Detection system and detection method

Abstract

The present disclosure relates to a detection system and a detection method. The present disclosure provides a detection system and a detection method. The detection method comprises: obtaining, by a key point obtaining module, a first key point, a second key point and a third key point of a face based on an image containing the face, wherein the third key point is obtained based on a preset position on a face midline and the first key point and the second key point are obtained based on a pair of positions outside the face midline; obtaining, by a first calculation module, a vector based on the first key point and the second key point; obtaining, by a second calculation module, a binary linear equation based on the vector and the third key point; and substituting coordinates of the first key point and coordinates of the second key point into the binary linear equation to obtain a first value and a second value, respectively, and determining a state of the face based on the first value and the second value.

Description

Technical Field

[0001] This invention relates to the field of image recognition, specifically a technique for determining the state of a face in an image by using key points on the face. Background Technology

[0002] Traditionally, interpupillary distance (IPD) is a commonly used metric for frontal face assessment. However, this metric does not take into account the distance issue that occurs during face detection. The IPD of a frontal face at a distance may be the same as that of a profile face at a close distance, thus making it impossible to accurately select frontal face samples for face recognition. Summary of the Invention

[0003] In view of this, some embodiments of the present invention provide a training system, a training method, and a recognition system to improve the problems of the prior art.

[0004] Some embodiments of the present invention provide a detection system, which includes a key point acquisition module, a first calculation module, a second calculation module, and a judgment module. The key point acquisition module is configured to receive an image containing a face, and obtain a first key point, a second key point, and a third key point of the face based on the image. The key point acquisition module obtains the third key point based on a preset position on the midline of the face and obtains the first and second key points based on two paired positions outside the midline of the face. The first calculation module is configured to obtain a vector based on the first and second key points. The second calculation module is configured to obtain a two-variable linear function based on the vector and the third key point. The judgment module is configured to substitute the coordinates of the first key point and the second key point into the two-variable linear function to obtain a first value and a second value, respectively, and judge the state of the face based on the first value and the second value.

[0005] Some embodiments of the present invention provide a detection method applicable to a detection system. The detection system includes a key point acquisition module, a first calculation module, a second calculation module, and a judgment module. The detection method includes the following steps: the key point acquisition module receives an image containing a face, and obtains a first key point, a second key point, and a third key point of the face based on the aforementioned image. The key point acquisition module obtains the third key point based on a preset position on the midline of a face and obtains the first and second key points based on two paired positions outside the midline of the face. The first calculation module obtains a vector based on the first and second key points. The second calculation module obtains a linear equation in two variables based on the aforementioned vector and the third key point. The judgment module substitutes the coordinates of the first and second key points into the linear equation in two variables to obtain a first value and a second value, respectively, and judges the state of the face based on the first and second values.

[0006] Based on the above, the detection system and detection method provided in some embodiments of the present invention can quickly determine the state of the face in the image by performing simple algebraic operations and judgments on the first key point, second key point and third key point of the detected face. Attached Figure Description

[0007] Figure 1 This is a block diagram of a detection system drawn according to some embodiments of the present invention.

[0008] Figure 2 These are schematic diagrams of a human face and key points illustrated according to some embodiments of the present invention.

[0009] Figure 3 These are schematic diagrams of a human face and key points illustrated according to some embodiments of the present invention.

[0010] Figure 4-1 This is a schematic diagram illustrating the correspondence between preset positions and human faces, based on some embodiments of the present invention.

[0011] Figure 4-2 This is a schematic diagram illustrating the correspondence between preset positions and human faces, based on some embodiments of the present invention.

[0012] Figure 5 The module block diagram is obtained based on key points shown in some embodiments of the present invention.

[0013] Figure 6 These are schematic diagrams of the structure of an electronic device illustrated according to some embodiments of the present invention.

[0014] Figure 7 This is a flowchart illustrating a detection method based on some embodiments of the present invention.

