A workwear identification method, device, equipment and storage medium

Abstract

The application discloses a work uniform identification method and device, equipment and a storage medium, and relates to the technical field of computers, which comprises the following steps: performing human body identification on a real-time picture and extracting feature information of the identified human body, performing clustering operation on the feature information to obtain a clustering result; storing a target work uniform and corresponding feature information determined based on the clustering result in a preset work uniform library to complete the establishment of the work uniform library; after the establishment of the work uniform library is completed, extracting feature information of a current target human body in the real-time picture, and identifying whether the current target human body wears the target work uniform according to the feature information of the current target human body; if the current target human body does not wear the target work uniform, determining whether the current target human body appears for the first time, and if not, eliminating the current target human body. The application can automatically generate and update the work uniform library, and realize the deduplication of personnel without work uniforms, so as to prevent problems such as resource occupation and inaccurate detection results caused by repeated statistics.

Description

Technical Field

[0001] This invention relates to the field of computer technology, and in particular to a method, apparatus, device, and storage medium for identifying work clothes. Background Technology

[0002] Currently, due to standardized production practices, most companies, factories, and construction sites require uniforms. Uniformity in clothing enhances a company's image and helps convey a fair and just corporate culture. In certain specialized industries, such as electronics factories, medical facilities, and food processing, specific work uniforms are necessary to protect employee safety, and these settings have more stringent requirements for uniform attire.

[0003] Currently, uniform inspection relies on two methods: manual inspection and intelligent management. Manual management is inefficient and costly, making it unsustainable for large companies. Intelligent management requires pre-deployment, meaning samples need to be manually created before deployment. This process can lead to issues like insufficient quantity and poor quality; factors such as sample completeness and pixel size affect sample metrics. Therefore, manual sample creation places high demands on deployment personnel and is a complex process. After sample creation, personnel must manually access the interface to import the pre-set samples into the server performing uniform recognition for deployment. If multiple servers are involved in uniform recognition, deployment must be performed on each server individually, resulting in a large workload, complex deployment methods, high personnel skill requirements, high deployment costs, and challenges for subsequent equipment maintenance. Existing work uniform recognition technologies require pre-setting a corresponding work uniform database, and updating the database also requires manual updating, making automation impossible. For example, it requires pre-arranging personnel to collect work uniform styles, and to ensure sample diversity, it is also necessary to manually collect images from various preset angles under different lighting and scenarios. When work uniform types need to be updated, samples need to be collected again, making subsequent equipment maintenance cumbersome and inefficient. On the other hand, existing work uniform detection mostly uses single-frame recognition and alarm methods. During the recognition and detection process, it is easy to repeatedly detect the same target not wearing a work uniform and trigger an alarm, thus repeatedly counting people who are not wearing work uniforms. Summary of the Invention

[0004] In view of this, the purpose of this invention is to provide a method, apparatus, device, and storage medium for identifying work uniforms, which can avoid the need for a pre-set work uniform database, realize the process of automatically generating and updating the work uniform database, and also achieve deduplication of personnel not wearing work uniforms, thereby preventing problems such as resource consumption and inaccurate detection results caused by duplicate statistics. The specific solution is as follows:

[0005] Firstly, this application discloses a method for identifying work clothes, including:

[0006] Human body recognition is performed on real-time images, and the corresponding feature information of the recognized human body is extracted. The feature information is then clustered to obtain the corresponding clustering results.

[0007] The target work clothes corresponding to the identified human body determined based on the clustering results and the corresponding feature information are stored in a preset work clothes library to complete the establishment of the work clothes library;

[0008] Once the work uniform library is established, the feature information of the current target human body in the real-time image is extracted, and the current target human body is identified as to whether the current target human body is wearing the target work uniform based on the feature information of the current target human body, so as to obtain the corresponding current identification result.

[0009] If the current identification result indicates that the current target human body is not wearing the target work clothes, then it is determined whether the current target human body is appearing for the first time. If so, the current target human body is stored in the preset alarm list database; otherwise, the current identification result is removed and alarms are prohibited from being triggered.

[0010] Optionally, the step of extracting the corresponding feature information of the identified human body and performing a clustering operation on the feature information to obtain the corresponding clustering results includes:

[0011] Extract the corresponding feature information of the identified human body, including human body orientation features, and perform clustering operation on the human body orientation features based on preset clustering rules to obtain a clustering result containing several clustering feature categories.

[0012] Optionally, the method further includes:

[0013] Determine if the lighting in the current scene has changed;

[0014] If so, then based on the preset feature supplementation rules, the feature information of the identified human body in the changed current scene lighting is extracted, so as to perform clustering operation on the feature information in the changed current scene lighting.

[0015] Optionally, the step of storing the target work clothes corresponding to the identified human body determined based on the clustering results and the corresponding feature information into a preset work clothes library to complete the establishment of the work clothes library includes:

[0016] Determine whether there exists a target cluster feature category in the clustering result whose feature quantity meets the preset work uniform determination conditions;

[0017] If so, the work clothes corresponding to the feature information of the target cluster feature category are determined as the target work clothes corresponding to the identified human body, and the target work clothes and the corresponding feature information are stored in the preset work clothes library to complete the establishment of the work clothes library.

[0018] Optionally, the step of identifying whether the current target human body is wearing the target work clothes based on the feature information of the current target human body to obtain the corresponding current identification result includes:

[0019] The similarity between the feature information of the current target human body and the feature information of the target work clothes is determined by feature comparison operation to see if it is not less than a preset feature similarity threshold.

[0020] If the similarity is not less than the preset feature similarity threshold, then it is determined that the current target human body is wearing the target work clothes, and the work clothes recognition operation for the current target human body ends.

