test management assistance method and system based on gesture recognition. Background Art

[0002] The site distribution of large-scale tests is wide, and there are many participating units and personnel. During the test process, complex tasks such as test equipment preparation, communication networking, safety guarantee, test implementation, test data collection and analysis, and tests need to be completed. The types and natures of the tasks undertaken by each test sub-site are different, and their work progress is also different. The overall test synchronization problem is very prominent. To enable the staff to understand the overall test progress, it is necessary to broadcast and explain the work progress of this test site and other test sites in real time, and a large amount of resources and manpower are required for the management work during the test process. Due to reasons such as the long physical distance between test sites and the backward communication technology, problems such as untimely information transmission, low work efficiency, and potential safety hazards in the test sites are still inevitable in practice. Summary of the Invention

[0003] The purpose of the present invention is to provide a test management assistance method and system based on gesture recognition, so that during the test process, test managers can query the test progress through gestures, avoid using traditional walkie-talkie communication methods to transmit information, and improve the test work efficiency.

[0004] To achieve the above invention purpose, the first aspect of the present invention provides a test management assistance method based on gesture recognition, and the method includes:

[0005] Collect video information through cameras set in each workplace in the test site, obtain the working status information of each working device in the test site, and identify the test progress and working status of the test site based on the video information and the working status information;

[0006] Identify whether there are staff gestures in the video information, recognize the staff gestures, and obtain gesture instruction information, where the staff gestures include at least one of static gestures and dynamic gestures;

[0007] Determine the current test progress and working status of the test site according to the gesture instruction information, record and prompt the test manager of the current test progress and working status of the test site.

[0008] Further, identifying the test progress and working status of the test site based on the video information and the working status information specifically includes:

[0009] Input the video information and the working status information of the working device into the device analysis model, predict the next action of the device, and obtain device action prediction information;

[0010] Input video information into the behavior analysis model to analyze the behavior of the staff and obtain the personnel behavior analysis results;

[0011] Input the personnel behavior analysis results and the current test subject information into the intention analysis model to predict the intentions and next actions of the staff and obtain the personnel action prediction information;

[0012] Input the work status information, equipment action prediction information, personnel behavior analysis results, and personnel action prediction information into the progress recognition model for processing to obtain the test site progress status information, where the test site progress status information includes the current test progress, current work status, and next action prediction information of the test site.

[0013] Further, the method further includes:

[0014] Identify whether there are gestures of the test management personnel in the video information, recognize the gestures of the test management personnel, and obtain the gesture instruction information;

[0015] Query the current test progress and work status of the test site according to the gesture instruction information, and prompt the test management personnel of the current test progress and work status of the test site.

[0016] Further, recognizing the gestures of the staff or the test management personnel to obtain the gesture instruction information specifically includes the following operations:

[0017] Obtain the gesture depth image information from the video information;

[0018] Use the threshold segmentation algorithm to segment the staff's gestures from the image background to obtain a separate gesture image;

[0019] Extract features from the separate gesture image through the feature extraction algorithm based on image appearance to obtain the gesture image features;

[0020] Input the gesture image features into the gesture recognition model to obtain the gesture action information, where the gesture recognition model is a hybrid network of the ViT model and ResNet;

[0021] Query the corresponding gesture instruction information according to the gesture action information, and there is a one-to-one correspondence between the gesture action information and the gesture instruction information.

[0022] Further, the pre-set gesture action information and its corresponding gesture instruction information are stored in pairs in the database, and the storage specifically includes the following operations:

[0023] Perform a hash operation on all pairs of gesture action information and gesture instruction information to obtain a hash value;

[0024] Encrypt all pairs of gesture action information and gesture instruction information to obtain encrypted data;

[0025] Upload the hash value and identification information to the blockchain for storage;

[0026] Save the encrypted data in the local database.

