CN212456080U

CN212456080U - Terminal support based on AI visual convolution neural network technique

Info

Publication number: CN212456080U
Application number: CN202020859071.8U
Authority: CN
Inventors: 张业超; 刘峰; 王振鹏
Original assignee: Shangshi Hongtu Shenzhen Technology Co ltd
Current assignee: Shangshi Hongtu Shenzhen Technology Co ltd
Priority date: 2020-05-20
Filing date: 2020-05-20
Publication date: 2021-02-02
Anticipated expiration: 2030-05-20

Abstract

The utility model provides a terminal support based on AI vision convolution neural network technique, include: a holding member for holding the fixed terminal; the camera is arranged on the clamping component and basically has the same direction as the clamping component, and is used for acquiring environmental information; the base bears the clamping component and is in rotary connection with the clamping component; the driver is used for driving the clamping component to rotate relative to the base; and the AI chip module is provided with a visual convolution neural network model and is used for controlling the driver to rotate so as to enable the camera to follow the human body structure when the environmental information acquired by the camera comprises the human body structure information through calculation and judgment of the convolution neural network model. The utility model discloses can follow the user, improve terminal bracket's intelligence and practicality.

Description

Terminal support based on AI visual convolution neural network technique

Technical Field

The utility model relates to a support especially relates to a terminal support based on AI vision convolution neural network technique.

Background

The terminal support is a device for clamping and fixing the terminal, and the application field of the terminal support is wide, and the terminal support can be applied to live video or video shooting. The existing support can only be shot by the clamped terminal device through the active angle adjustment of a user, and is very inconvenient.

SUMMERY OF THE UTILITY MODEL

In view of this, in order to solve one of the technical problems in the related art to a certain extent, it is necessary to provide a terminal support based on an AI visual convolutional neural network technology, which can follow a user and improve intelligence and practicability of the terminal support by applying the existing AI technology to the terminal support.

The utility model provides a terminal support based on AI vision convolution neural network technique, include:

a holding member for holding the fixed terminal;

the camera is arranged on the clamping component and basically has the same direction as the clamping component, and is used for acquiring environmental information;

the base bears the clamping component and is in rotary connection with the clamping component;

the driver is used for driving the clamping component to rotate relative to the base;

and the AI chip module is provided with a visual convolution neural network model and is used for controlling the driver to rotate so as to enable the camera to follow the human body structure when the environmental information acquired by the camera comprises the human body structure information through calculation and judgment of the convolution neural network model.

Further, the AI chip module comprises a CMOS image sensor and a GPU for operating a convolutional neural network model, and the CMOS image sensor and the GPU are combined together in an SOC mode.

Further, the human body structure comprises a human face and/or a human figure.

Further, when the environment information acquired by the camera does not include the human body structure information through the calculation of the convolutional neural network model and is judged, the driver is controlled to continuously rotate so as to detect the human body structure.

Further, the terminal support still includes the adapter, the adapter is used for gathering environmental sound information, AI chip module still has AI voice model for according to the environmental sound information that the adapter gathered to the terminal sends and takes a picture or video instruction.

Furthermore, the base is provided with an interface for connecting a tripod.

According to the above technical scheme, the utility model discloses fix the terminal on clamping part, calculate and judge whether including human structure information in the environmental information that the camera acquireed through AI chip module, when calculating and judging including human structure information in the environmental information that the camera acquireed through convolution neural network model, then rotate so that the camera follows human structure through the control driver, and then make and fix the terminal camera on the support by the centre gripping can follow the user so that the user is shot all the time.

Drawings

Fig. 1 is a schematic structural diagram of the present invention.

Fig. 2 is a schematic circuit diagram of the present invention.

Fig. 3 is a specific circuit diagram of the present invention.

The following detailed description will further illustrate the invention in conjunction with the above-described figures.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments of the present invention without any inventive step, are within the scope of the present invention. It is to be understood that the drawings are provided solely for the purposes of reference and illustration and are not intended as a definition of the limits of the invention. The connection relationships shown in the drawings are for clarity of description only and do not limit the manner of connection.

It will be understood that when an element is referred to as being "connected" to another element, it can be directly connected to the other element or intervening elements may also be present. Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. It should also be noted that, unless expressly stated or limited otherwise, the terms "mounted," "connected," and "connected" are to be construed broadly and can include, for example, fixed connections, removable connections, or integral connections; either mechanically or electrically, and may be internal to both elements. The specific meanings of the above terms in the present invention can be understood in specific cases to those skilled in the art. The terminology used in the description of the invention herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention.

It should be noted that in the description of the present invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc. indicate orientations or positional relationships based on the orientations or positional relationships shown in the drawings, which are only for convenience of describing the present invention and simplifying the description, but do not indicate or imply that the referred device or element must have a specific orientation, be configured in a specific orientation, and operate, and thus, should not be construed as limiting the present invention. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance.

As shown in fig. 1 and fig. 2, the utility model provides a terminal bracket based on AI visual convolution neural network technique is applied to terminal bracket with AI visual convolution neural network technique.

The terminal support includes: a holding member 10 for holding the fixed terminal; the camera 20 is arranged on the clamping component 10 and has the same direction as the clamping component 10, and the camera 20 is used for acquiring environmental information; a base 30 for carrying the clamping member 10 and rotatably connecting with the clamping member 10; an actuator 40 for driving the clamping member 10 to rotate relative to the base 30; the AI chip module 50 has a visual convolutional neural network model, and is configured to control the driver 40 to rotate so that the camera 20 follows the human body structure when it is calculated and determined through the convolutional neural network model that the environmental information acquired by the camera 20 includes the human body structure information.

