Method and system for measuring height of human body by monocular

Abstract

The method comprises the following steps: S1, selecting a calibration object, carrying out parameter calibration on shooting equipment to obtain calibration data, and obtaining a proportionality coefficient according to the calibration data; s2, acquiring a face image, and inputting the face image into the YOLO network model to obtain face detection data; s3, if the face detection data only comprises eye data, calculating a first human body height, and executing S4; if the face detection data comprises eye data and mouth data, calculating a second human body height, then calculating a third human body height based on multi-feature fusion, and executing S4; and S4, carrying out sliding mean filtering operation to obtain a final height measurement result. According to the human body height algorithm based on face detection provided by the invention, the human body height can be measured by using an incomplete human body picture, and the detection deviation caused by the distance between the human body and the camera can be corrected through the mean value filtering operation, so that the human body height can be accurately measured and calculated under the condition that the calculated amount is small.

Description

Technical Field

[0001] The present invention relates to the technical field of height measurement, and in particular to a method and system for monocularly measuring human height. Background Art

[0002] At present, there are mainly the following methods for measuring human height, namely, the detection method based on monocular face key point recognition, the detection method based on binocular human body contour detection algorithm, and the detection method based on ultrasonic sensor. The specific methods are as follows: 1) The detection method based on face key point recognition is to obtain preprocessed data according to the image information detected in a specific detection area and the position of the identification block, and then input the preprocessed data and the feature information of the object to be measured into the model based on face key point detection height to obtain the height data of the object to be measured; 2) The detection method based on binocular human body contour detection algorithm uses the images collected by two cameras at the same time, reconstructs the three-dimensional space model of the photographed object through the installation height and relative position of the two cameras, extracts the human body contour through the human body contour detection algorithm, and calculates the height of the human body based on the model composed of parameters such as camera height and angle; 3) The detection method based on ultrasonic sensor requires an ultrasonic detector, including a base and a metal rod more than two meters high. There is an ultrasonic generator located directly above the base in the vertical direction of the metal rod. The person being tested stands on the base, and the height is calculated by the time difference between transmitting and receiving ultrasonic waves.

[0003] The existing methods have the following defects: 1) Although the detection method based on facial key point recognition has low requirements for equipment, the measurement needs to be carried out in a specific detection area according to the set identification blocks to calculate the height normally. The style and placement of the identification blocks are fixed, the applicable scene is single, and the versatility is low; 2) The equipment based on the ultrasonic sensor detection method is usually integrated with the weight detection device. The equipment is bulky, the height is more than two meters, it is very inconvenient to carry, and the cost is expensive. In addition, this method is difficult to eliminate the influence of hairstyle, and the influence on measurement accuracy is greater when the hair is erect; 3) The solution using depth camera and image recognition requires multi-camera support, and the measurement result depends on the accurate recognition of the human body contour. In actual situations, the accurate recognition of the human body contour is very difficult and is greatly affected by factors such as clothing and background. In addition, this method uses the upper edge of the head detected by the camera as the top of the head, rather than the actual top of the head, and cannot eliminate the influence of hairstyle. Summary of the invention

[0004] The present invention provides a method and system for monocularly measuring human height, so as to overcome the technical problems of low applicability, inconvenience and inaccurate measurement results in the prior art.

[0005] In order to achieve the above object, the technical solution of the present invention is:

[0006] A method for monocularly measuring human height, the specific steps comprising:

[0007] S1: Select a calibration object, calibrate the parameters of the camera based on the calibration object, obtain calibration data, and obtain a proportional coefficient based on the calibration data;

[0008] S2: Obtain a facial image based on a shooting device, and input the facial image into a YOLO network model to obtain facial detection data;

[0009] S3: If the facial detection data only includes eye data, the first human height is calculated based on the eye data, and S4 is executed; if the facial detection data includes eye data and mouth data, the second human height is calculated based on the eye data and the mouth data, and then the third human height of multi-feature fusion is calculated based on the first human height and the second human height, and S4 is executed;

[0010] S4: performing a sliding mean filtering operation on the first human height, the second human height and the third human height to obtain a final height measurement result.