[0027] Further, the encryption specifically includes the following operations:

[0028] Compress all gesture action information to a low-dimensional space through an encoder, then reconstruct the data through a decoder, and extract the common features of all gesture action information;

[0029] Run the pseudo-random function for the first time, use the common features of all gesture action information as the input key of the pseudo-random function, dock the identification information of the user who sets the gesture action information and gesture instruction information with r + 1, where r is the number of iterative runs of the pseudo-random function, and use the result of the docking as the salt value and input key to calculate together in the pseudo-random function to obtain the output output[0] of the pseudo-random function for this run;

[0030] In subsequent iterations, run the pseudo-random function multiple times, use the output of the previous run of the pseudo-random function as the new salt value for this run, and input it together with the input key into the pseudo-random function for calculation again. After multiple runs, obtain the output set output[1,...,R - 1] of the pseudo-random function, where R represents the maximum number of iterations;

[0031] In each run of the pseudo-random function, perform an exclusive OR operation on all the outputs of the pseudo-random function up to this run to obtain the exclusive OR result E[r];

[0032] Concatenate all the exclusive OR results obtained in each run of the pseudo-random function, perform a hash operation on the concatenated result, and use the hash operation result of the concatenated result as the encryption key to encrypt all pairs of gesture action information and gesture instruction information.

[0033] Further, the method further includes: controlling the working device according to the gesture instruction information, specifically including the following operations:

[0034] Identify the video information from which the gesture instruction information comes to obtain the face information of the staff who issues the gesture instruction;

[0035] Identify the face information of the staff to determine the staff seat, and further determine the operation authority of the staff according to the staff seat;

[0036] Judge whether the operation authority of the staff matches the gesture instruction information. If it matches, control the working device according to the gesture instruction information.

[0037] In the second aspect of the present invention, a test management assistance system based on gesture recognition is provided. The system includes:

[0038] An acquisition module, configured to collect video information through cameras arranged in various workplaces within the test site, obtain the working status information of various working devices within the test site, and identify the test progress and working status of the test site based on the video information and the working status information;

[0039] A gesture recognition module, configured to identify whether there are staff gestures in the video information, recognize the staff gestures, and obtain gesture command information, where the staff gestures include at least one of static gestures and dynamic gestures;

[0040] A progress prompt module, configured to determine the current test progress and working status of the test site according to the gesture command information, record and prompt the test management personnel of the current test progress and working status of the test site.

[0041] Compared with the prior art, the beneficial effects of the present invention are:

[0042] A test management assistance method and system based on gesture recognition provided by the present invention, after collecting video information through cameras arranged in various workplaces within the test site, further obtains the working status information of various working devices within the test site, and identifies the test progress and working status of the test site based on the video information and the working status information; at the same time, the present invention can also recognize the staff gestures in the video information, obtain gesture command information, and determine the current test progress and working status of the test site according to the gesture command information of the staff, thereby simplifying the operation of the staff reporting the test progress, and the staff can focus more on each important link of the test system to ensure the smooth operation of the entire test system, and further improve the test efficiency. Description of the Drawings

[0043] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only the preferred embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0044] Figure 1 It is a schematic diagram of the overall process of a test management assistance method based on gesture recognition provided by an embodiment of the present invention. Detailed Embodiments

[0045] The principles and features of the present invention will be described below in conjunction with the drawings. The listed embodiments are only used to explain the present invention and are not used to limit the scope of the present invention.

[0046] Reference Figure 1 , this embodiment provides an experimental management assistance method based on gesture recognition, and the method includes:

[0047] S101. Collect video information through cameras set in various workplaces in the test field, obtain the working state information of various working devices in the test field, and identify the test progress and working state of the test field based on the video information and the working state information.

[0048] Exemplarily, the camera can adopt a depth camera. The working state information of the working device can include start, shutdown, power on, power off, etc.

[0049] S102. Identify whether there are staff gestures in the video information, recognize the staff gestures, and obtain gesture instruction information, where the staff gestures include at least one of static gestures and dynamic gestures.

[0050] S103. Determine the current test progress and working state of the test field according to the gesture instruction information, record and prompt the test manager of the current test progress and working state of the test field.

[0051] In this embodiment, the current test progress and working state of the test field can be determined by recognizing the gestures of the staff. That is to say, the gestures made by the staff facing the camera correspond to a certain test progress or working state. Exemplarily, the gestures made by the staff can be static or dynamic. Static gestures are to maintain a fixed gesture action; dynamic gestures are multiple continuously changing gesture actions. At the same time, it can be understood that the gesture actions made by the staff can be made with one hand or both hands. Combining the foregoing embodiments, the staff can make a static gesture with one hand and a dynamic gesture with the other hand, or both hands can make static or dynamic gestures, so as to increase the amount of content that the gesture instructions can carry.

