Gain adjustment system and method based on picture ratio of monitoring picture human shape model

Abstract

The invention discloses a gain adjustment system and method based on the picture ratio of a human shape model of a monitoring picture, and the system is applied to a network monitoring camera, and comprises a camera which is used for capturing the monitoring picture; the microphone pickup module is used for picking up sound signals so as to realize clear voice communication in the two-way talkback module; the two-way talkback module is used for realizing two-way voice communication between the network monitoring camera and a user; the AI human shape detection module is used for detecting a human shape model in the monitoring picture; the calculation module is used for calculating the distance to the camera according to the proportion of the human shape model in the monitoring picture; and the gain adjustment module is used for dynamically adjusting the gain of the microphone pickup module according to the distance obtained by the calculation module. According to the method and the device, the gain of the microphone can be effectively adjusted under the condition that the human shape is detected or not detected so as to adapt to different monitoring scenes and requirements, so that users can obtain the optimal hearing or visual experience.

Description

Technical Field

[0001] The present invention relates to the technical field of network cameras, and in particular to a gain adjustment system and method based on a humanoid model screen ratio of a monitoring screen. Background Art

[0002] Existing network surveillance camera systems are usually equipped with fixed microphone gain settings. The microphone has a default gain that can provide clear voice communication within a specific distance range. However, when the distance of the person changes, the fixed gain setting may cause the audio quality to degrade. For example, when the person is close to the camera, a fixed high gain setting may cause the sound to be distorted or too loud; and when the person is far away from the camera, insufficient gain may cause the sound signal to be weak and difficult to hear. Therefore, the existing technology cannot dynamically adapt to the changes in the distance of the person in different monitoring scenarios, affecting the practicality and user experience of the two-way intercom function. Summary of the invention

[0003] In view of the problems existing in the prior art, the present invention provides a gain adjustment system and method based on the proportion of the human model screen in the monitoring screen. First, the MIC pickup module is started and initialized to ensure that the audio input is ready. Subsequently, the system enters the two-way intercom mode to provide support for real-time communication. Immediately afterwards, the AI ​​human detection module is activated and starts to perform the human detection task. Once a human figure is detected, the system will calculate the proportion of the human figure in the screen, which is to more accurately evaluate the distance between the human figure and the camera. By traversing the human figure proportion data, the system can accurately find the corresponding distance information. If the AI ​​human detection module fails to detect a human figure, the system will directly query the distance and gain table and output a preset gain value. Regardless of whether a human figure is detected, the system will adjust the gain of the audio or video according to the queried distance information to optimize the output effect. This intelligent adjustment ensures that users can get the best auditory or visual experience in different scenarios. Finally, when the gain adjustment is completed, the whole process ends smoothly, and the system waits for the next initialization instruction. The design of this process reflects the advantages of intelligence and automation, and improves the response speed and user experience of the system.

[0004] In order to achieve the above object, the present invention provides a gain adjustment system based on the proportion of a human-shaped model in a monitoring screen, which is applied to a network monitoring camera and includes:

[0005] Camera, used to capture surveillance images;

[0006] A microphone pickup module is used to pick up sound signals so as to achieve clear voice communication in a two-way intercom module;

[0007] Two-way intercom module, used to realize two-way voice communication between network surveillance cameras and users;

[0008] AI humanoid detection module, used to detect humanoid models in surveillance images;

[0009] A calculation module, used to calculate the distance from the camera according to the proportion of the humanoid model in the monitoring screen;

[0010] The gain adjustment module is used to dynamically adjust the gain of the microphone pickup module according to the distance obtained by the calculation module.