[0015] Figure 8 This is a flowchart illustrating a detection method based on some embodiments of the present invention.

[0016] Figure 9 This is a flowchart illustrating a detection method based on some embodiments of the present invention.

[0017] Figure 10 This is a flowchart illustrating a detection method based on some embodiments of the present invention. Detailed Implementation

[0018] The foregoing descriptions and other technical contents, features, and effects of this invention will be clearly presented in the following detailed description of embodiments with reference to the accompanying drawings. The thickness or dimensions of the elements in the drawings are exaggerated, omitted, or approximated for the understanding and reading of those skilled in the art. The dimensions of each element are not their actual dimensions and are not intended to limit the implementation of this invention; therefore, they have no substantial technical significance. Any modifications to the structure, changes in proportions, or adjustments in size, without affecting the effects and objectives achieved by this invention, should still fall within the scope of the technical content disclosed in this invention. The same reference numerals will be used to denote the same or similar elements in all drawings. The term "connection" as used in the following embodiments can refer to any direct or indirect, wired or wireless connection means. In this document, the terms "first" or "second," and similar ordinal numbers, are used to distinguish or refer to elements or structures that are the same or similar, and do not necessarily imply the order of these elements in the system. It should be understood that, in certain situations or configurations, ordinal numbers can be used interchangeably without affecting the implementation of this invention.

[0019] Figure 1 This is a block diagram of a detection system drawn according to some embodiments of the present invention. Figure 2 These are schematic diagrams of a human face and key points illustrated according to some embodiments of the present invention. Figure 3 These are schematic diagrams of a human face and key points illustrated according to some embodiments of the present invention. Please also refer to... Figures 1 to 3 The detection system 100 includes a key point acquisition module 101, a first calculation module 102, a second calculation module 103, and a judgment module 104.

[0020] A keypoint acquisition module 101 is configured to receive an image 105 containing a face, such as image 201 or image 301. The keypoint acquisition module 101 is configured to acquire a first keypoint, a second keypoint, and a third keypoint of the face based on multiple preset locations and based on the received image 105 containing the face. Specifically, the keypoint acquisition module 101 acquires the third keypoint based on a preset location on the midline of the face and acquires the first and second keypoints based on two paired preset locations outside the midline of the face.

[0021] In some embodiments of the present invention, the aforementioned plurality of preset positions include the positions of the nose and the two eyes, wherein the nose is located on the midline of the face, and the two eyes are located in pairs outside the midline of the face. Figure 2The illustrated embodiment illustrates that the keypoint acquisition module 101 acquires a first keypoint 202, a second keypoint 203, and a third keypoint 204 corresponding to the positions of the two eyes in the image 201, based on preset positions of the nose and two eyes. The aforementioned acquisition of the first keypoint 202, second keypoint 203, and third keypoint 204 corresponding to the positions of the two eyes in the image 201 represents the acquisition of the coordinates of each of the first keypoint 202, second keypoint 203, and third keypoint 204. The third keypoint 204 corresponding to the position of the nose lies on the face midline 206. Figure 3 The illustrated embodiment illustrates that the aforementioned plurality of preset positions include the positions of the nose and the two eyes. The keypoint acquisition module 101 obtains, based on the positions of the nose and eyes, a first keypoint 302, a second keypoint 303, and a third keypoint 304 corresponding to the positions of the two eyes in the image 301. The aforementioned first keypoint 302, second keypoint 303, and third keypoint 304 corresponding to the positions of the two eyes in the obtained image 301 represent the coordinates of each of the first keypoint 302, second keypoint 303, and third keypoint 304. The third keypoint 304 corresponding to the position of the nose lies on the face midline 306.