The base 30 may serve as a support for supporting the clamping member 10, and an interface (which may be provided on a bottom surface of the base 30) for connecting a tripod may be provided on the base 30 to connect the base 30 to a tripod.

In the present invention, the AI chip module 50 may adopt an NB1001 chip, and a specific circuit structure is shown in fig. 3. The AI chip module 50 comprises a CMOS image sensor and a GPU running a convolutional neural network model, the CMOS image sensor and the GPU are combined together in an SOC mode, space required by system construction is simplified, lower power consumption can be provided, the size required by mainboard design can be reduced as much as possible, the area required by PCB manufacturing is reduced, and the system is more environment-friendly.

The visual convolutional neural network model of the AI chip module 50 may be a visual convolutional neural network model that is already available in the prior art. The visual convolution neural network model belongs to a mature algorithm model, and can detect interesting features including human body structures, such as human faces and/or human shapes, based on a deep neural network algorithm. The AI chip module 50 may select an existing single-algorithm or dual-algorithm visual convolutional neural network model.

Based on a visual convolutional neural network model of a single algorithm, a feature of interest, such as a human face or a human figure, can be selected, feature coordinates are labeled in an image, and the movement of the feature is calculated. The visual convolution neural network model based on the dual algorithm can simultaneously calculate the face feature and the human shape feature in the environment information image acquired by the camera 20, and specify the coordinates of the face and the human shape in the image to calculate the movement of the face and the human shape. The GPIO interface through AI chip module 50 exports the instruction of control height level to control driver 40 rotates, makes hold assembly 10 for base 30 rotates, and then realizes that camera 20 follows the people's face or humanoid, realizes that the centre gripping is fixed terminal camera 20 on the hold assembly 10 can keep aiming at the user automatically, need not user manual operation, and the user need not oneself or other people come manual regulation at the process of live broadcast, improves live broadcast or video shooting effect, and is more convenient and intelligent, improves terminal support's practicality and intellectuality greatly.

Based on the existing visual convolution neural network model, the gesture can be taken as a characteristic to perform gesture recognition. For example, if the bottom layer is calibrated to be in a photographing mode that the right hand is lifted over the top of the head, and the gesture that the human body in the image is right hand and over the top of the head is acquired, an instruction for photographing or video can be sent to the terminal through the USB60 or in a wireless mode, so that the terminal can photograph or video.

The terminal holder has a self-test mode, which can be turned on when the terminal holder is powered on (the terminal holder is powered on and off through the power-on button 90 arranged on the base 30) or when the terminal holder fails to follow the power. The method specifically comprises the following steps: when the environmental information acquired by the camera 20 does not include the human structure information through calculation of the convolutional neural network model and judgment, the driver 40 is controlled to continuously rotate to detect the human structure, the rotation can be performed by rotating the driver by 350 degrees left and right until the environmental information acquired by the camera 20 includes the human structure information through calculation of the convolutional neural network model and judgment, the self-checking mode is stopped when the calibration is successful, and at this time, the indicator lamp 70 can be normally turned on to indicate that the calibration is successful.

During the self-test mode, the indicator light 70 may flash, and if the self-test is not successful after a predetermined time period, the low power consumption standby mode may be entered, or the power-off may be performed. Under the low-power consumption mode, be favorable to reducing the battery energy consumption, improve duration, improve the life of battery, reduce the discarded pollution of battery.

The terminal cradle may further include a sound pickup 80, the sound pickup 80 is used to collect environmental sound information, and the AI chip module 50 further has an AI voice model for sending an instruction to photograph or video to the terminal according to the environmental sound information collected by the sound pickup 80. The bottom layer is added with trained voice labels such as 'photo', 'video' and the like, when a user sends voice commands such as 'photo', 'video' and the like, and after calculation and recognition are successful, an AI chip module 50 sends commands for taking pictures or videos to the terminal, so that the terminal takes pictures or videos.

Of course, the utility model discloses bluetooth function can also be increased for connect the bluetooth remote controller and rotate or the control of shooing at terminal in order to realize terminal support's remote control.

Throughout the description and claims of this application, the words "comprise/comprises" and the words "have/includes" and variations of these are used to specify the presence of stated features, values, steps or components but do not preclude the presence or addition of one or more other features, values, steps, components or groups thereof.

Some features of the invention, which are, for clarity, described in the context of separate embodiments, may also be provided in combination in a single embodiment. Conversely, certain features of the invention, which are, for brevity, described in the context of a single embodiment, may also be provided separately or in any suitable combination in different embodiments.

The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents and improvements made within the spirit and principle of the present invention are intended to be included within the scope of the present invention.

Claims

1. A terminal support based on AI visual convolution neural network technique, characterized by that includes:

a holding member for holding the fixed terminal;

2. The AI visual convolutional neural network technology-based terminal support of claim 1, wherein the AI chip module comprises a CMOS image sensor and a GPU running a convolutional neural network model, and the CMOS image sensor and the GPU are bonded together by means of an SOC.

3. The AI visual convolutional neural network technology-based terminal support of claim 1, wherein the human body structure comprises a human face and/or a human figure.

4. The AI visual convolutional neural network technology-based terminal support of claim 1, wherein when it is calculated and determined by a convolutional neural network model that the environmental information acquired by the camera does not include human structure information, the driver is controlled to continuously rotate to detect the human structure.

5. The AI visual convolutional neural network technology-based terminal support of claim 1, further comprising a sound pickup for collecting environmental sound information, wherein the AI chip module further has an AI voice model for sending an instruction to take a picture or video to the terminal according to the environmental sound information collected by the sound pickup.

6. The AI visual convolutional neural network technology-based terminal support of claim 1, wherein an interface for connecting a tripod is provided on the base.