[0021] If the similarity is less than the preset feature similarity threshold, it is determined that the current target human body is not wearing the target work clothes.

[0022] Optionally, determining whether the current target human body is appearing for the first time includes:

[0023] A facial recognition operation is performed on the current target human body to determine the target identity corresponding to the current target human body, and the current target human body is compared with the preset alarm list database to determine whether the current target human body is appearing for the first time.

[0024] Optionally, after storing the current target human body in a preset alarm list database, the method further includes:

[0025] Clustering operation is performed on the feature information of the current target human body to determine whether the cluster feature category corresponding to the feature information of the current target human body exists in the cluster feature category list of the preset alarm list library;

[0026] If so, the number of features in the cluster feature category in the cluster feature category list will be increased by a preset accumulation threshold;

[0027] If not, the clustering feature category is stored in the clustering feature category list;

[0028] Determine whether there exists a cluster feature category in the list of cluster feature categories whose number of features is greater than a preset feature number threshold;

[0029] If so, the process jumps back to the step of performing human body recognition on the real-time image and extracting the corresponding feature information of the recognized human body to perform clustering operation on the feature information to obtain the corresponding clustering result, so as to update the work clothes database.

[0030] Secondly, this application discloses a workwear identification device, comprising:

[0031] The feature clustering module is used to perform human body recognition on real-time images, extract the corresponding feature information of the recognized human body, and perform clustering operations on the feature information to obtain the corresponding clustering results.

[0032] The work uniform library establishment module is used to store the target work uniforms corresponding to the identified human body determined based on the clustering results and the corresponding feature information into a preset work uniform library to complete the establishment of the work uniform library.

[0033] The work uniform recognition module is used to extract the feature information of the current target human body in the real-time image after the work uniform database is established, and to identify whether the current target human body is wearing the target work uniform based on the feature information of the current target human body, so as to obtain the corresponding current recognition result.

[0034] The result rejection module is used to determine whether the current target human body is appearing for the first time if the current identification result indicates that the current target human body is not wearing the target work clothes. If so, the current target human body is stored in the preset alarm list library; otherwise, the current identification result is rejected and the alarm is prohibited from being triggered.

[0035] Thirdly, this application discloses an electronic device, including:

[0036] Memory, used to store computer programs;

[0037] A processor is used to execute the computer program to implement the aforementioned workwear identification method.

[0038] Fourthly, this application discloses a computer-readable storage medium for storing a computer program, which, when executed by a processor, implements the aforementioned work uniform identification method.

[0039] As can be seen, this application performs human body recognition on real-time footage and extracts the corresponding feature information of the recognized human body. It then performs clustering operations on the feature information to obtain corresponding clustering results. The target work clothes corresponding to the recognized human body, determined based on the clustering results, and the corresponding feature information are stored in a preset work clothes library to complete the establishment of the work clothes library. After the work clothes library is established, the feature information of the current target human body in the real-time footage is extracted, and the current target human body is identified as to whether it is wearing the target work clothes, based on the feature information, to obtain the corresponding current identification result. If the current identification result indicates that the current target human body is not wearing the target work clothes, it is determined whether the current target human body is appearing for the first time. If so, the current target human body is stored in a preset alarm list library; otherwise, the current identification result is removed and alarm triggering is prohibited. Therefore, this application automatically determines the target uniform by extracting and learning human features from real-time images. After the uniform library is created, uniform identification is performed, and duplicate identifications of personnel not wearing uniforms are removed. In this way, the process of determining uniforms and establishing a corresponding uniform library can be automated through monitoring and identification of real-time images. This facilitates rapid deployment and maintenance, reduces time and maintenance costs, and avoids problems such as the inability to update uniforms in a timely manner due to the need to pre-define the uniform library. Furthermore, this application can perform deduplication of personnel not wearing uniforms, which greatly facilitates the management of personnel not wearing uniforms. Attached Figure Description

[0040] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0041] Figure 1 This is a flowchart of a work uniform identification method disclosed in this application;

[0042] Figure 2 This is a flowchart of a specific work uniform identification method disclosed in this application;

[0043] Figure 3 This application discloses an automatic workwear import process flowchart;

[0044] Figure 4 This is a flowchart of a specific work uniform identification method disclosed in this application;

[0045] Figure 5 This is a flowchart of a specific work uniform identification method disclosed in this application;

[0046] Figure 6 This application discloses a flowchart for work uniform identification and updating.

[0047] Figure 7 This is a schematic diagram of the structure of a work uniform identification device disclosed in this application;

[0048] Figure 8 This is a structural diagram of an electronic device disclosed in this application. Detailed Implementation

[0049] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0050] Currently, uniform inspection relies on two methods: manual inspection and intelligent management. Manual management is inefficient and costly, making it unsustainable for large companies. Intelligent management requires pre-deployment, meaning samples need to be manually created and pre-imported into the server for uniform recognition, a complex and costly process. Existing uniform recognition technologies require pre-setting a uniform database, which also needs manual updating, making automation impossible. For example, personnel need to be pre-arranged to collect uniform styles, and to ensure sample diversity, images need to be taken from various angles under different lighting and scenarios. When updating uniform types, samples need to be collected again, making subsequent equipment maintenance cumbersome and inefficient. Furthermore, existing uniform detection methods often use single-frame recognition and alarms, which can easily lead to multiple detections of the same person not wearing a uniform, resulting in duplicate counts of ununiformed individuals.

[0051] To solve the above problems, see [link to relevant documentation]. Figure 1 As shown in the figure, this application discloses a method for identifying work clothes, including:

[0052] Step S11: Perform human body recognition on the real-time image, extract the corresponding feature information of the recognized human body, and perform clustering operation on the feature information to obtain the corresponding clustering results.