[0022] In some embodiments of the present invention, the aforementioned plurality of preset positions include the positions of the nose, the left corner of the mouth, and the right corner of the mouth, wherein the nose is located on the midline of the face, and the left and right corners of the mouth are located in pairs outside the midline of the face. Figure 2 The illustrated embodiment illustrates that the key point acquisition module 101 acquires an image 201 based on preset positions of the nose, left corner of the mouth, and right corner of the mouth, corresponding to a first key point 207, a second key point 208, and a third key point 204 corresponding to the position of the nose. Figure 3 The illustrated embodiment illustrates that the aforementioned multiple preset positions include the positions of the nose, the left corner of the mouth, and the right corner of the mouth. The key point acquisition module 101 obtains an image 301 based on the positions of the nose, the left corner of the mouth, and the right corner of the mouth, corresponding to a first key point 307, a second key point 308, and a third key point 304 corresponding to the position of the nose.

[0023] Figure 5 The module block diagram is obtained based on key points illustrated in some embodiments of the present invention. Please refer to... Figure 5 ,exist Figure 5 In the illustrated embodiment, the keypoint acquisition module 101 includes a trained neural network module 1011. The trained neural network module 1011 is configured to receive the aforementioned image 105 containing a face and output the coordinates of a first keypoint, a second keypoint, and a third keypoint of the face so that the keypoint acquisition module 101 obtains the first keypoint, the second keypoint, and the third keypoint.

[0024] In some embodiments of the present invention, the aforementioned neural network module 1011 includes a trained multi-task cascaded convolutional network. After receiving an image 105 containing a face, the trained multi-task cascaded convolutional network can output the coordinates of key points on the face image 105 corresponding to the positions of the nose, left corner of the mouth, right corner of the mouth, and two eyes. The key point acquisition module 101 then obtains a first key point, a second key point, and a third key point from the output of the multi-task cascaded convolutional network based on multiple preset positions.

[0025] Figure 4-1 This is a schematic diagram illustrating the correspondence between preset positions and human faces, based on some embodiments of the present invention. Figure 4-2 This is a schematic diagram illustrating the correspondence between preset positions and human faces, based on some embodiments of the present invention. In some embodiments of the present invention, the aforementioned plurality of preset positions are selected from preset position 4-1 to preset position 4-68. For example... Figure 4-2 As shown, preset positions 4-9, 4-28 to 4-31, 4-34, 4-52, 4-58, 4-63 and 4-67 are located on the midline of the face. Figure 4-1 and Figure 4-2 The preset positions 4-1 to 4-8, 4-10 to 4-27, 4-32 to 4-33, 4-35 to 4-51, 4-53 to 4-57, 4-59 to 4-62, 4-64 to 4-66, and 4-68 are depicted in pairs at symmetrical positions on the face. In some embodiments of the present invention, the third key point is selected from one of the aforementioned preset positions 4-9, 4-28 to 4-31, 4-34, 4-52, 4-58, 4-63, and 4-67, for example, preset position 4-28. The first key point and the second key point are selected from pairs of preset positions among preset positions 4-1 to preset positions 4-8, preset positions 4-10 to preset positions 4-27, preset positions 4-32 to preset positions 4-33, preset positions 4-35 to preset positions 4-51, preset positions 4-53 to preset positions 4-57, preset positions 4-59 to preset positions 4-62, preset positions 4-64 to preset positions 4-66 and preset positions 4-68, such as preset positions 4-3 and preset positions 4-15.

[0026] In some embodiments of the present invention, the keypoint acquisition module 101 calls the FacemarkLBF category in OpenCV to obtain the coordinates of keypoints on the face of image 105 corresponding to the aforementioned preset positions 4-1 to 4-68. The keypoint acquisition module 101 then selects the coordinates corresponding to the aforementioned preset positions from the obtained keypoint coordinates corresponding to the aforementioned preset positions 4-1 to 4-68 based on multiple preset positions to obtain the first keypoint, the second keypoint, and the third keypoint. It is worth noting that the keypoint acquisition module 101 can also call modules from other image processing software to obtain the coordinates of keypoints on the face of image 105 corresponding to the aforementioned preset positions 4-1 to 4-68; the present invention is not limited to the FacemarkLBF category in OpenCV.

