Large screen control method and terminal
By combining voiceprint recognition and gesture recognition, using wake-up voice and camera to detect standing images, generating control commands to control the large screen, solving the problem of large screen control efficiency and accuracy, and achieving contactless and efficient control.
Patent Information
- Application Number
- CN202310501121.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2023-05-06
- Publication Date
- 2024-11-08
AI Technical Summary
In the control of large-screen terminals, the prior art has problems such as low control efficiency and difficult to accurately identify gesture recognition, especially in noise interference and complex environments, resulting in low control accuracy and efficiency.
By obtaining the user's wake-up voice, determining its voiceprint feature vector, and combining the camera to collect the image to be tested for stand detection and gesture recognition, target control commands are generated to control the large screen, achieving contactless control and avoiding environmental interference.
It improves the accuracy and efficiency of large-screen control, ensures the accuracy and reliability of control commands, and is suitable for long-distance control scenarios such as education.
Smart Images

Figure CN118918888A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of intelligent terminals, and in particular relates to a large-screen control method and a terminal. Background Art
[0002] With the vigorous development of information technology, large-screen smart terminals with advantages in information presentation and human-computer interaction have been widely used in various social places. Especially in the field of education, with the popularization of large-screen education on the market, teachers have realized paperless teaching, replaced the traditional teaching mode, and created an intelligent education environment. In the use of large-screen terminal equipment, common human-computer interaction methods include mouse, keyboard, touch and other operation methods.
[0003] However, in order to ensure the accuracy of the mouse on a large screen, the mouse movement speed must not be too fast, which will lead to inefficient mouse control. The finger touch and swipe control method not only requires standing in front of the large screen, which easily blocks the screen and affects the audience's viewing experience, but also requires swiping over a large area in front of the huge screen, and the experience is far inferior to small-screen devices such as mobile phones and tablets.
[0004] Therefore, in this scenario, long-distance contactless control of the large screen is a more appropriate control method. Common long-distance contactless control methods include voice control and gesture control. Voice control has the advantages of fast speed, high efficiency, and can be controlled at any position at a long distance without blocking the screen. However, voice control is also prone to noise interference, which leads to a decline in experience. It is not flexible enough in text processing, and some simple instructions may be ambiguous. Gesture control is easy to operate and is suitable for controlling scenes such as document slides. However, due to the long distance, the gestures are easily interfered by a large amount of background and difficult to recognize. Gestures suitable for operators are often too simple and easy to be misrecognized. At the same time, complex gestures are not easily accepted by operators. Summary of the invention
[0005] The technical problem to be solved by the present invention is to provide a large-screen control method and terminal, which can effectively improve the accuracy and efficiency of control.
[0006] In order to solve the above technical problems, a technical solution adopted by the present invention is:
[0007] A large screen control method, comprising the steps of:
[0008] Obtaining a user's wake-up voice and determining a voiceprint feature vector corresponding to the wake-up voice;
[0009] Determine whether the voiceprint feature vector corresponding to the wake-up voice exists in a preset voiceprint library, and if so, collect a frame of the image to be tested of the user;
[0010] Determine whether the image to be tested is a standing image, and if so, perform gesture recognition based on the standing image to obtain a gesture recognition result;
[0011] A target control instruction is generated according to the gesture recognition result, and the large screen is controlled according to the target control instruction.
[0012] In order to solve the above technical problems, another technical solution adopted by the present invention is:
[0013] A large-screen control terminal includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the following steps when executing the computer program:
[0014] Obtaining a user's wake-up voice and determining a voiceprint feature vector corresponding to the wake-up voice;
[0015] Determine whether the voiceprint feature vector corresponding to the wake-up voice exists in a preset voiceprint library, and if so, collect a frame of the image to be tested of the user;
[0016] Determine whether the image to be tested is a standing image, and if so, perform gesture recognition based on the standing image to obtain a gesture recognition result;
[0017] A target control instruction is generated according to the gesture recognition result, and the large screen is controlled according to the target control instruction.