[0011] Preferably, the AI ​​humanoid detection module includes:

[0012] An image processing unit, used for preprocessing the monitoring images captured by the camera;

[0013] A deep learning algorithm unit for identifying and locating a humanoid model from the processed image;

[0014] A human figure detection algorithm unit is used to perform human figure detection tasks and identify human figures in the picture;

[0015] A bounding box determination unit, used to determine the bounding box of each human figure in the picture;

[0016] A feature extraction unit for extracting key features from a human figure so as to more accurately identify and track the human figure;

[0017] The tracking unit is used to track the position and movement of the human figure in the monitoring image to maintain continuous detection of the human figure.

[0018] Preferably, the deep learning algorithm unit is trained to recognize a variety of different human postures and shapes; the human detection algorithm unit is capable of distinguishing between human figures and other objects to reduce false detections.

[0019] Preferably, the feature extraction unit uses computer vision technology to extract features of the human figure, including contour, color and texture information; the tracking unit uses a motion estimation algorithm to track the movement of the human figure between consecutive frames; and the bounding box defines the vertical and horizontal range of the human figure.

[0020] Preferably, the calculation module includes:

[0021] A proportion calculation unit is used to calculate the area of ​​each figure's bounding box and compare it with the area of ​​the monitoring screen to obtain the proportion of the figure in the screen;

[0022] A data traversal unit, used for traversing a preset correspondence table between proportion and distance;

[0023] The distance matching unit is used to compare the real-time calculated proportions with the proportions in the table to find the closest match, thereby determining the exact distance between the human figure and the camera.

[0024] Preferably, the gain adjustment system based on the screen ratio of the humanoid model of the monitoring screen also includes a user interface for displaying the monitoring screen and providing an interactive interface for system operation.

[0025] The present invention also provides a gain adjustment method based on the screen ratio of a humanoid model in a monitoring screen, the method comprising the following steps:

[0026] Step S1: Initialization step, initializing the MIC pickup module to ensure that the audio input is ready;

[0027] Step S2 is an intercom mode step, entering a two-way intercom mode to provide support for real-time communication;

[0028] Step S3: AI detection step, activating the AI ​​human figure detection module and starting to perform the human figure detection task;

[0029] Step S4: If a human figure is detected, a proportion calculation is performed, and the proportion of the human figure model in the picture is calculated by the calculation module, and the distance between the human figure model and the camera is evaluated, and then the correspondence table of the evaluated distance and gain value is queried to adjust the gain of the audio or video to optimize the output effect;

[0030] If no human figure is detected, the preset distance and gain value correspondence table is directly queried to adjust the gain of the audio or video to optimize the output effect;

[0031] Step S5: After the gain adjustment is completed, the whole process ends and waits for the next initialization instruction.

[0032] Preferably, the human figure detection task of the AI ​​human figure detection module in step S3 includes the following steps:

[0033] Step S301: using an image processing unit to pre-process the surveillance image captured by the camera;

[0034] Step S302: using a deep learning algorithm unit to identify and locate a humanoid model from the processed image;

[0035] Step S303: using a human figure detection algorithm unit to perform a human figure detection task to identify a human figure in the picture;

[0036] Step S304: using a bounding box determination unit to determine a bounding box of each human figure in the picture, the bounding box defining the vertical and horizontal range of the human figure;

[0037] Step S305: extracting key features from the human figure using a feature extraction unit so as to more accurately identify and track the human figure;

[0038] Step S306: Using a tracking unit to track the position and movement of the human figure in the monitoring image to maintain continuous detection of the human figure.

[0039] Preferably, in step S4, the specific steps of calculating the proportion of the human figure in the picture and evaluating the distance between the human figure and the camera are as follows:

[0040] Step S401: using a proportion calculation unit to calculate the area of ​​each figure boundary box, and compare it with the area of ​​the monitoring screen to obtain the proportion of the figure in the screen;

[0041] Step S402: using a data traversal unit to traverse a preset correspondence table between proportions and distances;

[0042] Step S403: Use the distance matching unit to compare the proportion calculated in real time with the proportion in the table to find the closest match, thereby determining the exact distance between the human figure and the camera.