[0027] The following is a detailed description, with reference to the accompanying drawings, of some embodiments of the detection method of the present invention and how the various modules of the detection system 100 work together.

[0028] Figure 7 This is a flowchart illustrating a detection method based on some embodiments of the present invention. Please also refer to... Figures 1 to 3 as well as Figure 7 ,exist Figure 7 In the illustrated embodiment, the detection method includes steps S701 to S704. In step S701, the key point acquisition module 101 receives an image 105 containing a face. The key point acquisition module 101 then obtains the first key point, second key point, and third key point of the face based on the image 105, wherein the key point acquisition module 101 is based on a preset position on the midline of a face (e.g., Figure 2 The third keypoint (e.g., third keypoint 204 or third keypoint 304) is obtained from the position of the nose shown in the diagram, and based on two paired positions outside the midline of the face (e.g., Figure 2 The positions of the two eyes are shown to obtain a first key point (e.g., first key point 202 or first key point 302) and a second key point (e.g., second key point 203 corresponding to first key point 202 or second key point 303 corresponding to first key point 302).

[0029] In step S702, the first calculation module 102 obtains a vector based on the first keypoint and the second keypoint. For example, the first calculation module 102 obtains vector 205 based on the first keypoint 202 and the second keypoint 203. For example, the first calculation module 102 obtains vector 209 based on the first keypoint 207 and the second keypoint 208. For example, the first calculation module 102 obtains vector 305 based on the first keypoint 302 and the second keypoint 303. For example, the first calculation module 102 obtains vector 309 based on the first keypoint 307 and the second keypoint 308.

[0030] In step S703, the second calculation module 103 obtains a linear equation in two variables based on the aforementioned vector and the third key point. In step S704, the judgment module 104 substitutes the coordinates of the first key point and the second key point into the aforementioned linear equation in two variables to obtain a first value and a second value, respectively. The judgment module 104 judges the state of the face in the image 105 based on the aforementioned first value and the aforementioned second value. In some embodiments of the present invention, the aforementioned state of the face includes a frontal face state and a side face state.

[0031] Figure 8 This is a flowchart illustrating a detection method based on some embodiments of the present invention. Please also refer to... Figure 1 , Figure 7 as well as Figure 8 ,exist Figure 8 In the illustrated embodiment, step S704 includes steps S801 to S803. In step S801, the determination module 104 determines whether the product of the first value and the second value is positive. If the product of the first value and the second value is positive, the process proceeds to step S802; if the product of the first value and the second value is negative, the process proceeds to step S803. In step S802, the determination module 104 determines that the face is in a side profile state in response to the product of the first value and the second value being positive. In step S803, the determination module 104 determines that the face is in a frontal view state in response to the product of the first value and the second value being negative.

[0032] The following are also in the same format Figure 2 In image 201, the first key point 202, the second key point 203 corresponding to the positions of the two eyes, and the third key point 204 corresponding to the position of the nose, and Figure 3The image 301 contains key points 302 and 303 corresponding to the positions of the two eyes, and a key point 304 corresponding to the position of the nose. These are used as illustrative examples. The coordinates of key point 302 are (54, 56), key point 303 are (74, 56), key point 304 are (64, 64), key point 302 are (73, 51), key point 303 are (88, 55), and key point 304 are (91, 68). These coordinates are taken from the pixel coordinates of image 105, specifically the coordinates of the top-left corner of image 105, which are defined as (0, 0). The first coordinate component is the position counted from left to right based on the top-left corner of image 105, and the second coordinate component is the position counted from top to bottom based on the top-left corner of image 105. It is worth noting that since the method of the present invention does not produce different results due to the selection of different coordinate systems, the present invention is not limited to the aforementioned pixel coordinates.