[0053] In this embodiment, first, it is necessary to perform human body recognition on the real-time video of the monitoring device, and extract the corresponding feature information of the recognized human body, so as to perform clustering operations on the corresponding feature information of the recognized human body to obtain the corresponding clustering results. Among them, the above processing is performed using a preset human body recognition model, feature recognition model, and clustering model. There are no specific restrictions on the human body recognition model, feature recognition model, and clustering model, and users can select a more suitable model for processing based on the current scenario. It can be understood that there are many human bodies recognized in the real-time video of the monitoring device, and here it is necessary to extract the corresponding feature information of each recognized human body in order to perform clustering operations using a large amount of feature information. Further, the monitoring device supports multiple cameras to start monitoring simultaneously to achieve multi-camera linkage.

[0054] Step S12: Store the target work uniforms corresponding to the recognized human body determined based on the clustering results and the corresponding feature information in a preset work uniform library to complete the establishment of the work uniform library.

[0055] In this embodiment, after the clustering operation is completed, the target work uniforms corresponding to the recognized human body are determined according to the clustering results, and the target work uniforms and the corresponding feature information are stored in a preset work uniform library to complete the establishment of the work uniform library. Among them, the number of the target work uniforms can be selected based on the user's needs. The work uniform library is used to store all the target work uniforms. Users can view the work uniform pictures corresponding to all the target work uniforms stored in the work uniform library through a preset human-computer interaction interface. The corresponding feature information of the target work uniforms is also stored in the background database for subsequent work uniform recognition operations. In this way, by real-time extracting the feature information of the recognized human body and performing clustering operations, the process of automatically scanning and determining work uniforms to determine the work uniform library can be realized, avoiding the operation of users pre-inputting a specific work uniform library, effectively improving the deployment efficiency and saving costs.

[0056] Step S13: When the work uniform library is established, extract the feature information of the current target human body in the real-time video, and identify whether the current target human body is wearing the target work uniform according to the feature information of the current target human body to obtain the corresponding current recognition result.

[0057] In this embodiment, after the work uniform database is established, the feature information of the current target human body identified in the real-time image is extracted by the monitoring device. Based on the feature information of the current target human body, an identification operation is performed to determine whether the current human body is wearing the target work uniform, thereby obtaining the corresponding current identification result. It is understood that during the work uniform identification operation, it is necessary to identify and detect each human body in the real-time image to determine whether each human body appearing in the real-time image is wearing a work uniform. The current target human body indicates the human body currently being detected and identified. Once the detection of the current target human body is completed, the next current target human body can be identified and detected to obtain the corresponding current identification result.

[0058] Step S14: If the current identification result indicates that the current target human body is not wearing the target work clothes, then determine whether the current target human body is appearing for the first time. If so, store the current target human body in the preset alarm list library; otherwise, remove the current identification result and prohibit triggering the alarm.

[0059] In this embodiment, if the current identification result indicates that the current target human body is not wearing the target uniform, it is necessary to further determine whether the current target human body is appearing for the first time, that is, the first human body not wearing uniform identified. If so, the current target human body is stored in the preset alarm list for subsequent alarm operation. Otherwise, the current identification result is removed and alarm triggering is prohibited. In this way, by removing the current target human bodies that are not wearing uniform that have already been counted, the problem of inaccurate data caused by repeated counting of current target human bodies not wearing uniform can be prevented, the problem of multiple reports for the same target is solved, and the user's review time is saved.

[0060] In this embodiment, it should be noted that the work uniform identification method is deployed in a specific device. Users can directly use the work uniform identification method provided by this solution to perform work uniform identification operations by connecting the work uniform identification device provided by this solution to the server, which effectively improves deployment efficiency.

[0061] As can be seen, this application performs human body recognition on real-time footage and extracts the corresponding feature information of the recognized human body. It then performs clustering operations on the feature information to obtain corresponding clustering results. The target work clothes corresponding to the recognized human body, determined based on the clustering results, and the corresponding feature information are stored in a preset work clothes library to complete the establishment of the work clothes library. After the work clothes library is established, the feature information of the current target human body in the real-time footage is extracted, and the current target human body is identified as to whether it is wearing the target work clothes, based on the feature information, to obtain the corresponding current identification result. If the current identification result indicates that the current target human body is not wearing the target work clothes, it is determined whether the current target human body is appearing for the first time. If so, the current target human body is stored in a preset alarm list library; otherwise, the current identification result is removed and alarm triggering is prohibited. Therefore, this application automatically determines the target uniform by extracting and learning human features from real-time images. After the uniform library is created, uniform identification is performed, and duplicate identifications of personnel not wearing uniforms are removed. In this way, the process of automatically determining uniforms and establishing a corresponding uniform library can be achieved through monitoring and identification of real-time images, which is convenient for rapid deployment. Furthermore, this application can perform deduplication of personnel not wearing uniforms, which greatly facilitates the management of personnel not wearing uniforms by enterprise managers.

[0062] As can be seen from the above embodiments, this application can determine the corresponding target uniforms by extracting features and clustering them from real-time images, and after the uniform identification operation, deduplicating the target human bodies that are repeatedly counted. Next, the process of establishing the uniform database will be described in detail. See [link to documentation]. Figure 2 As shown in the figure, this application discloses a specific method for identifying work clothes, including:

[0063] Step S21: Perform human body recognition on the real-time image, extract the corresponding feature information of the recognized human body including human body orientation features, and perform clustering operation on the human body orientation features based on preset clustering rules to obtain a clustering result containing several clustering feature categories.