[0011] Furthermore, in S1, the process of selecting a calibration object, calibrating the parameters of the photographing device based on the calibration object, obtaining calibration data, and obtaining the proportional coefficient according to the calibration data is as follows:

[0012] Select an object with a regular shape as the calibration object and measure the distance u from the geometric center of the calibration object to the camera. sta , in cm; the length of the calibration object o sta , in cm; the actual length of the calibration object on the display screen i s_sta , unit is cm;

[0013] Get the average pupil distance of humans e , unit is cm;

[0014] Measure the distance H from the camera to the ground cg , unit is cm;

[0015] Measure and determine the actual width w of the image on the display screen in cm, and obtain the number of pixels w of the image width p , the unit is pixel;

[0016] The proportional coefficient is obtained according to the above calibration data, including:

[0017] High-order calibration factors:

[0018]

[0019] Low-order calibration factors:

[0020]

[0021] System pupil distance factor:

[0022] γ=o e -0.82834.

[0023] Further, if the facial detection data only includes eye data, the formula for calculating the height of the first person based on the eye data is:

[0024] H p1 =α·i p_ch1 ·i p_e -3 +β·i p_ch1 ·i p_e -2 +γ·i p_ch1 ·i p_e -1 +H cg

[0025] i p_ch1 is the distance from the center of the screen to the first pixel on the top of the head, and the calculation formula is:

[0026] i p_ch1 =y mid +y jaw -y le -y re

[0027] In the formula, y mid is the ordinate of the center of the image; jaw The vertical coordinate of the bottom edge of the face detection frame; y le y is the ordinate of the center of the left eye detection frame; re is the vertical coordinate of the center of the right eye detection frame; i p_e is the pixel distance between the two eyes, and i p_e =|x le -x re |, x le is the horizontal coordinate of the center of the left eye detection frame, x re H is the horizontal coordinate of the center of the right eye detection frame; cg is the distance from the camera to the ground; y m It is the vertical coordinate of the center of the mouth detection frame.

[0028] Furthermore, if the facial detection data includes eye data and mouth data, the formula for calculating the height of the second person based on the eye data and mouth data is:

[0029] H p2 =α·i p_ch2 ·i p_e -3 +β·i p_ch2 ·ip_e -2 +γ·i p_ch2 ·i p_e -1 +H cg

[0030] i p_ch2 The second pixel distance from the center of the screen to the top of the head is calculated as:

[0031] i p_ch2 =y mid +1.4y m -1.2y le -1.2y re

[0032] In the formula, y mid is the ordinate of the center of the image; m y is the ordinate of the center of the mouth detection frame; le y is the ordinate of the center of the left eye detection frame; re is the vertical coordinate of the center of the right eye detection frame; i p_e is the pixel distance between the two eyes, and i p_e =|x le -x re |, x le is the horizontal coordinate of the center of the left eye detection frame, x re H is the horizontal coordinate of the center of the right eye detection frame; cg is the distance from the camera to the ground.

[0033] Furthermore, the formula for calculating the third person height of multi-feature fusion based on the first person height and the second person height is:

[0034] H p3 =α·i p_ch3 ·i p_e -3 +β·i p_ch3 ·i p_e -2 +γ·i p_ch3 ·i p_e -1 +H cg

[0035] i p_ch3 The third pixel distance from the center of the screen to the top of the head is calculated as:

[0036]

[0037] In the formula, y m i d is the ordinate of the center of the image; m y is the ordinate of the center of the mouth detection frame; ley is the ordinate of the center of the left eye detection frame; re is the vertical coordinate of the center of the right eye detection frame; i p_e is the pixel distance between the two eyes, and i p_e =|x le -x re |, x le is the horizontal coordinate of the center of the left eye detection frame, x re H is the horizontal coordinate of the center of the right eye detection frame; cg is the distance from the camera to the ground.

[0038] Further, in S4, a sliding mean filtering operation is performed on the predicted first human height, second human height, and third human height to obtain a final height measurement result including:

[0039] The formula for performing a sliding mean filtering operation on the first human height, the second human height, and the third human height is:

[0040]

[0041] In the formula, H p (t) is the height prediction result after filtering in the tth frame, h p (t) is the height prediction result of the unfiltered wave in the tth frame, n is the number of times d∈[60,90] is satisfied in all time frames; i p_e is the pixel distance between the two eyes, and i p_e =|x le -x re |, x le is the horizontal coordinate of the center of the left eye detection frame, x re It is the horizontal coordinate of the center of the right eye detection frame.