[0052] As a possible implementation manner, identifying the test progress and working state of the test field based on the video information and the working state information specifically includes:

[0053] S201. Input the video information and the working state information of the working device into the device analysis model, predict the next action of the device, and obtain device action prediction information.

[0054] In this implementation manner, the device analysis model is a neural network model pre-trained with labeled sample video information and sample working state information.

[0055] S202. Input the video information into the behavior analysis model to analyze the current behavior of the staff and obtain the personnel behavior analysis result. The personnel behavior analysis result is used to characterize the current behavior content of the staff.

[0056] In this embodiment, the behavior analysis model is a neural network model pre-trained with labeled sample video information and sample personnel behavior information.

[0057] S203. Input the personnel behavior analysis result and the current test subject information into the intention analysis model to predict the intention and next action of the staff and obtain the personnel action prediction information.

[0058] In this embodiment, the intention analysis model is a neural network model pre-trained with labeled sample personnel behavior data, sample test subject data, sample personnel intention and next action data.

[0059] S204. Input the working state information, equipment action prediction information, personnel behavior analysis result, and personnel action prediction information into the progress recognition model for processing to obtain the test site progress state information. The test site progress state information includes the current test progress, current working state, and next action prediction information of the test site.

[0060] In this embodiment, the progress recognition model is a neural network model pre-trained with labeled sample working state information, sample equipment action prediction information, sample personnel behavior analysis result, sample personnel action prediction information, and sample test site progress state information.

[0061] In this embodiment, a set of gesture actions can be predefined to represent different test progress. When a staff member makes one of these gesture actions in front of the camera, by recognizing the staff member's gesture, the test progress, status, etc. information of the corresponding test site can be quickly determined and recorded, and then the test manager can be prompted with the current test progress and working state of the corresponding test site, so that the test manager can quickly understand the progress status of different test sites at the critical nodes of the test.

[0062] In another possible embodiment, the method further includes:

[0063] S301. Identify whether there is a test manager's gesture in the video information, recognize the test manager's gesture, and obtain the gesture instruction information.

[0064] S302. Query the current test progress and working state of the test site according to the gesture instruction information, and prompt the test manager with the current test progress and working state of the test site.

[0065] In this embodiment, the test manager can actively query the test progress and working status of the test site through gestures, avoiding the use of traditional walkie-talkie communication methods to transmit information. The advantage of gesture recognition is that it does not occupy voice communication resources and can quickly and accurately issue instructions in a noisy working environment, thus helping to improve work efficiency.

[0066] As a further possible embodiment, the gestures of the staff or the test manager are recognized to obtain gesture instruction information, which specifically includes the following operations:

[0067] S401. Obtain gesture depth image information from the video information.

[0068] S402. Use the threshold segmentation algorithm to segment the staff's gesture from the image background to obtain a separate gesture image.

[0069] The purpose of this step is to remove background interference by segmenting the gesture from the image background. The threshold segmentation algorithm is a region-based image segmentation technique, and its principle is to divide the image pixel points into several categories. The threshold segmentation method combines known information such as depth data and the position and shape of the gesture, and performs gesture segmentation and extraction by setting a specific depth threshold. Exemplarily, the threshold segmentation algorithm can adopt several methods such as the maximum entropy method, the cross entropy method, and the maximum inter-class variance method, and this embodiment does not make specific limitations on this. The threshold segmentation technology has low computational requirements and a simpler algorithm implementation method, and is particularly suitable for processing images with different gray levels.

[0070] S403. Extract features from the separate gesture image through a feature extraction algorithm based on image appearance to obtain gesture image features.

[0071] Image appearance feature extraction is a method of extracting visually visible appearance or apparent features from an image. This technology is mainly used to capture the basic characteristics of the image, such as points, lines, angles, segmented regions, shapes, and histograms. These basic characteristics of the image are achieved through various means such as obtaining the feature attributes of the image, the grayscale image, and the variable template. The feature extraction technology based on image appearance not only has low computational requirements but also can provide rich information expression.

[0072] S404. Input the gesture image features into the gesture recognition model to obtain gesture action information, and the gesture recognition model is a hybrid network of ViT (Vision Transformer) model and ResNet.