[0043] Preferably, when no human figure is detected, a preset distance and gain value correspondence table is queried, comprising the following steps:

[0044] Default strategy step S404, selecting a default distance value and a corresponding gain value according to preset parameters of the camera and environmental conditions;

[0045] In step S405 of querying a gain table, a preset distance-gain value correspondence table is searched, the table including microphone gain settings at different distances;

[0046] In the gain adjustment step S406, the gain of the microphone is adjusted according to the queried default distance value to ensure that a reasonable audio collection effect is maintained when no human figure is detected.

[0047] The technical solution of the present invention has the following beneficial effects:

[0048] The present invention can effectively adjust the gain of the microphone in both the case of detecting a human figure and the case of not detecting a human figure, so as to adapt to different monitoring scenes and requirements. Such a design not only improves the adaptability and flexibility of the system, but also enables users to obtain the best hearing or visual experience.

[0049] The present invention can dynamically adjust the gain of the microphone according to the proportion of the human figure in the monitoring screen. The human figure in the screen is identified by the AI ​​human figure detection module, and its proportion is calculated, and the system can evaluate the distance between the human figure and the camera. Then, the calculation module queries the appropriate gain value from the preset distance and gain value corresponding table based on this distance information, and the gain adjustment module adjusts the gain of the microphone in real time to optimize the pickup of the audio signal; when the AI ​​human figure detection module fails to detect the human figure, the system will directly query the distance and gain table and output a preset gain value to adjust the gain of the audio or video to optimize the output effect. This design enables the two-way intercom function of the network surveillance camera to adapt to different monitoring scenes and changes in the distance between the speaker and the camera, significantly improving the clarity of voice communication and the user's auditory experience. In addition, the present invention also improves the intelligence level of the system, reduces the audio signal quality problems caused by environmental changes or human movement, and enhances the adaptability and reliability of the system.

[0050] Intelligent audio adjustment: Human shape detection is achieved through the AI ​​human shape detection module, and the audio gain is intelligently adjusted according to the proportion of the human shape in the picture, improving the intelligence level of audio processing.

[0051] Improved real-time performance: After detecting a human figure, the system quickly calculates the proportion and adjusts the gain, reducing human intervention and improving the system's real-time response capabilities.

[0052] Optimized user experience: Regardless of whether a human figure is detected or not, the system can automatically adjust the gain to ensure that users can get the best hearing experience in different scenarios.

[0053] Automated process: The entire process from initialization to gain adjustment is completed automatically, reducing operational complexity and improving the ease of use of the system.

[0054] Scenario adaptability: The system can adapt to different usage scenarios and provide appropriate gain adjustment strategies regardless of whether human detection succeeds or fails.

[0055] The system is able to dynamically adjust the gain of the microphone to adapt to the changes in the distance between the human figure and the camera, thereby optimizing the pickup of sound signals from users in the monitoring area. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0057] The present invention is further described below in conjunction with the accompanying drawings and specific embodiments.

[0058] Reference Figure 1The present invention provides a gain adjustment system based on the proportion of a humanoid model in a monitoring screen, which is applied to a network monitoring camera and includes:

[0059] Camera, used to capture surveillance images;

[0060] A microphone pickup module is used to pick up sound signals so as to achieve clear voice communication in a two-way intercom module;

[0061] Two-way intercom module, used to realize two-way voice communication between network surveillance cameras and users;

[0062] AI humanoid detection module, used to detect humanoid models in surveillance images;

[0063] A calculation module, used to calculate the distance from the camera according to the proportion of the humanoid model in the monitoring screen;

[0064] The gain adjustment module is used to dynamically adjust the gain of the microphone pickup module according to the calculated distance.

[0065] Furthermore, the AI ​​humanoid detection module includes:

[0066] The image processing unit is used to pre-process the surveillance images captured by the camera, including but not limited to denoising, contrast enhancement, color correction and resolution adjustment. These pre-processing steps ensure that subsequent human detection and recognition work is performed on the basis of optimized image quality, thereby improving the accuracy and efficiency of detection.