[0042] A system for monocularly measuring human height, used to implement a method for monocularly measuring human height, comprising:

[0043] A proportional coefficient determination module, used to calibrate the parameters of the photographing device based on the calibration object, obtain calibration data, and obtain the proportional coefficient according to the calibration data;

[0044] YOLO network model, used to obtain face detection data based on face images;

[0045] A first human height prediction module, used to predict the first human height according to the facial detection data;

[0046] A second human body height prediction module, used to predict the second human body height according to the face detection data;

[0047] A third person height prediction module, used to predict the third person height based on the face detection data;

[0048] The sliding mean filtering module is used to perform a sliding mean filtering operation on the first human height, the second human height and the third human height to obtain a final height measurement result.

[0049] Beneficial effect: The present invention first calibrates the parameters of the shooting equipment to obtain calibration data, and obtains the proportional coefficient according to the calibration data; obtains the facial image based on the shooting equipment, and inputs the facial image into the YOLO network model to obtain facial detection data, obtains the first human height, the second human height and the third human height based on the facial detection data, and performs a sliding mean filtering operation on the first human height, the second human height and the third human height to obtain the final height measurement result. The human height algorithm based on face detection proposed by the present invention can use incomplete human photos to measure human height, and can correct the detection deviation caused by the distance between the human body and the camera through the mean filtering operation, so that the human height can be accurately calculated with a small amount of calculation. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0051] Figure 1 A flow chart of a method for monocularly measuring human height in the present invention;

[0052] Figure 2 This is a display screen diagram of a photographing device when performing parameter calibration in an embodiment of the present invention;

[0053] Figure 3 This is a diagram showing the relative relationship between the calibration object and the camera when performing parameter calibration in an embodiment of the present invention.

[0054] In the figure: 1. display screen; 2. image; 3. calibration object; 4. camera; 5. geometric center of the calibration object. DETAILED DESCRIPTION

[0055] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0056] According to the golden ratio rule of human face, the vertical distance from pupil to chin and hairline to pupil is 1 / 0.618, the vertical distance from pupil to middle of lip / the vertical distance from middle of lip to chin is 1 / 0.618, so the vertical distance from pupil to middle of lip and hairline to chin is about 0.3, and the vertical distance from pupil to chin and hairline to chin is about 0.4. According to the study of human interpupillary distance (Dodgson, NA (2004, May). Variation and extrema of human interpupillary distance. In Stereoscopic displays and virtual reality systems XI (Vol. 5291, pp. 36-46). SPIE.), the average interpupillary distance of adults is about 63 mm. This algorithm combines the above-mentioned biological characteristic ratios of facial organs and the average interpupillary distance o after experimental fine-tuning. e Combined with the YOLO target detection algorithm, it is used to calculate the distance between the human body and the camera and the height.

[0057] Based on the above principle, this embodiment provides a method for monocularly measuring human height. Figure 1 As shown, the specific steps include:

[0058] S1: Select a calibration object, calibrate the parameters of the camera based on the calibration object, obtain calibration data, and obtain a proportional coefficient based on the calibration data;

[0059] In a specific embodiment, in S1, a calibration object is selected, and the parameters of the photographing device are calibrated based on the calibration object to obtain calibration data, and the process of obtaining the proportional coefficient according to the calibration data is as follows:

[0060] like Figure 2 and Figure 3 As shown, a regular-shaped object is selected as the calibration object 3, and the distance u from the geometric center 5 of the calibration object to the camera 4 is measured and determined. sta , in cm; the length of the calibration object o sta , in cm; the actual length i of the calibration object on the display screen 1 s_sta , in cm; specifically, the display screen can be a display screen of a shooting device, a mobile phone or a computer display screen. The method can also measure the height of a person using only one camera and a camera stand, without requiring other equipment or special environments, and has the characteristics of low cost, high portability and high versatility.