[0073] ViT transfers the Transformer architecture in natural language processing to image recognition tasks by splitting an image into multiple small patches and then treating these patches as elements in a sequence. This approach bypasses the traditional convolutional neural network (CNN) methods that rely on local receptive fields and layer-by-layer feature abstraction.

[0074] Compared with ResNet, Transformer can capture the relationships between image patches through its global attention mechanism, which is crucial for complex feature representation in image classification tasks. However, existing vision transformers have some deficiencies in processing images. Especially during the process of splitting an image into patches, structural information within the patches may be lost, and their global attention mechanism may neglect the importance of local features.

[0075] To overcome these limitations, this embodiment integrates the advantages of ResNet into the Transformer architecture. This method finely extracts features by introducing convolutional operations and designs special modules to hierarchically extract local and global features, which not only improves the performance of the model but also reduces the consumption of computing resources.

[0076] To avoid losing two-dimensional spatial features in patches, including details and edge information, etc., the hybrid network adopts a traditional convolutional starting structure (Convstem), which realizes downsampling and extraction of detailed features through stacking multiple 3x3 convolutional layers. To capture multi-scale features, the main body of the network adopts a staged Transformer structure, with 2x2 convolution used for downsampling with a stride of 2 before each stage and increasing the number of feature channels. Here, the hybrid network consists of a perception unit (LPU), a lightweight self-attention module (LMHSA), and an inverse residual feed-forward network (IRFFN). The following is the operation calculation of ViT integrating ResNet:

[0077] First, the input image passes through a lightweight perception unit, usually a convolutional layer with a 3x3 convolutional kernel, to capture local features and reduce the computational amount. This unit serves as the initial feature extractor of the network and provides a preliminary feature representation for subsequent Transformer modules. Assuming the input image is I, a 3x3 convolutional lightweight convolutional layer is applied to extract the local features Flpu of the image. The calculation process:

[0078] Flpu = Conv3×3(I) Flpu = Conv3×3

[0079] Next, the feature map is fed into the lightweight self-attention module. In this module, the feature map is split into multiple small blocks or patches, and then these patches are processed through the self-attention mechanism to capture global dependencies. The LMHSA is designed to reduce the computational amount, adopting strategies such as shrinking the attention window or reducing the number of heads to lower the attention calculation complexity. The specific steps are as follows: divide the output of the LPU into multiple non-overlapping patches, apply the self-attention mechanism to each patch, calculate the attention weights of each patch for other patches, and implement the calculation of attention weights through matrix multiplication and the softmax function.

[0080] After the self-attention module, the feature map enters the inverse residual feed-forward network. This network contains multiple feed-forward network layers of Transformer, which are organized in a residual connection manner, allowing gradients to flow more efficiently, thus alleviating the vanishing gradient problem in deep networks. Each feed-forward network layer usually contains two fully connected layers, connected by activation functions such as ReLU. The specific steps are as follows: use depthwise separable convolution instead of traditional convolution to reduce the number of parameters and computational amount; use residual connections to directly add the input to the output of the convolutional layer to prevent the vanishing gradient problem.

[0081] In the entire hybrid network, the local feature extraction ability of CNN and the global feature modeling ability of Transformer are effectively combined. The local features extracted by the LPU are combined with the global context features generated by the LMHSA, and then further feature refinement and fusion are performed through the IRFFN. The outputs of the LPU, LMHSA, and IRFFN can be fused by concatenation.

[0082] S405. Query the corresponding gesture instruction information according to the gesture action information. The gesture action information and the gesture instruction information are in a one-to-one correspondence relationship.

[0083] In this embodiment, the pre-set gesture action information and its corresponding gesture instruction information are stored in pairs in the database. The storage specifically includes the following operations:

[0084] S501. Perform a hash operation on all pairs of gesture action information and gesture instruction information to obtain a hash value.

[0085] S502. Encrypt all pairs of gesture action information and gesture instruction information to obtain encrypted data.

[0086] S503. Upload the hash value and the identification information to the blockchain for storage.

[0087] S504. Save the encrypted data in the local database.

[0088] In this embodiment, the hash values of the gesture action information and the gesture instruction information are stored on the blockchain, so that it can be verified later whether the gesture action information and the gesture instruction information have been tampered with according to the hash values. At the same time, the gesture action information and the gesture instruction information are encrypted to improve the security of the source data.