[0067] A deep learning algorithm unit for identifying and locating humanoid models from processed images. The unit is capable of identifying and locating humanoids in a variety of postures and shapes, and maintains high accuracy even in complex backgrounds or lighting conditions.

[0068] The human figure detection algorithm unit is used to perform human figure detection tasks and identify human figures in the picture. The human figure detection algorithm unit is used to perform human figure detection tasks in the image processed by the deep learning algorithm unit. It can identify human figures in the picture and distinguish between human figures and other objects to reduce false detections, and provide accurate human figure position information to provide a basis for subsequent bounding box determination and feature extraction.

[0069] The bounding box determination unit is used to determine the bounding box of each human figure in the picture; the bounding box defines the vertical and horizontal range of the human figure.

[0070] The feature extraction unit is used to extract key features from the human figure so as to more accurately identify and track the human figure; the feature extraction unit uses computer vision technology to extract features of the human figure, including contour, color and texture information, which are used to improve the accuracy of human figure recognition and support long-term tracking of the movement of the human figure in the picture.

[0071] A tracking unit is used to track the position and movement of a human figure in the monitoring image to maintain continuous detection of the human figure. The tracking unit uses a motion estimation algorithm to track the movement of the human figure between consecutive frames to ensure that the system continuously detects and analyzes the human figure, and can maintain tracking even when the human figure moves or the image changes.

[0072] Through the collaborative work of the above units, the AI ​​humanoid detection module can accurately detect and identify humanoids in complex monitoring environments, providing real-time and precise data support for the system's gain adjustment. This not only optimizes the pickup of audio signals, but also improves the intelligence level of the monitoring system and the user's interactive experience. In addition, the design of the module takes into account the adaptability of different environments and scenarios, ensuring the stability and reliability of the system under various conditions.

[0073] Furthermore, the calculation module includes:

[0074] The proportion calculation unit is used to calculate the area of ​​each figure's bounding box and compare it with the area of ​​the monitoring screen to obtain the proportion of the figure in the screen. The coordinates of the bounding box can be obtained through image processing technology, and its area can be calculated and compared with the area of ​​the entire monitoring screen to obtain the proportion of the figure in the screen.

[0075] The data traversal unit is used to traverse a preset table of proportions and distances; the unit is responsible for traversing a preset table of proportions and distances, which is based on the viewing angle and resolution parameters of the camera, and the standard proportions of the human figure in the picture at different distances.

[0076] The distance matching unit is used to compare the real-time calculated proportions with the proportions in the table to find the closest match, thereby determining the exact distance between the human figure and the camera. This process is the key to dynamically adjusting the gain value, ensuring that the system can adjust the audio gain in real time according to the changes in the distance of the human figure and optimize the pickup of the sound signal.

[0077] In this embodiment, the calculation module works closely with the AI ​​humanoid detection module to receive the humanoid bounding box information provided by the bounding box determination unit. Once the AI ​​humanoid detection module identifies and determines the humanoid bounding box, the calculation module immediately starts its calculation process to ensure that the system can quickly respond to changes in the position and distance of the humanoid. In addition, the output of the calculation module will be directly input into the gain adjustment module to adjust the gain of the microphone pickup module. If the humanoid is close to the camera, the gain will be reduced to avoid excessive sound; if the humanoid is far away, the gain will be increased to ensure the clarity of the sound signal. Through the precise calculation of the calculation module, the system can dynamically adjust the audio gain according to the proportion of the humanoid in the monitoring screen, thereby providing the best voice communication experience in different monitoring scenarios. This intelligent adjustment process not only improves the adaptability of the system, but also improves the practicality and user satisfaction of the surveillance camera in the two-way intercom function.

