Gesture action detection method and gesture action detection device

Abstract

The invention relates to the technical field of action detection, and discloses a gesture action detection method and a gesture action detection device, which are applied to a vision detection scene, and the method comprises the following steps: obtaining monitoring information for a preset area in real time; judging whether a vision detection condition triggered by the detected person exists or not based on the monitoring information; if yes, locking the detected person; and acquiring gesture recognition information of the detected person, and determining a detection result of a gesture action based on the gesture recognition information. By adopting a non-inductive mode, automatic identification, locking and gesture identification are performed on the detected person, so that the vision detection result is quickly generated, the operation complexity is greatly reduced, and the user experience is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of action detection, and in particular to a method for detecting a gesture action and a device for detecting a gesture action. Background Art

[0002] With the continuous advancement of technology, electronic devices are playing an increasingly important role in our daily lives. As electronic devices become more popular, the time people spend using electronic screens has also increased significantly. However, this frequent use also brings health risks, especially damage to the eyes. Many people neglect proper rest and eye protection when using electronic devices for a long time, which makes the eyes more vulnerable to damage. The fast pace of modern life and high-intensity work pressure also increase the burden on the eyes, causing visual health problems to become more and more common.

[0003] In order to protect people's vision, especially children's vision, more frequent and more accurate vision tests are needed. In traditional vision testing methods, the person being tested mainly uses a vision chart to test their vision under the guidance of medical professionals in medical institutions such as hospitals and clinics. This testing method relies on the participation of professionals, so the labor cost of this method is high. Due to differences in personnel's technical capabilities, work experience and working conditions, the accuracy and precision of vision testing are insufficient, and consistency is lacking.

[0004] In order to solve the problem that traditional eye chart testing is completely dependent on manual work, technicians have developed an eye chart with automatic indication. Although there is no need for professionals to indicate the sight marks on the eye chart, the person being tested needs to use special equipment to input the interpretation results of the sight marks. This is different from the traditional method of indicating with hands, which changes the user's usage habits and increases the learning threshold, so it cannot meet actual needs. Summary of the invention

[0005] In order to overcome the above-mentioned technical problems existing in the prior art, the embodiments of the present invention provide a gesture action detection method and a gesture action detection device, which adopt a non-contact method to automatically identify, lock and gesture recognize the person being detected, thereby quickly generating vision detection results, greatly reducing the complexity of operation and improving user experience.

[0006] In order to achieve the above-mentioned purpose, an embodiment of the present invention provides a method for detecting gesture actions, which is applied to vision detection scenarios, and the method includes: obtaining monitoring information for a preset area in real time; judging whether there is a person being detected who triggers vision detection conditions based on the monitoring information; if so, locking the person being detected; obtaining gesture recognition information of the person being detected, and determining the detection result of the gesture action based on the gesture recognition information.

[0007] Preferably, the determining whether there is a person being detected who triggers the vision detection condition based on the monitoring information includes: determining a preset circled area, determining whether there is a person being detected standing in the preset circled area based on the monitoring information, and if there is a person being detected standing in the preset circled area, determining that the person being detected has triggered the vision detection condition; and / or: determining whether there is a person being detected wearing a specific identifier based on the monitoring information, the specific identifier having specified identifier information and / or specified shape features, and if there is a person being detected wearing the specific identifier, determining that the person being detected has triggered the vision detection condition; and / or: determining whether there is a person being detected who performs a set action based on the monitoring information, and if there is a person being detected who performs the set action, determining that the person being detected who performs the set action for the first time has triggered the vision detection condition.

[0008] Preferably, the method also includes: determining a random action before determining whether the person being detected performs a set action based on the monitoring information; determining guidance information based on the random action, and outputting the guidance information; determining whether the person being detected performs an action corresponding to the guidance information; and if so, determining that the person being detected performs the set action.