[0018] The beneficial effects of the present invention are: obtaining the user's wake-up voice, and determining the voiceprint feature vector corresponding to the wake-up voice, when the voiceprint feature vector corresponding to the wake-up voice exists in the preset voiceprint library, collecting a frame of the user's image to be tested, when the image to be tested is a standing image, performing gesture recognition based on it, obtaining a gesture recognition result, generating a target control instruction according to the gesture recognition result, and controlling the large screen according to the target control instruction, thereby controlling the large screen by combining voiceprint recognition, standing detection and gesture recognition, realizing contactless large-screen control, and avoiding as much as possible the problem of difficulty in accurately identifying control instructions caused by objective environmental interference factors, thereby effectively improving the accuracy and efficiency of control. BRIEF DESCRIPTION OF THE DRAWINGS
[0019] Figure 1 This is a flowchart of a large screen control method according to an embodiment of the present invention;
[0020] Figure 2 This is a schematic diagram of the structure of a large-screen control terminal according to an embodiment of the present invention;
[0021] Figure 3 This is a control flow chart of a large-screen control method according to an embodiment of the present invention;
[0022] Figure 4 This is a flowchart of the wake-up word + voice command control in the large-screen control method of an embodiment of the present invention;
[0023] Figure 5 This is a control flow chart of wake-up word + standing detection + gesture recognition in the large-screen control method of an embodiment of the present invention. DETAILED DESCRIPTION
[0024] In order to explain the technical content, achieved objectives and effects of the present invention in detail, the following is an explanation in conjunction with the implementation modes and the accompanying drawings.
[0025] Please refer to Figure 1 , an embodiment of the present invention provides a large screen control method, comprising the steps of:
[0026] Obtaining a user's wake-up voice and determining a voiceprint feature vector corresponding to the wake-up voice;
[0027] Determine whether the voiceprint feature vector corresponding to the wake-up voice exists in a preset voiceprint library, and if so, collect a frame of the image to be tested of the user;
[0028] Determine whether the image to be tested is a standing image, and if so, perform gesture recognition based on the standing image to obtain a gesture recognition result;
[0029] A target control instruction is generated according to the gesture recognition result, and the large screen is controlled according to the target control instruction.
[0030] From the above description, it can be seen that the beneficial effects of the present invention are: obtaining the user's wake-up voice, and determining the voiceprint feature vector corresponding to the wake-up voice, when the voiceprint feature vector corresponding to the wake-up voice exists in the preset voiceprint library, collecting a frame of the user's image to be tested, when the image to be tested is a standing image, performing gesture recognition based on it, obtaining a gesture recognition result, generating a target control instruction according to the gesture recognition result, and controlling the large screen according to the target control instruction, thereby controlling the large screen by combining voiceprint recognition, standing detection and gesture recognition to realize contactless large-screen control, and avoiding as much as possible the problem of difficulty in accurately identifying control instructions caused by objective environmental interference factors, thereby effectively improving the accuracy and efficiency of control.
[0031] Further, the determining whether the image to be tested is a standing image, and if so, performing gesture recognition based on the standing image to obtain a gesture recognition result includes:
[0032] Perform standing detection on the image to be tested, and determine whether there is a standing person in the image to be tested. If yes, determine the image to be tested as a standing image, and obtain multiple consecutive frames of standing images corresponding to the standing image;
[0033] Extracting a plurality of consecutive frames of standing person images of the user from the plurality of consecutive frames of standing images;
[0034] Gesture recognition is performed based on multiple consecutive frames of standing person images of the user to obtain a gesture recognition result.
[0035] From the above description, it can be seen that in the teaching scenario, the teacher is at a long distance relative to the large screen, and there are interferences from a large number of people in the background and a complex environment. By performing standing detection on a captured frame of the image to be tested, the user who is standing in the current scene can be identified. This user is the user who needs to control the large screen. Subsequently, only separate gesture recognition is performed on this user. This can support long-distance and accurate gesture recognition, eliminate the interference of the external environment on gesture recognition, and thus improve the accuracy and efficiency of large-screen control.