[0061] Get the average pupil distance of humans e , unit is cm;

[0062] Measure the distance H from the camera to the groundcg , unit is cm;

[0063] Measure and determine the actual width w of image 2 on the display screen, in cm, and obtain the number of pixels w of the image width p , the unit is pixel;

[0064] The proportional coefficient is obtained according to the above calibration data, including:

[0065] High-order calibration factors:

[0066]

[0067] Low-order calibration factors:

[0068]

[0069] System pupil distance factor:

[0070] γ=o e -0.82834.

[0071] S2: Acquire a facial image based on a shooting device, and input the facial image into a YOLO network model to obtain facial detection data; specifically, in this embodiment, a YOLO5 network model is used to acquire facial detection data.

[0072] S3: If the facial detection data only includes eye data, the first human height is calculated based on the eye data, and S4 is executed; if the facial detection data includes eye data and mouth data, the second human height is calculated based on the eye data and the mouth data, and then the third human height of multi-feature fusion is calculated based on the first human height and the second human height, and S4 is executed;

[0073] In a specific embodiment, if the facial detection data only includes eye data, the formula for calculating the height of the first person based on the eye data is:

[0074] H p1 =α·i p_ch1 ·i p_e -3 +β·i p_ch1 ·i p_e -2 +γ·i p_ch1 ·i p_e -1 +H cg

[0075] i p_ch1 is the distance from the center of the screen to the first pixel on the top of the head, and the calculation formula is:

[0076] i p_ch1 =y mid +yjaw -y le -y re

[0077] In the formula, y mid is the ordinate of the center of the image; jaw The vertical coordinate of the bottom edge of the face detection frame; y le y is the ordinate of the center of the left eye detection frame; re is the vertical coordinate of the center of the right eye detection frame; i p_e is the pixel distance between the two eyes, and i p_e =|x le -x re |, x le is the horizontal coordinate of the center of the left eye detection frame, x re H is the horizontal coordinate of the center of the right eye detection frame; cg is the distance from the camera to the ground; y m It is the vertical coordinate of the center of the mouth detection frame.

[0078] In a specific embodiment, if the facial detection data includes eye data and mouth data, the formula for calculating the height of the second person based on the eye data and mouth data is:

[0079] H p2 =α·i p_ch2 ·i p_e -3 +β·i p_ch2 ·i p_e -2 +γ·i p_ch2 ·i p_e -1 +H cg

[0080] i p_ch2 The second pixel distance from the center of the screen to the top of the head is calculated as:

[0081] i p_ch2 =y mid +1.4y m -1.2y le -1.2y re

[0082] In the formula, y mid is the ordinate of the center of the image; m y is the ordinate of the center of the mouth detection frame; le y is the ordinate of the center of the left eye detection frame; re is the vertical coordinate of the center of the right eye detection frame; i p_e is the pixel distance between the two eyes, and i p_e =|x le -x re |, x leis the horizontal coordinate of the center of the left eye detection frame, x re H is the horizontal coordinate of the center of the right eye detection frame; cg is the distance from the camera to the ground.

[0083] In a specific embodiment, the formula for calculating the third person height of multi-feature fusion based on the first person height and the second person height is:

[0084] H p3 =α·i p_ch3 ·i p_e -3 +β·i p_ch3 ·i p_e -2 +γ·i p_ch3 ·i p_e -1 +H cg

[0085] i pch3 The third pixel distance from the center of the screen to the top of the head is calculated as:

[0086]

[0087] In the formula, y mid is the ordinate of the center of the image; m y is the ordinate of the center of the mouth detection frame; le y is the ordinate of the center of the left eye detection frame; re is the vertical coordinate of the center of the right eye detection frame; i p_e is the pixel distance between the two eyes, and i p_e =|x le -x re |, x le is the horizontal coordinate of the center of the left eye detection frame, x re H is the horizontal coordinate of the center of the right eye detection frame; cg is the distance from the camera to the ground.

[0088] S4: performing a sliding mean filtering operation on the first human height, the second human height and the third human height to obtain a final height measurement result.

[0089] In a specific embodiment, in S4, a sliding mean filtering operation is performed on the first human height, the second human height and the third human height to obtain a final height measurement result including:

[0090] The formula for performing a sliding mean filtering operation on the first human height, the second human height, and the third human height is:

[0091]

[0092] In the formula, Hp (t) is the height prediction result after filtering in the tth frame, h p (t) is the height prediction result of the unfiltered wave in the tth frame, n is the number of times d∈[60,90] is satisfied in all time frames; i p_e is the pixel distance between the two eyes, and i p_e =|x le -x re |, x le is the horizontal coordinate of the center of the left eye detection frame, x re It is the horizontal coordinate of the center of the right eye detection frame.