[0009] Preferably, locking the person being detected includes: acquiring the appearance features, facial features, arm features and torso joint features of the person being detected based on the monitoring information; establishing a physical continuity relationship between the appearance features and the facial features, the arm features and the torso joint features; and locking the person being detected based on the physical continuity relationship, the appearance features, the facial features, the arm features and the torso joint features.

[0010] Preferably, the acquiring of gesture recognition information of the person being detected and determining the detection result of the gesture action based on the gesture recognition information include: determining the gesture recognition information based on the arm features; comparing the gesture recognition information with preset gesture actions to determine the indication action; determining the detected eyes based on the facial features, and generating the detection result of the gesture action based on the indication action and the detected eyes.

[0011] Preferably, the method also includes: extracting eye surface features based on the facial features; judging whether the detected eye is wearing physical glasses based on the eye surface features; if so, determining that the detected eye is in a vision correction state, and generating a detection result of a gesture action based on the vision correction state, the indication action and the detected eye; otherwise, obtaining the historical naked eye vision value of the person being detected, and generating a detection result of a gesture action based on the historical naked eye vision value, the indication action and the detected eye.

[0012] Preferably, the method also includes: after generating the detection result, outputting inquiry information on whether to retest; obtaining feedback information for the inquiry information; if the feedback information indicates that retest is required, deleting the detection result, and re-executing the gesture action detection for the person being detected; if the feedback information indicates that retest is not required, judging whether gesture action detection for all eyes of the person being detected is completed: if gesture action detection for all eyes is completed, releasing the locked state of the person being detected; if gesture action detection for all eyes is not completed, executing gesture action detection for the next eye of the person being detected.

[0013] Preferably, the method also includes: judging whether the person being detected has walked out of the preset circled area based on the monitoring information, and generating a first judgment result; judging whether the brightness of the current detection environment has changed significantly based on the monitoring information, and generating a second judgment result; judging whether the person being detected is frequently blocked based on the monitoring information, and generating a third judgment result; judging whether the person being detected has cancelled eye blocking based on the monitoring information, and generating a fourth judgment result; judging whether the person being detected has removed the specific identifier based on the monitoring information, and generating a fifth judgment result; generating a locked state monitoring result based on the first judgment result and / or the second judgment result and / or the third judgment result and / or the fourth judgment result and / or the fifth judgment result.

[0014] Preferably, the method further comprises: judging whether there is a locking abnormality based on the locking state monitoring result; if so, performing credibility analysis on the detection result to generate an indication credibility; and generating a new detection result based on the indication credibility.

[0015] Correspondingly, the present invention also provides a gesture action detection device, which includes: an information acquisition unit, used to acquire monitoring information for a preset area in real time; a judgment unit, used to judge whether there is a condition for a person being detected to trigger a vision detection based on the monitoring information; a locking unit, used to lock the person being detected when there is a condition for a person being detected to trigger a vision detection; and an action determination unit, used to acquire gesture recognition information of the person being detected, and generate a detection result of the gesture action based on the gesture recognition information.

[0016] Through the technical solution provided by the present invention, the present invention has at least the following technical effects: By improving the existing vision detection method and adopting a contactless collection method, the monitoring information of the person being detected is automatically collected, and the person being detected is locked and gestures are recognized based on the monitoring information, thereby greatly improving the user experience of vision detection, improving the efficiency of vision detection, reducing the detection cost, and effectively improving the accuracy of vision detection.

[0017] Other features and advantages of the embodiments of the present invention will be described in detail in the subsequent detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The accompanying drawings are used to provide a further understanding of the embodiments of the present invention and constitute a part of the specification. Together with the following specific implementations, they are used to explain the embodiments of the present invention, but do not constitute a limitation on the embodiments of the present invention. In the accompanying drawings:

[0019] Figure 1 is a specific implementation flow chart of a gesture action detection method provided by an embodiment of the present invention;

[0020] Figure 2 It is a structural schematic diagram of a gesture action detection device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0021] The specific implementation of the embodiment of the present invention is described in detail below in conjunction with the accompanying drawings. It should be understood that the specific implementation described here is only used to illustrate and explain the embodiment of the present invention, and is not used to limit the embodiment of the present invention.