[0033] Figure 9 This is a flowchart illustrating a detection method based on some embodiments of the present invention. Please also refer to... Figures 1 to 3 as well as Figure 9 ,exist Figure 9 In the illustrated embodiment, step S702 includes step S901, in which the first calculation module 102 subtracts the coordinates of the first key point from the coordinates of the second key point to obtain the aforementioned vector. Taking the aforementioned example, the first calculation module 102 subtracts the coordinates of the first key point 202 (54, 56) from the coordinates of the second key point 203 to obtain vector 205 as (20, 0). The first calculation module 102 subtracts the coordinates of the first key point 302 (73, 51) from the coordinates of the second key point 303 (88, 55) to obtain vector 305 as (15, 4).

[0034] Figure 10 This is a flowchart illustrating a detection method based on some embodiments of the present invention. Please also refer to... Figures 1 to 3 , Figures 1 to 8 as well as Figure 10 ,exist Figure 10 In the illustrated embodiment, the aforementioned step S703 includes steps S1001 to S1003. In step S1001, the second calculation module 103 sets the first component of the aforementioned vector as the coefficient of the first variable of an intermediate linear equation in two variables, and sets the second component of the aforementioned vector as the coefficient of the second variable of the aforementioned intermediate linear equation in two variables.

[0035] In step S1002, the second calculation module 103 substitutes the first and second coordinates of the third key point into the aforementioned intermediate linear equation in two variables to obtain the constant term of the intermediate linear equation in two variables. In step S1003, the first component of the aforementioned vector is set as the coefficient of the first variable of the linear equation in two variables, the second component of the aforementioned vector is set as the coefficient of the second variable of the linear equation in two variables, and the constant term of the aforementioned intermediate linear equation in two variables is set as the constant term of the linear equation in two variables.

[0036] The following is based on the aforementioned Figure 2 The image 201 uses the following key points to illustrate the positions of the two eyes: the first key point 202 (coordinates (54,56)), the second key point 203 (coordinates (74,56)), the third key point 204 (coordinates (64,64)), and the vector 205 (value (20,0)).

[0037] In step 1001, the second calculation module 103 substitutes the first component of vector 205 into the coefficient α1 of the first variable x in the intermediate linear equation (1) below, and substitutes the second component 0 of vector 205 into the coefficient β1 of the second variable y in the aforementioned intermediate linear equation (1).

[0038] α1x+β1y+γ1=0,……(1)

[0039] The equation 20x + γ1 = 0 is obtained. In step 1002, the second calculation module 103 substitutes the first coordinate component 64 and the second coordinate component 64 of the third key point 204 into the first variable x and the second variable y of the aforementioned intermediate linear equation in two variables, respectively, to solve for the value of the constant term γ1 of the intermediate linear equation in two variables, which is -1280. In step S1003, the first component of vector 205 is substituted into the coefficient α2 of the first variable x in the following linear equation in two variables (2), the second component of vector 205 is substituted into the coefficient β2 of the second variable y in the following linear equation in two variables (2), and the constant term γ1 = -1280 of the aforementioned intermediate linear equation in two variables is substituted into the constant term γ2 of the following linear equation in two variables (2).

[0040] α²x + β²y + γ²,……(2)

[0041] We obtain the linear equation in two variables 20x-1280.

[0042] It is worth noting that the judgment module 104 substitutes the coordinates (54, 56) of the first key point 202 into the first variable x and the second variable y of the linear equation 20x-1280, respectively, to obtain a first value of -200. Similarly, it substitutes the coordinates (74, 56) of the second key point 203 into the same equation to obtain a second value of 200. The judgment module 104 determines that the face in image 201 is in a frontal view state when the product of the first value -200 and the second value 200 is negative.

[0043] The following will refer to the aforementioned Figure 3 The image 301 uses the following key points to illustrate the positions of the two eyes: the first key point 302 (coordinates (73,51)), the second key point 303 (coordinates (88,55)), the third key point 304 (coordinates (91,68)), and the vector 305 (value (15,4)).