[0078] Furthermore, the gain adjustment system based on the screen ratio of the humanoid model of the monitoring screen also includes a user interface for displaying the monitoring screen and providing an interactive interface for system operation. The user interface displays the monitoring screen captured by the camera in real time, allowing the user to view the situation in the current monitoring area. The display function may include multi-screen display, screen zooming, and switching of different camera screens. System status indication: The interface provides real-time feedback on the system status, including audio connection status, AI detection status, gain adjustment information, etc., to ensure that the user can understand the working status of the system in real time.

[0079] The user interface works closely with the two-way intercom module, AI humanoid detection module, calculation module, and gain adjustment module. For example, when the user starts the two-way intercom through the interface, the two-way intercom module is activated, and the AI ​​humanoid detection module starts to analyze the monitoring screen. The calculation module calculates the humanoid proportion and distance based on the results of AI detection, and then the gain adjustment module adjusts the gain of the microphone based on the calculation results. In addition, the user interface can also display the calculation results provided by the calculation module, such as the humanoid proportion and estimated distance, to provide users with visual data support. Through the intuitive design and ease of use of the user interface, users can easily operate the system, monitor and adjust settings in real time. This not only improves the user's operating efficiency, but also enhances the user's sense of control and satisfaction with the system. The design of the user interface takes into account the skill levels and usage habits of different users, ensuring the ease of use and popularity of the system.

[0080] The present invention also provides a gain adjustment method based on the screen ratio of a humanoid model in a monitoring screen, the method comprising the following steps:

[0081] Step S1: Initialization step, initializing the MIC pickup module to ensure that the audio input is ready; this step ensures that the system can receive clear voice signals and provides a basis for subsequent voice communication.

[0082] Step S2 intercom mode step, enters two-way intercom mode to provide support for real-time communication; the system switches to two-way intercom mode, allowing real-time voice communication between the network surveillance camera and the user; activation of this mode provides necessary support for real-time communication and ensures smooth communication.

[0083] Step S3: AI detection step, activating the AI ​​human figure detection module and starting to perform the human figure detection task;

[0084] Step S4: If a human figure is detected, a proportion calculation is performed, and the proportion of the human figure model in the picture is calculated by the calculation module, and the distance between the human figure model and the camera is evaluated, and then the correspondence table of the evaluated distance and gain value is queried to adjust the gain of the audio or video to optimize the output effect;

[0085] If no human figure is detected, the preset distance and gain value correspondence table is directly queried to adjust the gain of the audio or video to optimize the output effect;

[0086] Step S5: After the gain adjustment is completed, the whole process ends and waits for the next initialization instruction;

[0087] Through this method, the network surveillance camera can intelligently adjust the audio gain in different surveillance scenarios, and can provide clear voice communication quality whether the human figure is close or far away. In addition, when no human figure is detected, the method can maintain a reasonable audio collection effect by querying the preset gain value, thereby improving the adaptability of the system and user experience.

[0088] Furthermore, in step S3, the human figure detection task of the AI ​​human figure detection module includes the following steps:

[0089] Step S301: Use an image processing unit to pre-process the surveillance image captured by the camera; the image processing unit pre-processes the surveillance image captured by the camera to enhance the image quality and prepare for subsequent human detection. The pre-processing includes operations such as denoising, brightness adjustment, contrast enhancement, color correction and image sharpening.

[0090] Step S302: using a deep learning algorithm unit to identify and locate a humanoid model from the processed image;

[0091] Step S303: Use the human figure detection algorithm unit to perform the human figure detection task and identify the human figure in the picture; the human figure detection algorithm unit uses the output of the deep learning algorithm to determine whether there is a human figure in the picture and identify the human figure. This step ensures the accuracy of human figure detection.