[0022] The terms "system" and "network" in the embodiments of the present invention can be used interchangeably. "Multiple" means two or more than two. In view of this, "multiple" can also be understood as "at least two" in the embodiments of the present invention. "And / or" describes the association relationship of associated objects, indicating that three relationships can exist. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / ", unless otherwise specified, generally indicates that the previous and subsequent associated objects are in an "or" relationship. In addition, it should be understood that in the description of the embodiments of the present invention, the words "first", "second", etc. are only used to distinguish the purpose of description, and cannot be understood as indicating or implying relative importance, nor can they be understood as indicating or implying order.

[0023] See also Figure 1 The embodiment of the present invention provides a method for detecting gestures, which is applied to vision detection scenarios. The method includes: S10: Acquire monitoring information for a preset area in real time; S20: determining whether there is a condition for the subject to trigger vision detection based on the monitoring information; S30: If yes, lock the detected person; S40: Acquire gesture recognition information of the detected person, and determine a detection result of the gesture action based on the gesture recognition information.

[0024] In a possible implementation, the embodiment of the present invention is applied to a vision test scenario of a vision chart with an automatic indication function. In this scenario, a gesture action detection method provided by the embodiment of the present invention is run in a processor with data processing capabilities. For example, the vision chart with an automatic indication function is a light box vision chart device with an automatic indication function of sight marks disclosed in patent 202321749533.0. Through this device, the person being tested can perform vision tests on their own, and no longer needs to be equipped with professionals on site for supervision and guidance, thereby greatly reducing the cost of vision testing and improving the convenience of vision testing.

[0025] Specifically, during the implementation process, the technician can pre-set the area where the person being tested needs to stand in the detection scene. For example, the preset area is the area 3-5m directly in front of the above-mentioned vision chart. The technician can use colored paint to mark an area on the ground in this area to prompt the person being tested to stand in the area, so as to facilitate locking and gesture recognition.

[0026] The vision detection method based on gesture recognition provided in an embodiment of the present invention can be applied to devices such as cameras and millimeter-wave radars that can be used to collect front image information or scanning information. The device can be set in an integrated manner on the above-mentioned vision chart and share the same power supply device to reduce the implementation complexity of the entire system; the device can also be set in an independent manner near the person being detected or a preset area, and face the person being detected to perform real-time collection of gesture movements. The device can be powered by a wired power supply device or a wireless power supply device, which will not be elaborated here.

[0027] During the application process, the device acquires monitoring information for the preset area in real time, including but not limited to image information, point cloud information, scanning information, etc. for the preset area. Then, it is determined whether the person being tested triggers vision testing conditions based on the monitoring information.

[0028] In actual applications, vision detection scenes are often public scenes, so it is inevitable that there are a large number of people standing in front of the device. Therefore, during the detection, if the person being detected is not accurately distinguished and identified, detection errors and misidentification will occur.

[0029] In an embodiment of the present invention, the determining whether there is a person being detected who triggers a vision detection condition based on the monitoring information includes: determining a preset circled area, determining whether there is a person being detected standing in the preset circled area based on the monitoring information, and if there is a person being detected standing in the preset circled area, determining that the person being detected has triggered the vision detection condition; and / or: determining whether there is a person being detected wearing a specific identifier based on the monitoring information, the specific identifier having specified identifier information or specified shape features, and if there is a person being detected wearing the specific identifier, determining that the person being detected has triggered the vision detection condition; and / or: determining whether there is a person being detected who performs a setting action based on the monitoring information, and if there is a person being detected who performs the setting action, determining the person being detected who performs the setting action for the first time as the person being detected who triggers the vision detection condition.

[0030] In a possible implementation, in order to reduce the complexity of user operations, improve user experience, and achieve seamless operation, the person being tested who triggers the vision detection conditions is automatically identified based on real-time collection of monitoring information, and gesture recognition is performed on the person.