[0089] As a further possible embodiment, the encryption specifically includes the following operations:

[0090] S601. Compress all gesture action information into a low-dimensional space through an encoder, and then reconstruct the data through a decoder to extract the common features of all gesture action information.

[0091] S602. Run the pseudo-random function for the first time. Use the common features of all gesture action information as the input key of the pseudo-random function. Connect the identification information of the user who sets the gesture action information and the gesture instruction information with r + 1, where r is the number of iterative runs of the pseudo-random function. Use the result of the connection as the salt value and input it into the pseudo-random function together with the input key for calculation to obtain the output output[0] of this run of the pseudo-random function.

[0092] S603. Run the pseudo-random function multiple times in subsequent iterations. Use the output of the previous run of the pseudo-random function as the new salt value of this run, and input it into the pseudo-random function together with the input key for calculation again. After multiple runs, obtain the output set output[1,..., R - 1] of the pseudo-random function, where R represents the maximum number of iterations.

[0093] S604. In each run of the pseudo-random function, perform an exclusive OR operation on all the outputs of the pseudo-random function up to this run to obtain the exclusive OR result E[r].

[0094] S605. Concatenate all the exclusive OR results obtained in each run of the pseudo-random function, perform a hash operation on the concatenated result, and use the hash operation result of the concatenated result as the encryption key to encrypt all pairs of gesture action information and gesture instruction information.

[0095] In this embodiment, by extracting the common features of all the gesture action information of the user as the input key of the pseudo-random function, and combining the output obtained from multiple iterative operations of the pseudo-random function as the salt value to generate the encryption key for encryption. Since the amplitudes, speeds, etc. of different users' gesture actions are different, even if the gesture actions are the same, the generated encryption keys will still be different. Therefore, the uniqueness and security of the encryption key are greatly improved, and the security of gesture-related data can be better protected.

[0096] As yet another possible implementation, the method further includes: controlling the working device according to the gesture instruction information, which specifically includes the following operations:

[0097] S701. Identify the video information of the source of the gesture instruction information to obtain the face information of the staff member who issued the gesture instruction.

[0098] S702. Identify the face information of the staff member to determine the staff member's seat, and further determine the staff member's operation authority according to the staff member's seat.

[0099] S703. Determine whether the staff member's operation authority matches the gesture instruction information. If they match, control the working device according to the gesture instruction information.

[0100] In this implementation, the staff member can control the operation of the working device through gesture instructions to improve the efficiency of the test work. When the staff member inputs gesture instructions to control the working device, the face image of the staff member is identified to verify whether the staff member has the operation authority to execute the corresponding gesture instructions, thereby improving the standardization and safety of the test process.

[0101] Based on the same inventive concept as the foregoing method embodiment, another embodiment of the present invention provides a test management assistance system based on gesture recognition. The system includes:

[0102] An acquisition and obtaining module, configured to collect video information through cameras arranged in various workplaces in the test field, obtain the working status information of various working devices in the test field, and identify the test progress and working status of the test field based on the video information and the working status information;

[0103] A gesture recognition module, configured to identify whether there is a staff member's gesture in the video information, recognize the staff member's gesture, and obtain gesture instruction information, where the staff member's gesture includes at least one of a static gesture and a dynamic gesture;

[0104] A progress prompt module, configured to determine the current test progress and working status of the test field according to the gesture instruction information, record and prompt the test manager of the current test progress and working status of the test field.

[0105] The working principle and technical effects of the system embodiment can both refer to the foregoing method embodiment and will not be elaborated herein.

[0106] The foregoing are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A test management auxiliary method based on gesture recognition, characterized in that: The method comprises: The video information is collected by the cameras installed in each workplace in the test field, the working status information of each working equipment in the test field is obtained, and the test progress and working status of the test field are identified based on the video information and working status information; Identify whether there is a staff gesture in the video information, identify the staff gesture, and obtain gesture instruction information, where the staff gesture includes at least one of a static gesture and a dynamic gesture; The current test progress and working status of the test site are determined based on the gesture command information, and the current test progress and working status of the test site are recorded and prompted to the test management personnel.