[0092] Step S304: using a bounding box determination unit to determine a bounding box of each human figure in the picture, the bounding box defining the vertical and horizontal range of the human figure;

[0093] Step S305: extract key features from the human figure using a feature extraction unit to more accurately identify and track the human figure; the feature extraction unit extracts key features from the identified human figure, such as contour, color, texture, and motion features. These features are used to improve the accuracy of human figure recognition and support long-term tracking of the human figure;

[0094] Step S306: Use a tracking unit to track the position and movement of the human figure in the monitoring image to maintain continuous detection of the human figure. The tracking unit ensures that the system can maintain recognition and analysis of the human figure even when the human figure moves or the image changes.

[0095] Through the collaborative work of the above steps, the AI ​​human detection module can accurately detect and identify human figures in complex monitoring environments, providing real-time and accurate data support for the system's gain adjustment. This not only optimizes the pickup of audio signals, but also improves the intelligence level of the monitoring system and the user's interactive experience.

[0096] Furthermore, in step S4, the specific steps of calculating the proportion of the human figure in the picture and evaluating the distance between the human figure and the camera are as follows:

[0097] Step S401: using a proportion calculation unit to calculate the area of ​​each figure boundary box, and compare it with the area of ​​the monitoring screen to obtain the proportion of the figure in the screen;

[0098] Step S402: using a data traversal unit to traverse a preset table of correspondence between proportion and distance, the table is based on the viewing angle and resolution parameters of the camera, and the standard proportion of the human figure in the picture at different distances, and provides necessary reference information for the distance matching unit;

[0099] Step S403: Use the distance matching unit to compare the proportion calculated in real time with the proportion in the table to find the closest match, thereby determining the exact distance between the human figure and the camera; this step is the basis for dynamic gain adjustment, ensuring that the system can adjust the audio gain in real time according to the change in the distance of the human figure and optimize the pickup of the sound signal.

[0100] Through the precise calculation of the computing module, the system can dynamically adjust the audio gain according to the proportion of human figures in the monitoring screen, thereby providing the best voice communication experience in different monitoring scenarios. In addition, the design of this module takes into account the adaptability of different environments and scenarios, ensuring the stability and reliability of the system under various conditions. Through the collaborative work of these steps, the computing module not only optimizes the pickup of audio signals, but also improves the intelligence level of the monitoring system and the user's interactive experience.

[0101] Furthermore, if no human figure is detected, a preset distance and gain value correspondence table is queried, including the following steps:

[0102] Default strategy step S404, selects a default distance value and a corresponding gain value according to the preset parameters of the camera and environmental conditions; the system automatically selects a default distance value according to the preset parameters of the camera, such as focal length, viewing angle, and environmental conditions, such as light intensity and noise level. At the same time, the system will select a corresponding gain value based on historical data and machine learning algorithms. This step ensures that even if a human figure is not detected, the system can provide a reasonable starting point for adjusting the audio gain.

[0103] In step S405 of querying the gain table, a preset distance-gain value correspondence table is queried, which contains microphone gain settings at different distances. These settings are obtained based on a large amount of experimental data and scene simulations, and are intended to cover various possible monitoring environments and camera configurations.

[0104] The gain adjustment step S406 adjusts the gain of the microphone according to the default distance value queried to ensure that a reasonable audio collection effect can be maintained when no human figure is detected. This step is dynamic, allowing the system to maintain a reasonable audio collection effect according to the preset distance value when no human figure is detected.

[0105] In summary, through these steps, the system is able to maintain the quality and clarity of audio acquisition even when no human figure is detected. This improves the robustness and adaptability of the system and ensures stable audio output in various monitoring scenarios. In addition, this approach reduces the system's dependence on human detection and enhances the system's reliability in complex environments.

[0106] From the above embodiments, it can be seen that the present invention can effectively adjust the gain of the microphone in both the case of detecting a human figure and the case of not detecting a human figure, so as to adapt to different monitoring scenarios and requirements. Such a design not only improves the adaptability and flexibility of the system, but also allows users to obtain the best hearing or visual experience.