[0031] In the first embodiment, a preset circled area is determined. For example, the preset circled area is an area pre-circled with a specific color by a technician according to an agreement, and areas circled with other colors will not be included in the preset circled area. During the monitoring process, it is determined in real time based on the monitoring information whether there is a person being tested standing in the preset circled area. For example, a common recognition algorithm (such as Yolov5 algorithm, etc.) can be used to identify the human body in the preset area, and when it is identified that there is a human body standing in the preset circled area for a preset time (such as 5s), it can be determined that the person being tested has triggered the vision detection condition.

[0032] In the second embodiment, a person who is about to undergo vision testing is required to wear a specific identifier having specified identification information and / or specified shape features, such as a badge or a mask having a QR code with authentication information printed on the surface. If it is detected that a person being tested is wearing the specific identifier, it can be determined that the person being tested has triggered the vision testing condition.

[0033] In the third embodiment, the monitoring information is used to identify in real time a person who performs a set action in a preset area. For example, the set action is a pre-fixed trigger action, such as drawing a circle with the hand, making a check with the hand, covering one eye with the hand, etc. If it is detected that the person being tested has performed the set action, it is determined that the person being tested who performs the set action for the first time has triggered the vision detection condition.

[0034] In actual application, only using any one of the above judgment methods requires sufficient recognition accuracy in specific scenarios. For open environments, there are more interferences in the scenes, which will greatly reduce the reliability of recognition. Technical personnel can use any combination of the above judgment conditions according to actual needs to effectively improve recognition accuracy.

[0035] In an embodiment of the present invention, by setting the trigger conditions before the vision test according to the user experience, the person being tested can achieve filtering and identification between the person being tested and the person not being tested on the basis of greatly reducing the complexity of the operation and satisfying the naturalness of their operation. After the person being tested is locked without feeling, the gesture-based vision test can be started, thereby greatly improving the user experience and detection accuracy of the vision test.

[0036] In actual application, if a fixed gesture is used, once the fixed gesture becomes well-known, the person being tested may be disturbed by other people during normal vision testing (for example, someone may perform a fixed gesture with the intention of disrupting the person), thereby reducing the user experience.

[0037] In order to solve the above technical problems, in an embodiment of the present invention, the method also includes: before judging whether there is a person being detected who performs a set action based on the monitoring information, determining a random action; determining guidance information based on the random action, and outputting the guidance information; judging whether there is a person being detected who performs an action corresponding to the guidance information; and if so, determining that there is a person being detected who performs the set action.

[0038] In a possible implementation, a user interaction device may be provided in the vision detection scene, and the user interaction device includes but is not limited to a speaker, a display screen, a diode display, etc., for providing interactive information such as voice and text to the person to be detected, so as to guide the user to enter the vision detection process. In combination with this embodiment, while waiting for the next person to be detected, a random action may be determined first, such as storing multiple actions in an action library in advance, and after the detection of the previous person to be detected is completed, an action is randomly determined as a random action that triggers the vision detection condition this time, and then the above-mentioned interactive device is used to broadcast it, so as to guide the next person to be detected to implement the random action, so as to enter the vision detection process. When it is detected that someone has implemented the current random action, it is determined that there is a person to be detected that triggers the vision detection condition.

[0039] In the embodiment of the present invention, by adopting the random action method, the external interference caused by the fixed trigger action can be effectively avoided, ensuring that the person being tested can be locked first, improving the user experience, and ensuring the accuracy of subsequent vision testing.

[0040] During the implementation process, due to the complex environment on site, such as the presence of a large number of bystanders, passers-by, and people playing around, all of which may cause interference such as occlusion, collision, and imitation of movements to the person being tested. For example, when the occluding person's body color difference is similar to that of the person being tested, or the occluding person's body partially overlaps with the person being tested, etc., it may cause the target to be lost or tracked incorrectly. Therefore, in order to ensure the reliability of vision testing, the person being tested needs to be locked.