[0093] Specifically, in this embodiment, a mean filtering formula is set according to the 3σ principle, and sliding mean filtering is performed on the first human height, the second human height and the third human height to avoid the influence of fine-tuning of the camera focal length on the height measurement result as the distance between the human body and the camera changes.

[0094] This embodiment also provides a system for monocularly measuring human height, which is used to implement a method for monocularly measuring human height, including:

[0095] A proportional coefficient determination module, used to calibrate the parameters of the photographing device based on the calibration object, obtain calibration data, and obtain the proportional coefficient according to the calibration data;

[0096] YOLO network model, used to obtain face detection data based on face images;

[0097] A first human height prediction module, used to predict the first human height according to the facial detection data;

[0098] A second human body height prediction module, used to predict the second human body height according to the face detection data;

[0099] A third person height prediction module, used to predict the third person height based on the face detection data;

[0100] The sliding mean filtering module is used to perform a sliding mean filtering operation on the first human height, the second human height and the third human height to obtain a final height measurement result.

[0101] In order to verify the effectiveness of this embodiment, an experimental verification was conducted. The experiment measured 50 groups of samples, each containing 10 individuals, for a total of 500 individuals. Each individual was measured 5 times at different distances. Table 1 shows the true height and the average absolute error of two randomly selected groups of samples. It can be seen from Table 1 that the overall average absolute error is about 0.95 cm, which shows the effectiveness of this method in measuring human height.

[0102] Table 1:

[0103]

[0104] The present invention has strong embeddability and a wide range of applications. It can be applied in many scenarios. For example, it can be applied to medical measurement equipment to quickly measure the height of patients, which is convenient for doctors to evaluate and treat patients. It can also be deployed in cameras at sales locations to provide personalized services based on the height of customers. In addition, the method can also be integrated into smart furniture equipment to automatically adjust the height of furniture. The principle is the same as measuring human height.

[0105] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein by equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for monocularly measuring human height, characterized in that: The specific steps include: S1: Select a calibration object, calibrate the parameters of the camera based on the calibration object, obtain calibration data, and obtain a proportional coefficient based on the calibration data; S2: Obtain a facial image based on a shooting device, and input the facial image into a YOLO network model to obtain facial detection data; S3: If the facial detection data only includes eye data, the first human height is calculated based on the eye data, and S4 is executed; if the facial detection data includes eye data and mouth data, the second human height is calculated based on the eye data and the mouth data, and then the third human height of multi-feature fusion is calculated based on the first human height and the second human height, and S4 is executed; S4: performing a sliding mean filtering operation on the first human height, the second human height and the third human height to obtain a final height measurement result.

2. The method for monocularly measuring human height according to claim 1, characterized in that: In S1, a calibration object is selected, and the parameters of the photographing device are calibrated based on the calibration object to obtain calibration data, and the process of obtaining the proportional coefficient according to the calibration data is as follows: Select an object with a regular shape as the calibration object and measure the distance u from the geometric center of the calibration object to the camera. sta , in cm; the length of the calibration object o sta , in cm; the actual length of the calibration object on the display screen i s_sta , unit is cm; Get the average pupil distance of humans e , unit is cm; Measure the distance H from the camera to the ground cg , unit is cm; Measure and determine the actual width w of the image on the display screen in cm, and obtain the number of pixels w of the image width p , the unit is pixel; The proportional coefficient is obtained according to the above calibration data, including: High-order calibration factors: Low-order calibration factors: System pupil distance factor: c = o e -0.82834.