[0036] Furthermore, the performing gesture recognition based on the continuous multiple-frame standing person images of the user to obtain the gesture recognition result includes:
[0037] Sorting the plurality of consecutive frames of standing person images in chronological order to obtain sorted standing person images;
[0038] It is determined whether any preset dynamic gesture exists in the sorted standing person images. If so, the preset dynamic gesture existing in the sorted standing person images is used as a gesture recognition result.
[0039] From the above description, it can be seen that the user can use preset dynamic gestures to control the large screen. First, multiple consecutive frames of standing person images are sorted in chronological order, and then gesture recognition is performed. This can more accurately identify the preset dynamic gestures in the image, improve the recognition accuracy, and thus improve the accuracy of control.
[0040] Furthermore, before obtaining the user's wake-up voice and determining the voiceprint feature vector corresponding to the wake-up voice, the method includes:
[0041] Determining a plurality of preset dynamic gestures and control instructions corresponding to the preset dynamic gestures;
[0042] Generating a target control instruction according to the gesture recognition result includes:
[0043] Generate a target control instruction corresponding to the preset dynamic gesture present in the sorted standing person image.
[0044] From the above description, it can be seen that each preset dynamic gesture corresponds to a control instruction, and a target control instruction corresponding to the preset dynamic gesture in the sorted standing person image is generated. The control of the large screen can be completed according to the instruction, thereby realizing the large screen control simply and efficiently.
[0045] Furthermore, before obtaining the user's wake-up voice and determining the voiceprint feature vector corresponding to the wake-up voice, the method includes:
[0046] Obtaining a voiceprint registration request from a user, where the voiceprint registration request includes the user's voice and user personal information;
[0047] Extracting a voiceprint feature vector of the user's voice according to the voiceprint registration request;
[0048] Binding the voiceprint feature vector of the user's voice with the user's personal information to obtain the bound voiceprint feature vector and user's personal information;
[0049] The bound voiceprint feature vector and user personal information are saved in a preset voiceprint library.
[0050] From the above description, it can be seen that the user's voiceprint registration is performed in advance to generate a preset voiceprint library. Only users whose voiceprint feature vectors are stored in the preset voiceprint library can perform contactless operation of the large screen, preventing other unauthorized persons in the scene from arbitrarily operating the large screen, thereby ensuring the reliability and accuracy of the large screen control.
[0051] Please refer to Figure 2 , a large-screen control terminal, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the following steps when executing the computer program:
[0052] Obtaining a user's wake-up voice and determining a voiceprint feature vector corresponding to the wake-up voice;
[0053] Determine whether the voiceprint feature vector corresponding to the wake-up voice exists in a preset voiceprint library, and if so, collect a frame of the image to be tested of the user;
[0054] Determine whether the image to be tested is a standing image, and if so, perform gesture recognition based on the standing image to obtain a gesture recognition result;
[0055] A target control instruction is generated according to the gesture recognition result, and the large screen is controlled according to the target control instruction.
[0056] From the above description, it can be seen that the beneficial effects of the present invention are: obtaining the user's wake-up voice, and determining the voiceprint feature vector corresponding to the wake-up voice, when the voiceprint feature vector corresponding to the wake-up voice exists in the preset voiceprint library, collecting a frame of the user's image to be tested, when the image to be tested is a standing image, performing gesture recognition based on it, obtaining a gesture recognition result, generating a target control instruction according to the gesture recognition result, and controlling the large screen according to the target control instruction, thereby controlling the large screen by combining voiceprint recognition, standing detection and gesture recognition to realize contactless large-screen control, and avoiding as much as possible the problem of difficulty in accurately identifying control instructions caused by objective environmental interference factors, thereby effectively improving the accuracy and efficiency of control.
[0057] Further, the determining whether the image to be tested is a standing image, and if so, performing gesture recognition based on the standing image to obtain a gesture recognition result includes:
[0058] Perform standing detection on the image to be tested, and determine whether there is a standing person in the image to be tested. If yes, determine the image to be tested as a standing image, and obtain multiple consecutive frames of standing images corresponding to the standing image;
[0059] Extracting a plurality of consecutive frames of standing person images of the user from the plurality of consecutive frames of standing images;
[0060] Gesture recognition is performed based on multiple consecutive frames of standing person images of the user to obtain a gesture recognition result.