[0041] In an embodiment of the present invention, locking the person being detected includes: acquiring the appearance features, facial features, arm features and torso joint features of the person being detected based on the monitoring information; establishing a physical continuity relationship between the appearance features and the facial features, the arm features and the torso joint features; and locking the person being detected based on the physical continuity relationship, the appearance features, the facial features, the arm features and the torso joint features.

[0042] In a possible implementation, after determining that there is a condition for the detected person to trigger the vision detection, the appearance features, facial features and arm features of the detected person are immediately extracted from the monitoring information, such as using an existing target recognition algorithm to extract the above features, and then establish a physical continuity relationship between the appearance features, facial features, arm features and trunk joint features, that is, in the subsequent recognition process, only the appearance, face, arm and trunk joints with the above physical continuity relationship are determined as the relevant human body parts of the detected person, and locked accordingly. In one embodiment, during the vision detection process, another person passes in front of the detected person, thereby forming an occlusion of the face and part of the body of the detected person, and may cause the loss of the gesture recognition target. At this time, by further combining the arm features with a physical connection relationship, the detected person is continuously locked to avoid erroneous detection caused by target loss. At the same time, when interference such as occlusion and collision is detected, the corresponding voice or text information can be output through the above interaction device to prompt nearby personnel to avoid or avoid interference in time.

[0043] After locking the person being tested, gesture recognition is performed on the person during the vision test. Specifically, in this embodiment, the visual chart with the visual mark indication function is used to provide the person being tested with the visual mark to be identified, and the person being tested performs gestures according to the visual mark. For example, in one embodiment, the visual chart indicates a visual mark in the 4.9 row. After viewing the visual mark, the person being tested believes that the direction of the visual mark is to the right, so he slides his arm to the right. At this time, gesture recognition information is determined based on the collected arm features. In an embodiment of the present invention, the gesture recognition information of the person being tested is obtained, and the detection result of the gesture action is determined based on the gesture recognition information, including: determining the gesture recognition information based on the arm features; comparing the gesture recognition information with the preset gesture action to determine the indication action; determining the detected eye based on the facial features, and generating the detection result of the gesture action based on the indication action and the detected eye.

[0044] Specifically, the corresponding bone information can be extracted according to the arm features, and the gesture recognition information can be determined according to the movement of the bone information. Then the gesture recognition information is compared with the preset gesture action to determine the indication action. The preset gesture action can be a gesture action set in advance by the technician according to the gesture command (including but not limited to up, down, left, right, cancel, confirm, etc.). By comparison, the indication action of the person being tested is determined. At this time, further, the currently blocked eye is determined according to the collected facial features. For example, if the currently blocked eye is the right eye, the indication action is an indication action for detecting the left eye. In the subsequent implementation process, the person being tested repeats the response of the gesture action according to the instructions of the eye chart until the prompt completes the gesture action detection of this eye, thereby generating the corresponding detection result, for example, the detection result is all the indication actions of the detected eye.

[0045] In the subsequent analysis process, the doctor or the host computer can determine the vision test result of the person being tested based on whether each gesture of the person being tested matches the actual visual mark. For example, in the first embodiment, each gesture of the person being tested is uploaded to the cloud server or the main control platform in real time, and the corresponding prior direction information on the vision chart is also uploaded to the cloud server or the main control platform in real time. After obtaining the above information, the cloud server or the main control platform automatically compares to determine whether each gesture of the person being tested is consistent with the prior direction; in the second embodiment, a processor is configured on the vision chart. After indicating the visual mark, the gesture of the person being tested is wirelessly sent to the processor of the vision chart. At this time, the processor compares the indicated direction of the visual mark with the gesture of the person being tested to verify whether the gesture matches. After completing all indication actions, the vision test result of the person being tested can be generated immediately.

[0046] During the application process, the person being tested may be required to enter personal information into the system in advance, including but not limited to headshot information, fingerprint information, iris information, RFID information, and other information used to characterize personal identity. After the above vision test results are generated, the vision test results can be automatically bound to the person being tested and sent to the doctor; if the person being tested has not bound personal information in advance, the vision test results can be cached after they are generated, and the person being tested can be prompted to record or claim the vision test results by themselves.