[0107] The present invention can dynamically adjust the gain of the microphone according to the proportion of the human figure in the monitoring screen. The human figure in the screen is identified by the AI ​​human figure detection module, and its proportion is calculated, and the system can evaluate the distance between the human figure and the camera. Then, the calculation module queries the appropriate gain value from the preset distance and gain value corresponding table based on this distance information, and the gain adjustment module adjusts the gain of the microphone in real time to optimize the pickup of the audio signal; when the AI ​​human figure detection module fails to detect the human figure, the system will directly query the distance and gain table and output a preset gain value to adjust the gain of the audio or video to optimize the output effect. This design enables the two-way intercom function of the network surveillance camera to adapt to different monitoring scenes and changes in the distance between the speaker and the camera, significantly improving the clarity of voice communication and the user's auditory experience. In addition, the present invention also improves the intelligence level of the system, reduces the audio signal quality problems caused by environmental changes or human movement, and enhances the adaptability and reliability of the system.

[0108] Intelligent audio adjustment: Human shape detection is achieved through the AI ​​human shape detection module, and the audio gain is intelligently adjusted according to the proportion of the human shape in the picture, improving the intelligence level of audio processing.

[0109] Improved real-time performance: After detecting a human figure, the system quickly calculates the proportion and adjusts the gain, reducing human intervention and improving the system's real-time response capabilities.

[0110] Optimized user experience: Regardless of whether a human figure is detected or not, the system can automatically adjust the gain to ensure that users can get the best hearing experience in different scenarios.

[0111] Automated process: The entire process from initialization to gain adjustment is completed automatically, reducing operational complexity and improving the ease of use of the system.

[0112] Scenario adaptability: The system can adapt to different usage scenarios and provide appropriate gain adjustment strategies regardless of whether human detection succeeds or fails.

[0113] The system is able to dynamically adjust the gain of the microphone to adapt to the changes in the distance between the human figure and the camera, thereby optimizing the pickup of sound signals from users in the monitoring area.

[0114] The above description is only a preferred embodiment of the present invention, and does not limit the patent scope of the present invention. All equivalent structural changes made by using the contents of the present invention specification and drawings under the inventive concept of the present invention, or directly / indirectly applied in other related technical fields are included in the patent protection scope of the present invention.

Claims

1. A gain adjustment system based on the proportion of a humanoid model in a monitoring screen, applied to a network monitoring camera, characterized in that: include: Camera, used to capture surveillance images; A microphone pickup module is used to pick up sound signals so as to realize voice communication in a two-way intercom module; Two-way intercom module, used to realize two-way voice communication between network surveillance cameras and users; AI humanoid detection module, used to detect humanoid models in surveillance images; A calculation module, used to calculate the distance from the camera according to the proportion of the humanoid model in the monitoring screen; The gain adjustment module is used to dynamically adjust the gain of the microphone pickup module according to the distance obtained by the calculation module.

2. The gain adjustment system based on the proportion of the human-shaped model on the monitoring screen according to claim 1 is characterized in that: AI human detection module includes: An image processing unit, used for preprocessing the monitoring images captured by the camera; A deep learning algorithm unit for identifying and locating a humanoid model from the processed image; A human figure detection algorithm unit is used to perform human figure detection tasks and identify human figures in the picture; A bounding box determination unit, used to determine the bounding box of each human figure in the picture; A feature extraction unit for extracting key features from a human figure so as to more accurately identify and track the human figure; The tracking unit is used to track the position and movement of the human figure in the monitoring image to maintain continuous detection of the human figure.

3. The gain adjustment system based on the proportion of the human-shaped model on the monitoring screen according to claim 2 is characterized in that: The deep learning algorithm unit is trained to recognize a variety of different human postures and shapes; the human detection algorithm unit can distinguish between human figures and other objects to reduce false detections.