[0061] From the above description, it can be seen that in the teaching scenario, the teacher is at a long distance relative to the large screen, and there are interferences from a large number of people in the background and a complex environment. By performing standing detection on a captured frame of the image to be tested, the user who is standing in the current scene can be identified. This user is the user who needs to control the large screen. Subsequently, only separate gesture recognition is performed on this user. This can support long-distance and accurate gesture recognition, eliminate the interference of the external environment on gesture recognition, and thus improve the accuracy and efficiency of large-screen control.
[0062] Furthermore, the performing gesture recognition based on the continuous multiple-frame standing person images of the user to obtain the gesture recognition result includes:
[0063] Sorting the plurality of consecutive frames of standing person images in chronological order to obtain sorted standing person images;
[0064] It is determined whether any preset dynamic gesture exists in the sorted standing person images. If so, the preset dynamic gesture existing in the sorted standing person images is used as a gesture recognition result.
[0065] From the above description, it can be seen that the user can use preset dynamic gestures to control the large screen. First, multiple consecutive frames of standing person images are sorted in chronological order, and then gesture recognition is performed. This can more accurately identify the preset dynamic gestures in the image, improve the recognition accuracy, and thus improve the accuracy of control.
[0066] Furthermore, before obtaining the user's wake-up voice and determining the voiceprint feature vector corresponding to the wake-up voice, the method includes:
[0067] Determining a plurality of preset dynamic gestures and control instructions corresponding to the preset dynamic gestures;
[0068] Generating a target control instruction according to the gesture recognition result includes:
[0069] Generate a target control instruction corresponding to the preset dynamic gesture present in the sorted standing person image.
[0070] From the above description, it can be seen that each preset dynamic gesture corresponds to a control instruction, and a target control instruction corresponding to the preset dynamic gesture in the sorted standing person image is generated. The control of the large screen can be completed according to the instruction, thereby realizing the large screen control simply and efficiently.
[0071] Furthermore, before obtaining the user's wake-up voice and determining the voiceprint feature vector corresponding to the wake-up voice, the method includes:
[0072] Obtaining a voiceprint registration request from a user, where the voiceprint registration request includes the user's voice and user personal information;
[0073] Extracting a voiceprint feature vector of the user's voice according to the voiceprint registration request;
[0074] Binding the voiceprint feature vector of the user's voice with the user's personal information to obtain the bound voiceprint feature vector and user's personal information;
[0075] The bound voiceprint feature vector and user personal information are saved in a preset voiceprint library.
[0076] From the above description, it can be seen that the user's voiceprint registration is performed in advance to generate a preset voiceprint library. Only users whose voiceprint feature vectors are stored in the preset voiceprint library can perform contactless operation of the large screen, preventing other unauthorized persons in the scene from arbitrarily operating the large screen, thereby ensuring the reliability and accuracy of the large screen control.
[0077] The large-screen control method and terminal of the present invention can be applied to scenarios where a large screen needs to be controlled, such as large-screen control scenarios in the education industry, and are described below through specific implementation methods:
[0078] Embodiment 1
[0079] Please refer to Figure 1 , Figure 3-Figure 5 , a large screen control method of this embodiment includes the steps of:
[0080] S1. Determine a plurality of preset dynamic gestures and control instructions corresponding to the preset dynamic gestures.
[0081] The preset dynamic gestures can be flexibly set according to actual conditions; in an optional implementation, five preset dynamic gestures and control instructions corresponding to the preset dynamic gestures are determined, as shown in Table 1, so that the gesture recognition algorithm can be trained according to the preset dynamic gestures to improve the efficiency and accuracy of gesture recognition;
[0082] Table 1 Preset dynamic gestures and control instructions corresponding to the preset dynamic gestures
[0083]
[0084]
[0085] S2. Obtain a voiceprint registration request from the user, where the voiceprint registration request includes the user's voice and personal information;
[0086] In an optional implementation, the user personal information includes name.