[0047] In actual application, since there are no professional personnel on site to supervise and manage, it is impossible to confirm whether the person being tested is in a state of vision correction, such as whether the vision test is performed while wearing glasses. This has a certain impact on doctors and reduces the accuracy of diagnosis.

[0048] In order to improve the accuracy of subsequent doctor's diagnosis, in an embodiment of the present invention, the method also includes: extracting eye surface features based on the facial features; judging whether the detected eye is wearing physical glasses based on the eye surface features; if so, determining that the detected eye is in a vision correction state, and generating a detection result of a gesture action based on the vision correction state, the indication action and the detected eye; otherwise, obtaining the historical naked eye vision value of the person being detected, and generating a detection result of a gesture action based on the historical naked eye vision value, the indication action and the detected eye.

[0049] In a possible implementation, during the process of performing vision testing on the subject, eye surface features are further extracted based on the collected facial features and analyzed to determine whether the subject is wearing physical glasses. The eye surface features include but are not limited to eye socket features, nose bridge features, eye corner features, etc. For example, a glasses recognition learning model can be pre-trained, and the eye socket features are analyzed by the glasses recognition learning model to determine whether the subject is wearing physical glasses. If so, it can be determined that the subject is wearing glasses for the vision test, so according to the vision correction state, the indicating action, and the detection result of the gesture action currently occluding the eyes, the detection result characterizes that the result is a detection result based on vision correction.

[0050] In another embodiment, it is not possible to identify whether the person being tested is wearing glasses through eye socket features, but he or she may be wearing contact lenses. Therefore, it is impossible to directly determine whether he or she is in a state of vision correction. Therefore, the historical naked eye vision value of the person being tested is automatically obtained, and the detection result of the gesture action is generated based on the historical naked eye vision value, the indication action and the current eye occlusion. Specifically, if the vision of the person being tested is significantly higher than his or her historical naked eye vision value, it can be determined that the detection result is a correction-based detection result.

[0051] In an embodiment of the present invention, by further analyzing whether the patient is in a state of vision correction during the vision test and generating detection results of gesture actions, precise data support is provided for subsequent doctors to conduct accurate vision analysis, thereby improving the accuracy of vision testing.

[0052] During the application process, the detection effect of the person being detected may not be ideal due to various reasons, such as collision interference, too many occlusions, false actions, etc. Therefore, it is necessary to determine whether the current detection result is qualified. If the current detection result is unqualified, retesting is required.

[0053] In an embodiment of the present invention, the method also includes: after generating the detection result, outputting inquiry information on whether to retest; obtaining feedback information for the inquiry information; if the feedback information indicates that retest is required, deleting the detection result, and re-executing the gesture action detection for the person being detected; if the feedback information indicates that retest is not required, judging whether the gesture action detection for all eyes of the person being detected is completed: if the gesture action detection for all eyes is completed, releasing the locked state of the person being detected; if the gesture action detection for all eyes is not completed, executing the gesture action detection for the next eye of the person being detected.

[0054] In the embodiment of the present invention, after the test of the person being tested is completed, it is confirmed whether a retest is needed to ensure that the test results of each vision test meet their subjective expectations and objective facts, avoid the generation of erroneous test results, and improve the accuracy of the test. On this basis, the locked state of the person being tested is released, and the preset area is automatically monitored in real time to respond at any time to enter the next test process.

[0055] Furthermore, according to the above description, due to the complexity of the testing site, the testing process may be disturbed and affected, and the reliability of the test results may be reduced. If the doctor does not have relevant information as a reference, he may give an incorrect diagnosis, thereby reducing the accuracy and reliability of vision testing.