4. The gain adjustment system based on the proportion of the human-shaped model on the monitoring screen according to claim 2 is characterized in that: The feature extraction unit uses computer vision technology to extract features of the human figure, including contour, color and texture information; the tracking unit uses a motion estimation algorithm to track the movement of the human figure between consecutive frames; and the bounding box defines the vertical and horizontal range of the human figure.

5. The gain adjustment system based on the proportion of the human-shaped model in the monitoring screen according to claim 1 is characterized in that: The calculation module comprises: A proportion calculation unit is used to calculate the area of ​​each figure's bounding box and compare it with the area of ​​the monitoring screen to obtain the proportion of the figure in the screen; A data traversal unit, used for traversing a preset correspondence table between proportion and distance; The distance matching unit is used to compare the real-time calculated proportions with the proportions in the table to find the closest match, thereby determining the exact distance between the human figure and the camera.

6. The gain adjustment system based on the proportion of the humanoid model in the monitoring screen according to claim 1 is characterized in that: The gain adjustment system based on the screen ratio of the humanoid model of the monitoring screen also includes a user interface for displaying the monitoring screen and providing an interactive interface for system operation.

7. The gain adjustment method based on the screen ratio of the human-shaped model of the monitoring screen according to any one of claims 1 to 6, characterized in that: The method comprises the following steps: Step S1: Initialization step, initializing the MIC pickup module to ensure that the audio input is ready; Step S2 is an intercom mode step, entering a two-way intercom mode to provide support for real-time communication; Step S3: AI detection step, activating the AI ​​human figure detection module and starting to perform the human figure detection task; Step S4: If a human figure is detected, a proportion calculation is performed, and the proportion of the human figure model in the picture is calculated by the calculation module, and the distance between the human figure model and the camera is evaluated, and then the correspondence table of the evaluated distance and gain value is queried to adjust the gain of the audio or video to optimize the output effect; If no human figure is detected, the preset distance and gain value correspondence table is directly queried to adjust the gain of the audio or video to optimize the output effect; Step S5: After the gain adjustment is completed, the whole process ends and waits for the next initialization instruction.

8. The method according to claim 7 is characterized in that: In step S3, the human figure detection task of the AI ​​human figure detection module includes the following steps: Step S301: using an image processing unit to pre-process the surveillance image captured by the camera; Step S302: using a deep learning algorithm unit to identify and locate a humanoid model from the processed image; Step S303: using a human figure detection algorithm unit to perform a human figure detection task to identify a human figure in the picture; Step S304: using a bounding box determination unit to determine a bounding box of each human figure in the picture, the bounding box defining the vertical and horizontal range of the human figure; Step S305: extracting key features from the human figure using a feature extraction unit so as to more accurately identify and track the human figure; Step S306: Using a tracking unit to track the position and movement of the human figure in the monitoring image to maintain continuous detection of the human figure.

9. The method according to claim 7 is characterized in that: In step S4, the calculation module calculates the proportion of the human figure in the picture and evaluates the distance between the human figure and the camera. The specific steps are as follows: Step S401: using a proportion calculation unit to calculate the area of ​​each figure boundary box, and compare it with the area of ​​the monitoring screen to obtain the proportion of the figure in the screen; Step S402: using a data traversal unit to traverse a preset correspondence table between proportions and distances; Step S403: Use the distance matching unit to compare the proportion calculated in real time with the proportion in the table to find the closest match, thereby determining the exact distance between the human figure and the camera.

10. The gain adjustment method based on the screen ratio of the human-shaped model of the monitoring screen according to claim 7 is characterized in that: If no human figure is detected in step S4, a preset distance and gain value correspondence table is queried, including the following steps: Default strategy step S404, selecting a default distance value and a corresponding gain value according to preset parameters of the camera and environmental conditions; In step S405 of querying a gain table, a preset distance-gain value correspondence table is searched, the table including microphone gain settings at different distances; In the gain adjustment step S406, the gain of the microphone is adjusted according to the queried default distance value to ensure that a reasonable audio collection effect is maintained when no human figure is detected.

Images