[0087] S3. Extracting a voiceprint feature vector of the user's voice according to the voiceprint registration request;
[0088] S4, binding the voiceprint feature vector of the user's voice with the user's personal information to obtain the bound voiceprint feature vector and user's personal information;
[0089] S5. Save the bound voiceprint feature vector and user personal information to a preset voiceprint library, such as Figure 3 shown.
[0090] In an optional implementation, the user's voiceprint registration must belong to the same channel source as the device used in subsequent actual use, to ensure the accuracy of subsequent voiceprint feature vector judgment.
[0091] S6. Obtain the user's wake-up voice and determine the voiceprint feature vector corresponding to the wake-up voice, such as Figure 1 As shown;
[0092] The wake-up voice is a preset wake-up word spoken by the user.
[0093] S7, determine whether the voiceprint feature vector corresponding to the wake-up voice exists in the preset voiceprint library, if so, collect a frame of the user's image to be tested, if not, return to S6, such as Figure 3 As shown;
[0094] In an optional implementation, a frame of the image to be tested of the user is captured by a camera at a rate of 1FPS.
[0095] S8, determining whether the image to be tested is a standing image, if so, performing gesture recognition based on the standing image to obtain a gesture recognition result, such as Figure 5 As shown, specifically including:
[0096] S81, performing standing detection on the image to be tested, and determining whether there is a standing person in the image to be tested, if yes, executing S811-S813, if no, repeating S81;
[0097] S811, determining the image to be tested as a standing image, and acquiring a plurality of consecutive frames of standing images corresponding to the standing image;
[0098] In an optional implementation, a plurality of continuous frames of standing images corresponding to the standing image are acquired at a rate of 5 FPS through a camera.
[0099] S812, extracting a plurality of consecutive frames of standing person images of the user from the plurality of consecutive frames of standing images;
[0100] The standing person image is an image captured along the boundary of the standing person in the standing image.
[0101] In an optional embodiment, if multiple consecutive frames of standing person images of the user cannot be extracted within a preset time, multiple consecutive frames of standing images corresponding to the standing images will no longer be acquired. The preset time is 2 seconds. That is to say, if no standing person is detected within 2 seconds, gesture recognition will no longer be performed.
[0102] S813, performing gesture recognition based on the continuous multiple-frame standing person images of the user to obtain a gesture recognition result, specifically including:
[0103] S8131, sorting the plurality of consecutive frames of standing person images in chronological order to obtain sorted standing person images;
[0104] S8132. Determine whether there is any preset dynamic gesture in the sorted standing person image. If yes, take the preset dynamic gesture in the sorted standing person image as the gesture recognition result. If no, return to execute S812.
[0105] S9, generating a target control instruction according to the gesture recognition result, and controlling the large screen according to the target control instruction;
[0106] Specifically, a target control instruction corresponding to a preset dynamic gesture existing in the sorted standing person image is generated, and the large screen is controlled according to the target control instruction.
[0107] In an optional implementation, the large screen can also be remotely controlled using a wake-up word + voice command, such as Figure 4 As shown, S7-S9 can be replaced by:
[0108] S7, determining whether the voiceprint feature vector corresponding to the wake-up voice exists in a preset voiceprint library, if so, collecting the user's voice command, if not, returning to S6;
[0109] S8. Recognize the voice command, obtain the target control instruction, and control the large screen according to the target control instruction.