[0064] In this embodiment, the real-time monitoring device is first connected via ONVIF (Open Network Video Interface Forum) / RTSP (Real Time Streaming Protocol) to automatically activate the uniform detection function. Human body recognition is then performed on the real-time video feed from the monitoring device, extracting the corresponding feature information of the identified human body. This feature information includes human body orientation features, which include, but are not limited to, frontal, back, and side view features. A preset feature recognition model can determine the current orientation of the identified human body based on angle thresholds, thereby extracting the corresponding human body orientation features. Simultaneously, the frequency of each human body orientation feature is recorded, i.e., the number of corresponding image features. The human body orientation features are then clustered based on preset clustering rules to obtain a clustering result containing several cluster feature categories. The preset clustering rule involves performing feature comparison on the feature information corresponding to each human body. Features with a similarity higher than a preset threshold are grouped into a single cluster feature category and assigned the same cluster feature ID. The preset clustering rule can also specify the number of human body orientation feature images to be extracted. For example, the system default number of features is set to 40, which includes 10 frontal image features, 10 back image features, 10 left side image features, and 10 right side image features. Only when the number of each type of image feature reaches the number set in the preset clustering rule can the human body feature clustering operation be completed to ensure sample richness. It is understood that each cluster feature category uniquely corresponds to a cluster feature ID to ensure that the cluster feature category has a unique identifier. It should be noted that, as the system continuously identifies human bodies, when a new human body and its corresponding feature information are identified and input into the preset clustering model, the feature information corresponding to the human body is compared with the feature information corresponding to different existing cluster feature categories. The cluster feature category to which the human body's feature information belongs is determined based on feature similarity. If the similarity with each of the existing cluster feature categories is lower than a preset threshold, the human body's feature information is set as a new cluster feature category, and a new cluster feature ID is assigned to it. The preset threshold can be set by the user. In this way, identification and clustering based on different human body orientation features can improve the richness of the samples, enhance the model's generalization ability, and increase the accuracy of uniform recognition.

[0065] In this embodiment, the method may further include: determining whether the current scene lighting has changed; if so, extracting feature information of the identified human body in the changed current scene lighting based on a preset feature supplementation rule, and performing clustering operations on the feature information in the changed current scene lighting. Since the lighting varies throughout the day (morning, noon, evening), if a change in the current scene lighting is detected during the establishment of the workwear database, the image features of the identified human body in the changed current scene lighting are extracted to supplement the feature information of the identified human body, and clustering operations are performed on the feature information in the changed current scene lighting. As an example, this solution sets the lighting to four levels: bright, dim, artificial light, and sunlight. Because workwear may reflect light under different lighting conditions, causing partial color differences, this affects the feature vector value and the similarity comparison. Therefore, different feature similarity thresholds can be set for calculation based on different scene lighting to compensate for the deviation of the feature vector. The preset feature supplementation rule can be set to increase the number of image features extracted. The system's default increase is 10, and users can also set the number of features extracted based on the current scene lighting. In this way, by performing feature supplementation operations under different lighting conditions and setting different feature similarity thresholds for feature comparison operations, the influence of scene lighting on work clothes recognition can be avoided, the recognition and feature comparison accuracy can be improved, and clustering features can be strengthened.

[0066] Step S22: Determine whether there are target clustering feature categories in the clustering results that meet the preset determination conditions for the number of features.

[0067] In this embodiment, since there are many different clustering feature categories in the final clustering result, it is necessary to determine whether there are target clustering feature categories among the clustering feature categories in the clustering result that meet the preset uniform determination conditions. The preset uniform determination conditions are the target clustering feature categories with the most features. The number of target clustering feature categories can be determined according to the number of uniform styles used for the preset. For example, if the number of uniform styles set by the user is 2, then the two clustering feature categories with the most features are determined as target clustering feature categories. As an example, the default value of the number of uniform styles in this system is 1.

[0068] Step S23: If yes, then the work clothes corresponding to the feature information of the target cluster feature category are determined as the target work clothes corresponding to the identified human body, and the target work clothes and the corresponding feature information are stored in the preset work clothes library to complete the establishment of the work clothes library.

[0069] In this embodiment, if there exists a target clustering feature category whose feature quantity meets the preset uniform determination conditions, then the uniform corresponding to the feature information of the target clustering feature category is determined as the target uniform corresponding to the identified human body, and the target uniform and the corresponding feature information are stored in a preset uniform library to complete the establishment of the uniform library. It can be understood that after the uniform library is established, the feature information of other non-target clustering feature categories has no value and can be deleted to prevent resource occupation and improve system running speed.

[0070] See Figure 3 The diagram shows a flowchart of an automatic work uniform database construction method disclosed in this application, which includes: first, accessing a video stream, then performing target human body recognition and feature extraction operations, extracting the corresponding human body and features, performing target pose detection and scene lighting detection, and performing feature clustering operations. If there are features that reach the clustering threshold, the target work uniform is identified, and the target work uniform is added to the database to retain the target clustering features. This completes the automatic database construction operation, and the work uniform recognition and analysis operation is started.

[0071] As can be seen, this application can identify the target work clothes and establish a corresponding work clothes library by recognizing the features of different human body orientations and features under different scene lighting conditions, and performing corresponding clustering processing. Furthermore, when the lighting in the current scene changes, the influence of lighting on feature comparison can be prevented by increasing the extraction of feature information and changing the feature similarity threshold, thereby compensating for vector deviations. This not only improves the richness of the samples but also enhances the overall accuracy of recognition and feature comparison.