[0044] In step S1001, the second calculation module 103 substitutes the first component 15 of vector 305 into the coefficient α1 of the first variable x in the aforementioned intermediate linear equation (1), and substitutes the second component 4 of vector 305 into the coefficient β1 of the second variable y in the aforementioned intermediate linear equation (1), to obtain the equation 15x + 4y + γ1 = 0. In step S1002, the second calculation module 103 substitutes the first coordinate component 91 and the second coordinate component 68 of the third key point 304 into the first variable x and the second variable y of the aforementioned intermediate linear equation, respectively, to solve for the constant term γ1 of the intermediate linear equation, which has a value of -1637. In step 1003, the first component of vector 305 is substituted into the coefficient α2 of the first variable x in the above linear equation (2), the second component of vector 305 is substituted into the coefficient β2 of the second variable y in the above linear equation (2), and the constant term γ1=-1637 of the aforementioned intermediate linear equation is substituted into the constant term γ2 of the above linear equation (2) to obtain the linear equation 15x+4y-1637.

[0045] It is worth noting that the judgment module 104 substitutes the coordinates (73, 51) of the first key point 302 into the first variable x and the second variable y of the linear equation 15x + 4y - 1637, respectively, to obtain the first value -338. Similarly, it substitutes the coordinates (88, 55) of the second key point 303 into the same equation to obtain the second value -97. The judgment module 104 determines that the face in image 301 is in profile mode when the product of the first value -338 and the second value -97 is positive.

[0046] Figure 6 These are schematic diagrams illustrating the structure of an electronic device based on some embodiments of the present invention. For example... Figure 6 As shown, at the hardware level, electronic device 600 includes processors 601-1, 601-2 to 601-R, where R is a positive integer. It also includes internal memory 602 and non-volatile memory 603. Internal memory 602 is, for example, random-access memory (RAM). Non-volatile memory 603 is, for example, at least one disk storage device. Of course, electronic device 600 may also include hardware required for other functions.

[0047] Internal memory 602 and non-volatile memory 603 are used to store programs, which may include program code and computer operation instructions. Internal memory 602 and non-volatile memory 603 provide instructions and data to processors 601-1 to 601-R. Processors 601-1 to 601-R read the corresponding computer program from non-volatile memory 603 into internal memory 602 and then execute it, forming a detection system 100 at the logical level. Specifically, processors 601-1 to 601-R are used to execute... Figures 7 to 10 The steps described herein. Of course, each module of the detection system 100 can also be implemented in hardware, and this invention is not limited thereto.

[0048] Processors 601-1 to 601-R may be integrated circuit chips with signal processing capabilities. In implementation, the methods and steps disclosed in the foregoing embodiments can be performed through hardware integrated logic circuits or software instructions in processors 601-1 to 601-R. Processors 601-1 to 601-R may be general-purpose processors, including central processing units (CPUs), tensor processing units, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, capable of implementing or executing the methods and steps disclosed in the foregoing embodiments.

[0049] In some embodiments of the present invention, a computer-readable storage medium is also provided, which stores at least one instruction that, when executed by the processors 601-1 to 601-R of the electronic device 600, enables the processors 601-1 to 601-R of the electronic device 600 to perform the methods and steps disclosed in the foregoing embodiments.

[0050] Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other internal memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable media that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient media, such as modulated data signals and carrier waves.

[0051] Based on the above, the detection system and detection method provided in some embodiments of the present invention can quickly determine the state of the face by performing simple algebraic operations and judgments on the first key point, second key point and third key point of the detected face.

[0052] Although the present invention has been disclosed above by way of embodiments, it is not intended to limit the present invention. Anyone skilled in the art can make some modifications and refinements without departing from the spirit and scope of the present invention. Therefore, the scope of protection of the present invention shall be determined by the appended claims.