3. The method for monocularly measuring human height according to claim 2, characterized in that: If the facial detection data only includes eye data, the formula for calculating the height of the first person based on the eye data is: H p1 =α·i p_ch1 ·i p_e -3 +β·i p_ch1 ·i p_e -2 +γ·i p_ch1 ·i p_e -1 +H cg i p_ch1 is the distance from the center of the screen to the first pixel on the top of the head, and the calculation formula is: Yo p_ch1 =and mid +y jaw -and le -and re In the formula, y mid is the ordinate of the center of the image; jaw The vertical coordinate of the bottom edge of the face detection frame; y le y is the ordinate of the center of the left eye detection frame; re The vertical coordinate of the center of the right eye detection frame; i p_e is the pixel distance between the two eyes, and i p_e =|x le -x re |, x le is the horizontal coordinate of the center of the left eye detection frame, x re H is the horizontal coordinate of the center of the right eye detection frame; cg is the distance from the camera to the ground; y m It is the vertical coordinate of the center of the mouth detection frame.

4. The method for monocularly measuring human height according to claim 3, characterized in that: If the facial detection data includes eye data and mouth data, the formula for calculating the height of the second person based on the eye data and mouth data is: H p2 =α·i p_ch2 ·i p_e -3 +β·i p_ch2 ·i p_e -2 +γ·i p_ch2 ·i p_e -1 +H cg i p_ch2 The second pixel distance from the center of the screen to the top of the head is calculated as: <h2 style=";text-align:left;direction:ltr">i<h2 style=";text-align:left;direction:ltr"> p_ch2 <h2 style=";text-align:left;direction:ltr"> =y<h2 style=";text-align:left;direction:ltr"> mid <h2 style=";text-align:left;direction:ltr"> +1.4y<h2 style=";text-align:left;direction:ltr"> m <h2 style=";text-align:left;direction:ltr"> -1.2y<h2 style=";text-align:left;direction:ltr"> le <h2 style=";text-align:left;direction:ltr"> -1.2y<h2 style=";text-align:left;direction:ltr"> re In the formula, y mid is the ordinate of the center of the image; m y is the ordinate of the center of the mouth detection frame; le y is the ordinate of the center of the left eye detection frame; re The vertical coordinate of the center of the right eye detection frame; i p_e is the pixel distance between the two eyes, and i p_e =|x le -x re |, x le is the horizontal coordinate of the center of the left eye detection frame, x re H is the horizontal coordinate of the center of the right eye detection frame; cg is the distance from the camera to the ground.

5. The method for monocularly measuring human height according to claim 4, characterized in that: The formula for calculating the third person's height based on the first person's height and the second person's height is: H p3 =α·i p_ch3 ·i p_e -3 +β·i p_ch3 ·i p_e -2 +γ·i p_ch3 ·i p_e -1 +H cg i p_ch3 The third pixel distance from the center of the screen to the top of the head is calculated as: In the formula, y mid is the ordinate of the center of the image; m y is the ordinate of the center of the mouth detection frame; le y is the ordinate of the center of the left eye detection frame; re The vertical coordinate of the center of the right eye detection frame; i p_e is the pixel distance between the two eyes, and i p_e =|x le -x re |, x le is the horizontal coordinate of the center of the left eye detection frame, x re H is the horizontal coordinate of the center of the right eye detection frame; cg is the distance from the camera to the ground.

6. The method for monocularly measuring human height according to claim 5, characterized in that: In S4, a sliding mean filtering operation is performed on the predicted first human height, second human height, and third human height to obtain a final height measurement result including: The formula for performing a sliding mean filtering operation on the first human height, the second human height, and the third human height is: In the formula, H p (t) is the height prediction result after filtering in the tth frame, h p (t) is the height prediction result of the unfiltered wave in the tth frame, n is the number of times d∈[60,90] is satisfied in all time frames; i p_e is the pixel distance between the two eyes, and i p_e =|x le -x re |, x le is the horizontal coordinate of the center of the left eye detection frame, x re It is the horizontal coordinate of the center of the right eye detection frame.

7. A system for measuring human height with a single camera, used to implement the method of claims 1-6, characterized in that: include: A proportional coefficient determination module, used to calibrate the parameters of the photographing device based on the calibration object, obtain calibration data, and obtain the proportional coefficient according to the calibration data; YOLO network model, used to obtain face detection data based on face images; A first human height prediction module, used to predict the first human height according to the facial detection data; A second human body height prediction module, used to predict the second human body height according to the face detection data; A third person height prediction module, used to predict the third person height based on the face detection data; The sliding mean filtering module is used to perform a sliding mean filtering operation on the first human height, the second human height and the third human height to obtain a final height measurement result.

Images