[0072] As can be seen from the above embodiments, this application can determine the corresponding target uniforms and uniform library by clustering feature information based on different human body orientation features and current scene lighting. Next, uniform recognition can be performed. The process of uniform recognition will be described in detail below. See [link to documentation]. Figure 4 As shown in the figure, this application discloses a specific method for identifying work clothes, including:

[0073] Step S31: After the work clothes library is established, extract the feature information of the current target human body in the real-time image, and determine whether the similarity between the feature information of the current target human body and the feature information of the target work clothes is not less than the preset feature similarity threshold through feature comparison operation.

[0074] In this embodiment, after the work uniform database is established, the video stream from the monitoring device is connected to extract the feature information of the current target human body in the real-time image. A feature comparison operation is then performed to determine whether the similarity between the feature information of the current target human body and the feature information of the target work uniform in the database is not less than a preset feature similarity threshold. The feature extraction model used in extracting the feature information is the same as that used in step S21, and the preset feature similarity threshold can be set by the user based on their needs.

[0075] Step S32: If the similarity is not less than the preset feature similarity threshold, then determine that the current recognition result indicates that the current target human body is wearing the target work clothes, and end the work clothes recognition operation for the current target human body.

[0076] In this embodiment, if the similarity between the feature information of the current target human body and the feature information of the target work clothes is not less than the preset feature similarity threshold, then it is determined that the current recognition result indicates that the current target human body is wearing the target work clothes, and the work clothes recognition operation for the current target human body ends.

[0077] Step S33: If the similarity is less than the preset feature similarity threshold, then the current recognition result indicates that the current target human body is not wearing the target work clothes.

[0078] Step S34: Perform facial recognition on the current target human body to determine the target identity corresponding to the current target human body, and determine whether the current target human body is appearing for the first time by comparing it with the preset alarm list database.

[0079] In this embodiment, facial recognition is performed on the current target human body that is not wearing the target uniform, and the target identity of the current target human body is determined by comparing it with a preset employee list database. Then, it is determined whether the current target human body is appearing for the first time, i.e., whether it has been identified, by comparing it with a preset alarm list database. This comparison can be made by comparing alarm images detected in real-time with images stored in the preset alarm list database, or by comparing the corresponding employee information of the current target human body with the preset alarm list database after the target identity is determined. The preset employee list database is a list of company employees whose information has been pre-entered. The preset alarm list database stores employee information, corresponding alarm images, and characteristic information of the target human body for a certain period of time. The preset alarm list can be set to update every 24 hours. If the update time is set to 24 hours, the preset alarm list will be cleared every 24 hours for updating.

[0080] Step S35: If so, store the current target human body in the preset alarm list database.

[0081] In this embodiment, if the current target human body is appearing for the first time, the employee information corresponding to the current target human body, the corresponding alarm image identified, and the feature information corresponding to the current target human body are stored in the preset alarm list database for alarm operation.

[0082] Step S36: Otherwise, discard the current identification result and prevent the alarm from being triggered.

[0083] In this embodiment, if the current target human body is not appearing for the first time, that is, if the corresponding information of the current target human body already exists in the preset alarm list, the current identification result is removed and the alarm operation is prohibited. In this way, the duplicate detection of personnel can be deduplicated without having to judge whether there is a problem of multiple reports. At the same time, the identity information of personnel not wearing work clothes, such as name, employee number, department, etc., is reported, without having to search for personnel information level by level according to the image, which greatly facilitates the management of personnel not wearing work clothes.

[0084] As can be seen, this application determines whether the current target human body is wearing work clothes by comparing features to judge whether the feature similarity meets the preset feature similarity threshold, and performs face recognition operation on the person who is not wearing work clothes to determine whether the current target human body without work clothes is appearing for the first time. If it is not the first time, the current recognition result is removed. In this way, the duplicate detection of personnel can be deduplicated without judging whether there is a problem of multiple reports, which greatly facilitates the management of the situation of not wearing work clothes by the managers.

[0085] As can be seen from the above embodiments, this application can perform facial recognition on a target person not wearing work clothes to determine whether it is the first appearance and remove already recorded personnel. In addition to automatically generating a work clothes database and automatically recognizing work clothes, this method can also automatically update the work clothes database through real-time monitoring. The process of automatically updating the work clothes database will be described in detail below. See [link to documentation] Figure 5 As shown in the figure, this application discloses a specific method for identifying work clothes, including:

[0086] Step S41: Perform human body recognition on the real-time image, extract the corresponding feature information of the recognized human body, and perform clustering operation on the feature information to obtain the corresponding clustering results.

[0087] Step S42: Store the target work clothes corresponding to the identified human body determined based on the clustering results and the corresponding feature information into a preset work clothes library to complete the establishment of the work clothes library.

[0088] Step S43: After the work clothes library is established, extract the feature information of the current target human body in the real-time image, and identify whether the current target human body is wearing the target work clothes based on the feature information of the current target human body, so as to obtain the corresponding current identification result.

[0089] Step S44: If the current identification result indicates that the current target human body is not wearing the target work clothes, then perform a clustering operation on the feature information of the current target human body to determine whether the clustering feature category corresponding to the feature information of the current target human body exists in the clustering feature category list of the preset alarm list library.

[0090] In this embodiment, if the current identification result indicates that the current target human body is not wearing the target work clothes, then while storing the current target human body in the preset alarm list, it is necessary to perform a clustering operation on the feature information of the current target human body to determine whether the clustering feature category corresponding to the feature information of the current target human body exists in the clustering feature category list of the preset alarm list library, wherein all the clustering feature categories corresponding to the feature information in the preset alarm list library constitute a clustering feature category list.

[0091] Step S45: If yes, then increase the number of features of the cluster feature category in the cluster feature category list by a preset accumulation threshold.