[0056] To solve the above technical problems, in an embodiment of the present invention, the method also includes: judging whether the person being detected has walked out of the preset circled area based on the monitoring information, and generating a first judgment result; judging whether the brightness of the current detection environment has changed significantly based on the monitoring information, and generating a second judgment result; judging whether the person being detected is frequently blocked based on the monitoring information, and generating a third judgment result; judging whether the person being detected has cancelled the eye blocking based on the monitoring information, and generating a fourth judgment result; judging whether the person being detected has removed the specific identification object based on the monitoring information, and generating a fifth judgment result; generating a locked state monitoring result based on the first judgment result and / or the second judgment result and / or the third judgment result and / or the fourth judgment result and / or the fifth judgment result.

[0057] In one possible implementation, during the process of the subject's vision test, whether the subject is disturbed or affected is monitored in real time based on the monitoring information. Specifically, based on actual experience, multiple forms of possible interference or influence are determined, and each form is monitored. When it is identified that the subject is disturbed or affected by any of the above forms, a lock state monitoring result is generated.

[0058] Furthermore, in an embodiment of the present invention, the method also includes: judging whether there is a locking abnormality based on the locking status monitoring result; if so, performing credibility analysis on the detection result to generate an indication credibility; and generating a new detection result based on the indication credibility.

[0059] In a possible implementation, if the person being tested is subject to any of the above forms of interference or influence, it is determined that there is a locking anomaly, and a credibility analysis is immediately performed on the detection result of the gesture action generated this time. For example, a credibility analysis can be performed based on multiple factors such as the duration of the interference, the frequency, and the degree of influence on the detection result to generate an indication of credibility, and then the original detection result is updated based on the indication of credibility to generate a new detection result. In the subsequent diagnosis process, after obtaining the test results of the person being tested, the doctor analyzes the objective vision of the person being tested; on the other hand, the doctor evaluates the results based on their credibility during the detection process to provide a more accurate diagnosis conclusion, thereby effectively improving the accuracy of vision diagnosis and improving user experience.

[0060] On the other hand, see Figure 2An embodiment of the present invention further provides a gesture action detection device, the device comprising: an information acquisition unit, used to acquire monitoring information for a preset area in real time; a judgment unit, used to judge whether there is a condition for vision detection triggered by a person being detected based on the monitoring information; a locking unit, used to lock the person being detected when there is a condition for vision detection triggered by the person being detected; and an action determination unit, used to acquire gesture recognition information of the person being detected, and generate a detection result of the gesture action based on the gesture recognition information.

[0061] The optional implementation modes of the embodiments of the present invention are described in detail above in conjunction with the accompanying drawings. However, the embodiments of the present invention are not limited to the specific details in the above implementation modes. Within the technical concept of the embodiments of the present invention, various simple modifications can be made to the technical scheme of the embodiments of the present invention, and these simple modifications all belong to the protection scope of the embodiments of the present invention.

[0062] It should also be noted that the various specific technical features described in the above specific embodiments can be combined in any suitable manner without contradiction. To avoid unnecessary repetition, the embodiments of the present invention will not further describe various possible combinations.

[0063] Those skilled in the art can understand that all or part of the steps in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a program, and the program is stored in a storage medium, including a number of instructions to enable a single-chip microcomputer, a chip or a processor to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0064] In addition, various implementation modes of the embodiments of the present invention may be arbitrarily combined, and as long as they do not violate the concept of the embodiments of the present invention, they should also be regarded as the contents disclosed by the embodiments of the present invention.

Claims

1. A method for detecting gestures, applied to vision detection scenarios, characterized in that: The method comprises: Obtain monitoring information for preset areas in real time; Determining whether there is a condition for the subject to trigger vision detection based on the monitoring information; If yes, locking the detected person; Acquire gesture recognition information of the detected person, and determine a detection result of the gesture action based on the gesture recognition information.