[0110] Embodiment 2
[0111] This embodiment applies the large screen control method described in Embodiment 1 to a specific application scenario, as follows:
[0112] Scenario 1: The teacher uses the educational large screen that communicates with the large screen control terminal of the present invention to teach students. During the teaching process, the teacher wants to show the next page of the document to the students, but he is far away from the screen, and it takes time to return to the large screen control console computer to operate, and it will also cause the lecture to be interrupted. At this time, the teacher first speaks the wake-up voice for human-computer interaction, and uses voice commands to wake up the screen working state and switch it to the human-computer interaction state. Then he makes a gesture command to the camera: "Raise your right hand above your waist, open your palm and pause for 1 second, then slide up". After the terminal of the present invention recognizes and obtains the wake-up voice, it first calls the camera and the sound pickup mark to identify the user, that is, the person who issues the command, determines the voiceprint feature vector corresponding to the wake-up voice, and compares the voiceprint feature vector with the preset voiceprint library to confirm whether the person who issues the command has obtained the use permission, that is, whether it has been registered. After confirming that there is permission to use, the gesture action of the person who issues the command is identified, and the target control instruction is generated according to the gesture recognition result. The target control instruction is the command text that slides on the current document. Then, the text is converted into a pre-defined control instruction through semantic analysis and sent to the control system of the large screen. Finally, the control system performs the operation of turning up one page for the document to be demonstrated according to the control instruction. After completing the gesture command, the system automatically exits the human-computer interaction state.
[0113] Scenario 2: The teacher uses an educational large screen that communicates with the large screen control terminal of the present invention to teach students. During the course of teaching, the teacher wants to close the currently demonstrated document "xxx". The wake-up voice issued by the teacher wakes up the screen in a human-computer interaction state, and makes an action of "making a thumb gesture with the right hand, placing it in front of the chest and pausing for 1 second". After the terminal of the present invention recognizes the wake-up voice, it first calls the camera and the pickup mark to identify the person who issued the command, and compares the voiceprint feature vector with the preset voiceprint library to confirm whether the person who issued the command has obtained the use permission, that is, whether he has registered. After confirming that he has the use permission, the action of the person who issued the command is identified, and the gesture command is recognized through gesture recognition to execute the corresponding control action of exiting the editing mode of the document "xxx".
[0114] This way, there is no need to pause between gestures and voice commands, allowing efficient and accurate large-screen control.
[0115] Embodiment 3
[0116] Please refer to Figure 2 A resume generation terminal of this embodiment includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, each step of the resume generation method in the first embodiment is implemented.
[0117] In summary, the present invention provides a resume generation method and terminal, which obtain the user's wake-up voice and determine the voiceprint feature vector corresponding to the wake-up voice; determine whether the voiceprint feature vector corresponding to the wake-up voice exists in a preset voiceprint library, and if so, collect a frame of the user's image to be tested; determine whether the image to be tested is a standing image, and if so, perform gesture recognition based on the standing image to obtain a gesture recognition result; generate a target control instruction according to the gesture recognition result, and control the large screen according to the target control instruction. By combining voiceprint recognition, standing detection and gesture recognition to control the large screen, contactless large screen control is achieved, and the problem of difficulty in accurately identifying control instructions caused by objective environmental interference factors is avoided as much as possible, thereby effectively improving the accuracy and efficiency of control; the user can use preset dynamic gestures to control the large screen, first sort multiple consecutive frames of standing person images in chronological order, and then perform gesture recognition, which can more accurately identify the preset dynamic gestures in the image, improve the recognition accuracy, and thus improve the accuracy of control.
[0118] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent transformations made using the contents of the present invention's specification and drawings, or directly or indirectly applied in related technical fields, are also included in the patent protection scope of the present invention.
Claims
1. A large screen control method, characterized in that: Includes steps: Obtaining a user's wake-up voice and determining a voiceprint feature vector corresponding to the wake-up voice; Determine whether the voiceprint feature vector corresponding to the wake-up voice exists in a preset voiceprint library, and if so, collect a frame of the image to be tested of the user; Determine whether the image to be tested is a standing image, and if so, perform gesture recognition based on the standing image to obtain a gesture recognition result; A target control instruction is generated according to the gesture recognition result, and the large screen is controlled according to the target control instruction.
2. A large screen control method according to claim 1, characterized in that: The determining whether the image to be tested is a standing image, and if so, performing gesture recognition based on the standing image to obtain a gesture recognition result includes: Perform standing detection on the image to be tested, and determine whether there is a standing person in the image to be tested. If yes, determine the image to be tested as a standing image, and obtain multiple consecutive frames of standing images corresponding to the standing image; Extracting a plurality of consecutive frames of standing person images of the user from the plurality of consecutive frames of standing images; Gesture recognition is performed based on multiple consecutive frames of standing person images of the user to obtain a gesture recognition result.