[0092] In this embodiment, if the clustering feature category corresponding to the feature information of the current target human body exists in the clustering feature category list, then the number of features of the clustering feature category in the clustering feature category list is increased by a preset accumulation threshold, i.e., increased by 1, according to the clustering feature ID corresponding to the clustering feature category.

[0093] Step S46: If not, store the clustering feature category in the clustering feature category list.

[0094] In this embodiment, if the clustering feature category corresponding to the feature information of the current target human body does not exist in the clustering feature category list, then a new clustering feature ID is assigned to the clustering feature category corresponding to the feature information of the current target human body, and stored in the clustering feature category list.

[0095] Step S47: Determine whether there is a clustering feature category in the list of clustering feature categories with a feature quantity greater than a preset feature quantity threshold.

[0096] In this embodiment, since it is impossible for a large number of target human bodies not wearing work clothes to wear the same clothes at the same time, the number of features under each cluster feature ID of the cluster feature category list is basically 1. If the number of features of a certain cluster feature category reaches a certain threshold in a short period of time, it indicates that the company may have adjusted the style of work clothes. Therefore, it is necessary to perform a work clothes library update operation, that is, to detect the cluster feature category list to determine whether there is a cluster feature category in the cluster feature category list with a number of features greater than a preset feature number threshold. The default value of the preset feature number threshold is 5, which can be modified by the user according to the situation.

[0097] Step S48: If yes, then jump back to the step of performing human body recognition on the real-time image and extracting the corresponding feature information of the recognized human body to perform clustering operation on the feature information to obtain the corresponding clustering result, so as to update the work clothes database.

[0098] In this embodiment, if there is a clustering feature category in the clustering feature category list with a feature count greater than a preset feature count threshold, the process jumps back to step S41 to re-detect the current target uniform and update the uniform library. This way, by detecting the feature count of the clustering feature category corresponding to the feature information of personnel not wearing uniforms, it can determine whether uniform replacement is necessary, achieving automatic uniform style updates without requiring updates to each device individually, significantly reducing subsequent maintenance costs.

[0099] See Figure 6 The diagram shows a flowchart of a uniform identification and update process disclosed in this application, including: first, accessing a video stream to perform target human body identification and feature extraction operations; then, comparing the features of the target human body with the features of the uniform to determine whether the target human body is wearing a uniform; if so, ending the uniform identification operation for the target human body; otherwise, performing face recognition on the target human body and comparing it with the alarm list database for the day; if the comparison is successful, it indicates a duplicate alarm, the result is discarded and the uniform identification operation for the target human body ends; if the comparison is unsuccessful, it indicates that the target human body is detected for the first time, the corresponding information of the target human body is stored in the alarm list database and enters the waiting alarm area, and the feature information corresponding to the target human body is classified into one category; within a certain period of time, it is determined whether the number of alarms for the same type of features has reached a preset threshold; if so, it is determined that a uniform update operation is required, the uniform database is updated, re-clustered and added to the database for control, and pre-alarm information is deleted; otherwise, an alarm is triggered for personnel not wearing uniforms.

[0100] The specific processes of steps S41 to S43 can be found in the relevant content disclosed in the foregoing embodiments, and will not be repeated here.

[0101] As can be seen, this application determines whether a uniform update operation is needed by detecting the number of features in each category of the cluster feature category list in the preset alarm list library. If an update operation is needed, the uniforms are re-detected and identified to update the uniform library. In this way, the uniform style can be automatically updated without updating each device individually when updating uniforms, which greatly reduces subsequent maintenance costs.

[0102] refer to Figure 7 The present application also discloses a workwear identification device, comprising:

[0103] The feature clustering module 11 is used to perform human body recognition on real-time images, extract the corresponding feature information of the recognized human body, and perform clustering operations on the feature information to obtain the corresponding clustering results.

[0104] The work uniform library establishment module 12 is used to store the target work uniforms corresponding to the identified human body determined based on the clustering results and the corresponding feature information into a preset work uniform library to complete the establishment of the work uniform library.

[0105] The work uniform recognition module 13 is used to extract the feature information of the current target human body in the real-time image after the work uniform database is established, and to identify whether the current target human body is wearing the target work uniform based on the feature information of the current target human body, so as to obtain the corresponding current recognition result.

[0106] The result rejection module 14 is used to determine whether the current target human body is appearing for the first time if the current identification result indicates that the current target human body is not wearing the target work clothes. If so, the current target human body is stored in the preset alarm list library; otherwise, the current identification result is rejected and the alarm is prohibited from being triggered.

[0107] As can be seen, this application performs human body recognition on real-time footage and extracts the corresponding feature information of the recognized human body. It then performs clustering operations on the feature information to obtain corresponding clustering results. The target work clothes corresponding to the recognized human body, determined based on the clustering results, and the corresponding feature information are stored in a preset work clothes library to complete the establishment of the work clothes library. After the work clothes library is established, the feature information of the current target human body in the real-time footage is extracted, and the current target human body is identified as to whether it is wearing the target work clothes, based on the feature information, to obtain the corresponding current identification result. If the current identification result indicates that the current target human body is not wearing the target work clothes, it is determined whether the current target human body is appearing for the first time. If so, the current target human body is stored in a preset alarm list library; otherwise, the current identification result is removed and alarm triggering is prohibited. Therefore, this application automatically determines the target uniform by extracting and learning human features from real-time images. After the uniform library is created, uniform identification is performed, and duplicate identifications of personnel not wearing uniforms are removed. In this way, the process of determining uniforms and establishing a corresponding uniform library can be automated through monitoring and identification of real-time images. This facilitates rapid deployment and maintenance, reduces time and maintenance costs, and avoids problems such as the inability to update uniforms in a timely manner due to the need to pre-define the uniform library. Furthermore, this application can perform deduplication of personnel not wearing uniforms, which greatly facilitates the management of personnel not wearing uniforms.