2. The method according to claim 1, characterized in that The determining, based on the monitoring information, whether there is a condition for the subject to trigger vision detection includes: Determine a preset circled area, and determine whether a detected person is standing in the preset circled area based on the monitoring information, and if a detected person is standing in the preset circled area, determine that the detected person triggers a vision detection condition; and / or: Based on the monitoring information, it is determined whether the subject is wearing a specific identifier, the specific identifier having specified identification information and / or specified shape characteristics, and if the subject is wearing the specific identifier, it is determined that the subject triggers a vision detection condition; and / or: Based on the monitoring information, it is determined whether there is a person being tested who performs the setting action. If there is a person being tested who performs the setting action, it is determined that the person being tested who performs the setting action for the first time triggers the vision detection condition.

3. The method according to claim 2, characterized in that The method further comprises: Before determining whether the detected person performs a set action based on the monitoring information, determining a random action; determining guidance information based on the random action, and outputting the guidance information; Determining whether there is a detected person who performs an action corresponding to the guidance information; If so, it is determined that there is a person being detected who performs the set action.

4. The method according to claim 2, characterized in that: The step of locking the detected person comprises: Acquire the appearance features, facial features, arm features and trunk joint features of the detected person based on the monitoring information; Establishing a physical continuity relationship between the appearance feature and the facial feature, the arm feature and the trunk joint feature; The detected person is locked based on the physical continuity relationship, the appearance features, the facial features, the arm features and the torso joint features.

5. The method according to claim 4, characterized in that The acquiring the gesture recognition information of the detected person and determining the detection result of the gesture action based on the gesture recognition information includes: determining gesture recognition information based on the arm features; Comparing the gesture recognition information with a preset gesture action to determine an indicated action; A detected eye is determined based on the facial feature, and a detection result of a gesture action is generated based on the indicating action and the detected eye.

6. The method according to claim 5, characterized in that The method further comprises: extracting ocular surface features based on the facial features; Determining whether the detected eye is wearing glasses based on the eye surface features; If yes, determine that the detected eye is in a vision correction state, and generate a detection result of the gesture action based on the vision correction state, the indicating action, and the detected eye; Otherwise, the historical naked eye vision value of the person being tested is obtained, and a detection result of the gesture action is generated based on the historical naked eye vision value, the indicating action and the detected eye.

7. The method according to claim 1, characterized in that The method further comprises: After generating the test result, outputting inquiry information on whether to retest; Obtaining feedback information regarding the inquiry information; If the feedback information indicates that retesting is required, deleting the test result, and re-performing the gesture detection for the person being tested; If the feedback information indicates that retesting is not required, determining whether the gesture detection for all eyes of the detected person is completed: If the gesture detection for all eyes is completed, the locked state of the detected person is released; If the gesture detection for all eyes is not completed, the gesture detection for the next eye of the detected person is performed.

8. The method according to claim 2, characterized in that: The method further comprises: Based on the monitoring information, determine whether the detected person has walked out of the preset circled area, and generate a first determination result; Based on the monitoring information, determining whether the brightness of the current detection environment has changed significantly, and generating a second determination result; Determining whether the detected person is frequently blocked based on the monitoring information, and generating a third determination result; Determine whether the detected person has cancelled the eye covering based on the monitoring information, and generate a fourth determination result; Determine whether the detected person has removed the specific identification object based on the monitoring information, and generate a fifth determination result; A locking state monitoring result is generated based on the first judgment result and / or the second judgment result and / or the third judgment result and / or the fourth judgment result and / or the fifth judgment result.

9. The method according to claim 8, characterized in that The method further comprises: Determining whether there is a locking abnormality based on the locking status monitoring result; If so, performing credibility analysis on the detection result to generate an indication credibility; A new detection result is generated based on the indication credibility.

10. A gesture detection device, characterized in that: The device comprises: An information acquisition unit, used to acquire monitoring information for a preset area in real time; A judgment unit, used for judging whether a condition for vision detection is triggered by the person being detected based on the monitoring information; A locking unit, used to lock the subject when the subject triggers a vision detection condition; The action determination unit is used to obtain the gesture recognition information of the detected person and determine the detection result of the gesture action based on the gesture recognition information.

Images