2. The test management auxiliary method based on gesture recognition according to claim 1, characterized in that: Identify the test progress and working status of the test site based on video information and working status information, including: Input the video information and the working status information of the working equipment into the equipment analysis model, predict the next action of the equipment, and obtain the equipment action prediction information; Input the video information into the behavior analysis model, analyze the behavior of the staff, and obtain the personnel behavior analysis results; Input the personnel behavior analysis results and the current test subject information into the intention analysis model to predict the intention and next action of the staff and obtain the personnel action prediction information; The working status information, equipment action prediction information, personnel behavior analysis results, and personnel action prediction information are input into the progress recognition model for processing to obtain the test field progress status information, which includes the current test progress of the test field, the current working status, and the next action prediction information.

3. The test management auxiliary method based on gesture recognition according to claim 1, characterized in that: The method further comprises: Identify whether there is a test manager's gesture in the video information, identify the test manager's gesture, and obtain gesture instruction information; The current test progress and working status of the test site are queried according to the gesture command information, and the current test progress and working status of the test site are prompted to the test management personnel.

4. The test management auxiliary method based on gesture recognition according to claim 3 is characterized in that: Recognize the gestures of staff or test managers to obtain gesture instruction information, including the following operations: Obtaining gesture depth image information from video information; Use the threshold segmentation algorithm to segment the staff's gestures from the image background to obtain a separate gesture image; Extract features from individual gesture images using a feature extraction algorithm based on image appearance to obtain gesture image features; Inputting gesture image features into a gesture recognition model to obtain gesture action information, wherein the gesture recognition model is a hybrid network of a ViT model and a ResNet; The gesture instruction information corresponding to the gesture action information is searched, and the gesture action information and the gesture instruction information are in a one-to-one correspondence.

5. The test management auxiliary method based on gesture recognition according to claim 4 is characterized in that: The preset gesture action information and its corresponding gesture instruction information are stored in pairs in the database, and the storage specifically includes the following operations: Perform hash operations on all pairs of gesture action information and gesture command information to obtain hash values; Encrypting all pairs of gesture action information and gesture command information to obtain encrypted data; Upload the hash value and identification information to the blockchain for storage; Store encrypted data in a local database.

6. The test management assistance method based on gesture recognition according to claim 5, characterized in that: The encryption specifically includes the following operations: The encoder compresses all gesture information into a low-dimensional space, and then the decoder reconstructs the data to extract the common features of all gesture information. Run the pseudo-random function for the first time, use the common characteristics of all gesture action information as the input key of the pseudo-random function, connect the identification information of the user who sets the gesture action information and gesture command information with r+1, where r is the number of iterations of the pseudo-random function, and use the result of the connection as the salt value and the input key to input the pseudo-random function for calculation, and obtain the output output[0] of this pseudo-random function operation; In subsequent iterations, the pseudo-random function is run multiple times, and the output of the previous run of the pseudo-random function is used as the new salt value for this run. It is input into the pseudo-random function again together with the input key for calculation. After multiple runs, the output set of the pseudo-random function output[1,...,R-1] is obtained, where R represents the maximum number of iterations. When the pseudo-random function is run each time, all outputs of the pseudo-random function up to this run are XORed to obtain the XOR result E[r]; All XOR results obtained each time the pseudo-random function is run are concatenated, a hash operation is performed on the concatenated results, and the hash operation result of the concatenated results is used as an encryption key to encrypt all pairs of gesture action information and gesture command information.

7. The test management auxiliary method based on gesture recognition according to claim 1, characterized in that: The method further includes: controlling the working device according to the gesture instruction information, specifically including the following operations: Identify the video information of the source of the gesture command information and obtain the facial information of the staff member who issued the gesture command; Identify staff members’ facial information, determine their seats, and further determine their operating permissions based on their seats; Determine whether the operator's operating authority matches the gesture command information. If they match, control the work equipment according to the gesture command information.

8. A test management assistance system based on gesture recognition, characterized in that: The system comprises: The acquisition module is used to acquire video information through cameras installed at various work sites in the test site, acquire the working status information of various working equipment in the test site, and identify the test progress and working status of the test site based on the video information and working status information; A gesture recognition module, used to identify whether there is a staff gesture in the video information, recognize the staff gesture, and obtain gesture instruction information, wherein the staff gesture includes at least one of a static gesture and a dynamic gesture; The progress prompt module is used to determine the current test progress and working status of the test site according to the gesture command information, record and prompt the test management personnel of the current test progress and working status of the test site.

Images