[0053] [Symbol Explanation]

[0054] 100: Detection System

[0055] 101: Key Point Acquisition Module

[0056] 102: First computing module

[0057] 103: Second computing module

[0058] 104: Detect Module

[0059] 105,201,301: Images

[0060] 202,207,302,307: First key point

[0061] 203, 208, 303, 308: Second key point

[0062] 204, 304: Third key point

[0063] 206, 306: Face midline

[0064] 205,209,305,309: Vector

[0065] 1011: Neural Network Module

[0066] 4-1~4-68: Preset Positions

[0067] 600: Electronic Equipment

[0068] 601-1, 601-2 to 601-R: Processors

[0069] R: positive integer

[0070] 602: Internal Memory

[0071] 603: Non-volatile memory

[0072] S701~S704, S801~S803, S901, S1001~S1003: Steps

Claims

1. A detection system, comprising: A key point acquisition module is configured to receive an image containing a face and obtain a first key point, a second key point and a third key point of the face based on the image. The key point acquisition module obtains the third key point based on a preset position on the midline of the face and obtains the first key point and the second key point based on two preset positions outside the midline of the face. A first computing module is configured to obtain a vector based on the first key point and the second key point; A second computational module is configured to obtain a two-variable linear function based on the vector and the third key point; as well as A judgment module is configured to substitute the coordinates of the first key point and the second key point into the linear equation in two variables to obtain a first value and a second value respectively, and to judge the state of the face based on the first value and the second value.

2. The detection system of claim 1, wherein the judgment module is configured to determine that the face is in a side profile state when the product of the first value and the second value is positive, and to determine that the face is in a frontal view state when the product of the first value and the second value is negative.

3. The detection system of claim 1, wherein the first calculation module is configured to subtract the coordinates of the first key point from the coordinates of the second key point to obtain the vector.

4. The detection system of claim 1, wherein the second calculation module performs the following steps to obtain the linear equation in two variables: setting a first component of the vector as a coefficient of a first variable of an intermediate linear equation in two variables, setting a second component of the vector as a coefficient of a second variable of the intermediate linear equation in two variables; substituting a first coordinate component and a second coordinate component of the third key point into the intermediate linear equation in two variables to solve for a constant term of the intermediate linear equation in two variables; and setting the first component of the vector as a coefficient of a first variable of the linear equation in two variables, setting a second component of the vector as a coefficient of a second variable of the linear equation in two variables, and setting the constant term of the intermediate linear equation in two variables as a constant term of the linear equation in two variables.

5. The detection system as described in claim 1, wherein the preset position on the midline of the face is a nose position.

6. The detection system as claimed in claim 1, wherein the two preset positions outside the midline of the face are the positions of the two eyes.

7. The detection system of claim 1, wherein the key point acquisition module includes a neural network module configured to receive the image of the face and output a coordinate of the first key point, a coordinate of the second key point, and a coordinate of the third key point of the face.

8. A detection method applicable to a detection system, the detection system comprising a key point acquisition module, a first calculation module, a second calculation module, and a judgment module, the detection method comprising the following steps: (a) The key point acquisition module receives an image containing a face and obtains a first key point, a second key point and a third key point of the face based on the image, wherein the key point acquisition module obtains the third key point based on a preset position on the midline of the face and obtains the first key point and the second key point based on two preset positions outside the midline of the face. (b) The first computing module obtains a vector based on the first key point and the second key point; (c) The second computational module obtains a two-variable linear function based on the vector and the third key point; and (d) The judgment module substitutes the coordinates of the first key point and the second key point into the linear equation to obtain a first value and a second value respectively, and judges the state of the face based on the first value and the second value.

9. The detection method as claimed in claim 8, wherein the aforementioned step (d) comprises: (d1) The judgment module determines that the face is in a side profile state when the product of the first value and the second value is positive; and determines that the face is in a frontal state when the product of the first value and the second value is negative.

10. The detection method as claimed in claim 8, wherein step (b) comprises: (b1) The first calculation module subtracts the coordinate of the first key point from the coordinate of the second key point to obtain the vector.

Images