[0108] In some specific embodiments, the feature clustering module 11 can be used to extract the feature information of the identified human body, including the human body orientation features, and perform clustering operations on the human body orientation features based on preset clustering rules to obtain a clustering result containing several clustering feature categories.

[0109] In some specific embodiments, the work clothes identification device may further include:

[0110] The lighting detection module is used to determine whether the lighting in the current scene has changed;

[0111] The feature supplementation module is used to extract the feature information of the identified human body in the changed current scene light based on a preset feature supplementation rule when the current scene light changes, so as to perform clustering operation on the feature information in the changed current scene light.

[0112] In some specific embodiments, the service library establishment module 12 may specifically include:

[0113] The first quantity judgment unit is used to determine whether there are target cluster feature categories in the clustering results whose feature quantity meets the preset work uniform determination conditions;

[0114] The uniform determination unit is used to determine the uniform corresponding to the feature information of the target cluster feature category as the target uniform corresponding to the identified human body when there is a target cluster feature category among the several cluster feature categories whose feature quantity meets the preset uniform determination conditions, and to store the target uniform and the corresponding feature information into the preset uniform library to complete the establishment of the uniform library.

[0115] In some specific embodiments, the work uniform identification module 13 may specifically include:

[0116] The similarity judgment unit is used to determine, through feature comparison operation, whether the similarity between the feature information of the current target human body and the feature information of the target work clothes is not less than a preset feature similarity threshold.

[0117] The first determination unit is used to determine that the current target human body is wearing the target work clothes when the similarity is not less than the preset feature similarity threshold, and to end the work clothes recognition operation of the current target human body.

[0118] The second determination unit is used to determine that the current target human body is not wearing the target work clothes when the similarity is less than the preset feature similarity threshold.

[0119] In some specific embodiments, the result elimination module 14 can be used to perform facial recognition on the current target human body to determine the target identity corresponding to the current target human body, and to determine whether the current target human body is appearing for the first time by comparing it with the preset alarm list database.

[0120] In some specific embodiments, the work clothes identification device may further include:

[0121] The category determination module is used to perform clustering operations on the feature information of the current target human body to determine whether the clustering feature category corresponding to the feature information of the current target human body exists in the clustering feature category list of the preset alarm list library;

[0122] The quantity increase module is used to increase the number of features of the cluster feature category in the cluster feature category list by a preset accumulation threshold when the cluster feature category corresponding to the feature information of the current target human body exists in the cluster feature category list of the preset alarm list library.

[0123] The category storage module is used to store the cluster feature category in the cluster feature category list when the cluster feature category corresponding to the feature information of the current target human body does not exist in the cluster feature category list of the preset alarm list library;

[0124] The second quantity judgment module is used to determine whether there is a cluster feature category in the cluster feature category list whose feature quantity is greater than a preset feature quantity threshold;

[0125] The workwear library update module is used to update the workwear library when there is a clustering feature category in the clustering feature category list with a feature quantity greater than a preset feature quantity threshold. In this case, the module will jump back to the step of performing human body recognition on the real-time image and extracting the corresponding feature information of the recognized human body to perform clustering operation on the feature information to obtain the corresponding clustering result.

[0126] Furthermore, embodiments of this application also disclose an electronic device, Figure 8 This is a structural diagram of an electronic device 20 according to an exemplary embodiment. The content of the diagram should not be construed as limiting the scope of this application.

[0127] Figure 8 This is a schematic diagram of the structure of an electronic device 20 provided in an embodiment of this application. Specifically, the electronic device 20 may include: at least one processor 21, at least one memory 22, a power supply 23, a communication interface 24, an input / output interface 25, and a communication bus 26. The memory 22 stores a computer program, which is loaded and executed by the processor 21 to implement the relevant steps in the work uniform identification method disclosed in any of the foregoing embodiments. Alternatively, the electronic device 20 in this embodiment may specifically be an electronic computer.

[0128] In this embodiment, the power supply 23 is used to provide operating voltage for each hardware device on the electronic device 20; the communication interface 24 can create a data transmission channel between the electronic device 20 and external devices, and the communication protocol it follows can be any communication protocol applicable to the technical solution of this application, and is not specifically limited here; the input / output interface 25 is used to acquire external input data or output data to the outside world, and its specific interface type can be selected according to specific application needs, and is not specifically limited here.

[0129] In addition, the memory 22, as a carrier for resource storage, can be a read-only memory, random access memory, disk or optical disk, etc. The resources stored thereon can include operating system 221, computer program 222, etc., and the storage method can be temporary storage or permanent storage.

[0130] The operating system 221 is used to manage and control the various hardware devices on the electronic device 20 and the computer program 222, which may be Windows Server, Netware, Unix, Linux, etc. In addition to including a computer program capable of performing the work uniform identification method executed by the electronic device 20 as disclosed in any of the foregoing embodiments, the computer program 222 may further include a computer program capable of performing other specific tasks.

[0131] Furthermore, this application also discloses a computer-readable storage medium for storing a computer program; wherein, when the computer program is executed by a processor, it implements the aforementioned work uniform identification method. Specific steps of this method can be found in the corresponding content disclosed in the foregoing embodiments, and will not be repeated here.

[0132] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to in the method section.

[0133] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0134] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented directly by hardware, a software module executed by a processor, or a combination of both. The software module can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.