3. A large screen control method according to claim 2, characterized in that: The performing gesture recognition based on the continuous multiple-frame standing person images of the user to obtain the gesture recognition result includes: Sorting the plurality of consecutive frames of standing person images in chronological order to obtain sorted standing person images; It is determined whether any preset dynamic gesture exists in the sorted standing person images. If so, the preset dynamic gesture existing in the sorted standing person images is used as a gesture recognition result.
4. A large screen control method according to any one of claims 1 to 3, characterized in that: Before obtaining the user's wake-up voice and determining the voiceprint feature vector corresponding to the wake-up voice, the method includes: Determining a plurality of preset dynamic gestures and control instructions corresponding to the preset dynamic gestures; Generating a target control instruction according to the gesture recognition result includes: Generate a target control instruction corresponding to the preset dynamic gesture present in the sorted standing person image.
5. A large screen control method according to claim 1, characterized in that: Before obtaining the user's wake-up voice and determining the voiceprint feature vector corresponding to the wake-up voice, the method includes: Obtaining a voiceprint registration request from a user, where the voiceprint registration request includes the user's voice and user personal information; Extracting a voiceprint feature vector of the user's voice according to the voiceprint registration request; Binding the voiceprint feature vector of the user's voice with the user's personal information to obtain the bound voiceprint feature vector and user's personal information; The bound voiceprint feature vector and user personal information are saved in a preset voiceprint library.
6. A large-screen control terminal, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the following steps are implemented: Obtaining a user's wake-up voice and determining a voiceprint feature vector corresponding to the wake-up voice; Determine whether the voiceprint feature vector corresponding to the wake-up voice exists in a preset voiceprint library, and if so, collect a frame of the image to be tested of the user; Determine whether the image to be tested is a standing image, and if so, perform gesture recognition based on the standing image to obtain a gesture recognition result; A target control instruction is generated according to the gesture recognition result, and the large screen is controlled according to the target control instruction.
7. A large-screen control terminal according to claim 6, characterized in that: The determining whether the image to be tested is a standing image, and if so, performing gesture recognition based on the standing image to obtain a gesture recognition result includes: Perform standing detection on the image to be tested, and determine whether there is a standing person in the image to be tested. If yes, determine the image to be tested as a standing image, and obtain multiple consecutive frames of standing images corresponding to the standing image; Extracting a plurality of consecutive frames of standing person images of the user from the plurality of consecutive frames of standing images; Gesture recognition is performed based on multiple consecutive frames of standing person images of the user to obtain a gesture recognition result.
8. A large-screen control terminal according to claim 7, characterized in that: The performing gesture recognition based on the continuous multiple-frame standing person images of the user to obtain the gesture recognition result includes: Sorting the plurality of consecutive frames of standing person images in chronological order to obtain sorted standing person images; It is determined whether any preset dynamic gesture exists in the sorted standing person images. If so, the preset dynamic gesture existing in the sorted standing person images is used as a gesture recognition result.
9. A large-screen control terminal according to claim 8, characterized in that: Before obtaining the user's wake-up voice and determining the voiceprint feature vector corresponding to the wake-up voice, the method includes: Determining a plurality of preset dynamic gestures and control instructions corresponding to the preset dynamic gestures; Generating a target control instruction according to the gesture recognition result includes: Generate a target control instruction corresponding to the preset dynamic gesture present in the sorted standing person image.
10. A large-screen control terminal according to claim 6, characterized in that: Before obtaining the user's wake-up voice and determining the voiceprint feature vector corresponding to the wake-up voice, the method includes: Obtaining a voiceprint registration request from a user, where the voiceprint registration request includes the user's voice and user personal information; Extracting a voiceprint feature vector of the user's voice according to the voiceprint registration request; Binding the voiceprint feature vector of the user's voice with the user's personal information to obtain the bound voiceprint feature vector and user's personal information; The bound voiceprint feature vector and user personal information are saved in a preset voiceprint library.