[0135] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0136] The technical solutions provided in this application have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for identifying work uniforms, characterized in that, include: Human body recognition is performed on real-time images, and the corresponding feature information of the recognized human body is extracted. The feature information is then clustered to obtain the corresponding clustering results. The target work clothes corresponding to the identified human body determined based on the clustering results and the corresponding feature information are stored in a preset work clothes library to complete the establishment of the work clothes library; Once the work uniform library is established, the feature information of the current target human body in the real-time image is extracted, and the current target human body is identified as to whether the current target human body is wearing the target work uniform based on the feature information of the current target human body, so as to obtain the corresponding current identification result. If the current identification result indicates that the current target human body is not wearing the target work clothes, then it is determined whether the current target human body is appearing for the first time. If so, the current target human body is stored in the preset alarm list library; otherwise, the current identification result is removed and alarm triggering is prohibited. The step of storing the current target human body in a preset alarm list database further includes: Clustering operation is performed on the feature information of the current target human body to determine whether the cluster feature category corresponding to the feature information of the current target human body exists in the cluster feature category list of the preset alarm list library; If so, the number of features in the cluster feature category in the cluster feature category list will be increased by a preset accumulation threshold; If not, the clustering feature category is stored in the clustering feature category list; Determine whether there exists a cluster feature category in the list of cluster feature categories whose number of features is greater than a preset feature number threshold; If so, the process jumps back to the step of performing human body recognition on the real-time image and extracting the corresponding feature information of the recognized human body to perform clustering operation on the feature information to obtain the corresponding clustering result, so as to update the work clothes database.

2. The work uniform identification method according to claim 1, characterized in that, The step of extracting the corresponding feature information of the identified human body and performing clustering operations on the feature information to obtain the corresponding clustering results includes: Extract the corresponding feature information of the identified human body, including human body orientation features, and perform clustering operation on the human body orientation features based on preset clustering rules to obtain a clustering result containing several clustering feature categories.

3. The work uniform identification method according to claim 2, characterized in that, Also includes: Determine if the lighting in the current scene has changed; If so, then based on the preset feature supplementation rules, the feature information of the identified human body in the changed current scene lighting is extracted, so as to perform clustering operation on the feature information in the changed current scene lighting.

4. The work uniform identification method according to claim 2, characterized in that, The step of storing the target work clothes corresponding to the identified human body determined based on the clustering results and the corresponding feature information into a preset work clothes library to complete the establishment of the work clothes library includes: Determine whether there exists a target cluster feature category in the clustering result whose feature quantity meets the preset work uniform determination conditions; If so, the work clothes corresponding to the feature information of the target cluster feature category are determined as the target work clothes corresponding to the identified human body, and the target work clothes and the corresponding feature information are stored in the preset work clothes library to complete the establishment of the work clothes library.

5. The work uniform identification method according to claim 1, characterized in that, The step of identifying whether the current target human body is wearing the target work clothes based on the feature information of the current target human body, so as to obtain the corresponding current identification result, includes: The similarity between the feature information of the current target human body and the feature information of the target work clothes is determined by feature comparison operation to see if it is not less than a preset feature similarity threshold. If the similarity is not less than the preset feature similarity threshold, then it is determined that the current target human body is wearing the target work clothes, and the work clothes recognition operation for the current target human body ends. If the similarity is less than the preset feature similarity threshold, it is determined that the current target human body is not wearing the target work clothes.

6. The work uniform identification method according to claim 1, characterized in that, Determining whether the current target human body is appearing for the first time includes: A facial recognition operation is performed on the current target human body to determine the target identity corresponding to the current target human body, and the current target human body is compared with the preset alarm list database to determine whether the current target human body is appearing for the first time.

7. A work uniform identification device, characterized in that, include: The feature clustering module is used to perform human body recognition on real-time images, extract the corresponding feature information of the recognized human body, and perform clustering operations on the feature information to obtain the corresponding clustering results. The work uniform library establishment module is used to store the target work uniforms corresponding to the identified human body determined based on the clustering results and the corresponding feature information into a preset work uniform library to complete the establishment of the work uniform library. The work uniform recognition module is used to extract the feature information of the current target human body in the real-time image after the work uniform database is established, and to identify whether the current target human body is wearing the target work uniform based on the feature information of the current target human body, so as to obtain the corresponding current recognition result. The result rejection module is used to determine whether the current target human body is appearing for the first time if the current identification result indicates that the current target human body is not wearing the target work clothes. If so, the current target human body is stored in the preset alarm list library; otherwise, the current identification result is rejected and the alarm is prohibited from being triggered. The work clothes identification device further includes: The category determination module is used to perform clustering operations on the feature information of the current target human body to determine whether the clustering feature category corresponding to the feature information of the current target human body exists in the clustering feature category list of the preset alarm list library; The quantity increase module is used to increase the number of features of the cluster feature category in the cluster feature category list by a preset accumulation threshold when the cluster feature category corresponding to the feature information of the current target human body exists in the cluster feature category list of the preset alarm list library. The category storage module is used to store the cluster feature category in the cluster feature category list when the cluster feature category corresponding to the feature information of the current target human body does not exist in the cluster feature category list of the preset alarm list library; The second quantity judgment module is used to determine whether there is a cluster feature category in the cluster feature category list whose feature quantity is greater than a preset feature quantity threshold; The workwear library update module is used to update the workwear library when there is a clustering feature category in the clustering feature category list with a feature quantity greater than a preset feature quantity threshold. In this case, the module will jump back to the step of performing human body recognition on the real-time image and extracting the corresponding feature information of the recognized human body to perform clustering operation on the feature information to obtain the corresponding clustering result.

8. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor for executing the computer program to implement the workwear identification method as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, Used to store a computer program, which, when executed by a processor, implements the work clothes identification method as described in any one of claims 1 to 6.

Images