CN108564586A - A kind of body curve's measurement method and system based on deep learning - Google Patents
A kind of body curve's measurement method and system based on deep learning Download PDFInfo
- Publication number
- CN108564586A CN108564586A CN201810651248.2A CN201810651248A CN108564586A CN 108564586 A CN108564586 A CN 108564586A CN 201810651248 A CN201810651248 A CN 201810651248A CN 108564586 A CN108564586 A CN 108564586A
- Authority
- CN
- China
- Prior art keywords
- human body
- points
- point
- distance
- user
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0012—Biomedical image inspection
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/103—Detecting, measuring or recording devices for testing the shape, pattern, colour, size or movement of the body or parts thereof, for diagnostic purposes
- A61B5/107—Measuring physical dimensions, e.g. size of the entire body or parts thereof
- A61B5/1077—Measuring of profiles
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/80—Analysis of captured images to determine intrinsic or extrinsic camera parameters, i.e. camera calibration
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30196—Human being; Person
Abstract
The invention discloses a kind of body curve's measurement method and system based on deep learning input user's human body image, are detected to input picture and demarcate and then body curve's measurement data is calculated by obtaining human body segmentation and skeleton point regression data library.Acquisition methods proposed by the present invention, with reference to user's human body picture of object, can easily carry out contactless humanbody curved measurement by input tape, improve the precision and Discussing Convenience of measurement.
Description
Technical field
The present invention relates to technical field of computer vision, especially a kind of body curve's measurement method based on deep learning
And system.
Background technology
Human testing is widely used in life, such as monitoring system, safety monitor, automatic Pilot and driving auxiliary system
System, human-computer interaction, interaction entertainment, wired home and old man's auxiliary, therapeutic treatment etc., is widely applied and challenge has attracted very much
Researcher, which participates in, wherein to come.
Existing body curve measures the contact type measurement being confined to equipment, and needs the cooperation of itself calibrated
True measurement Human Height, as the measurement of slide calliper rule formula and ultrasonic sensing instrument for measuring height are required to human body station and are put down what equipment was specified
It could complete, and can only measure one by one on platform.
If existing patent CN2548558Y discloses a kind of human vertebra, back curve measurement module, by being evenly equipped with through-hole
Poroid template and the measuring rod that is plugged in each through-hole with transition fit system form.First by human vertebra, back when use
The measuring rod of curved measurement template all slides to the side of template, is then placed at the backrest of human body, it is desirable that being measured people will
Measuring point is lightly pressed in curved measurement template, and the measuring rod being arranged in template will act on it because of measuring point
The different and inside retractions of power, the measuring rod mould being distributed in template carves the curve that human body is measured position, according to measuring rod
Retraction amount can approximatively draw the vertebra of human body or the physiological curve at back.Above-mentioned measurement method and device measurement efficiency are low
And it is unfriendly to measurement object, long-range or non-contact measurement cannot be carried out.
Computer vision is to use a kind of simulation of computer and relevant device to biological vision.Its main task is exactly
It is handled by picture to acquisition or video to obtain the three-dimensional information of corresponding scene, just as the mankind and many other classes give birth to
As object is done daily.
Monocular vision refers to completing positioning work merely with a video camera.Because it only needs a visual sensor, monocular
Camera needs that target is identified, that is to say, that before measuring first identification target be vehicle, people or other.In this base
The parameter measurement to target is carried out on plinth again.
Binocular vision technology is a kind of important form of computer vision, based on principle of parallax and utilizes different location
Two cameras Same Scene is shot, by binocular calibration and matching technique, obtain the three-dimensional information of target in scene.
Binocular ranging refers to the alignment for establishing the two width projected image target points under different visual angles of the target point under same scene.
Single Binocular vision photogrammetry method has many advantages, such as that efficient, precision is suitable, system structure is simple, at low cost, very
It is suitable for online, non-contact product testing and the quality control at manufacture scene.
Invention content
The present invention is directed to solve at least some of the technical problems in related technologies.For this purpose, the present invention carries
Go out a kind of body curve's measurement method and system based on deep learning, can easily carry out contactless humanbody curve survey
Amount, improves the precision and Discussing Convenience of measurement.
The present invention proposes a kind of body curve's measurement method based on deep learning, including:
Step 1:Obtain human body segmentation and skeleton point regression data library;
Step 2:Input user's human body image;
Step 3:Calibration and extraction are detected to the image of input using human body segmentation and skeleton point regression data library,
User's body curve's measurement data is calculated.
Further, it obtains human body segmentation in step 1 and skeleton point regression data library specifically includes:Training stage and survey
The examination stage.
Further, the training stage includes:Human body picture is chosen as training sample;Human body picture is pre-processed;
CNN and RPN training is carried out to the obtained picture of pretreatment, obtains target category score, target frame, human body segmentation's region unit and
Skeleton point regression model.
Further, test phase includes:Input picture frame;Read trained target category score, target frame, people
Body cut zone block and skeleton point Parameters in Regression Model;It calls a pictures frame to do propagated forward, obtains the mesh of input picture frame
Mark category score, target frame, human body segmentation's area data and bone point data.
Further, the user's human body image inputted in step 2 includes being obtained using monocular cam or binocular camera
User's human body direct picture and user's human body side image.
Further, the user's human body image obtained using monocular cam includes user's human body front with object of reference
Image and user's human body side image.
Further, in step 3 to input picture be detected calibration include to object of reference angle point be detected calibration and
Human body contour outline and skeleton point are extracted;Wherein, include extracting human body segmentation's number of regions to human body contour outline and skeleton point extraction
According to, according to human body segmentation's area data, extract human body segmentation's edges of regions coordinate, by edge coordinate from the beginning it is open-top begin by up to
Under, Coutinuous store from left to right, and bone point coordinates is obtained according to the skeleton point regression model.
Further, when the object of reference is gridiron pattern, calibration is detected to object of reference angle point and is specifically included:Detection
Gridiron pattern in picture, according to the X-comers data detected, by each angular coordinate detected by from left to right, from upper
Sequential storage under calculates the length of X and each pixel expression of Y-direction, above-mentioned multiple length is averaged to obtain each
The physical length that pixel indicates.
Further, user's human body direct picture based on acquisition is calculated human body front measurement data and specifically includes:
Human body contour outline is extracted since cephalad apex, the first concave point near zone in definition right side is averagely worth to D points, definition left side the
One concave point near zone is averagely worth to C points, and it is that front neck is wide to define C points to D point X-direction differences, defines C points and is pushed up to head
Point is head height;It is A points that the upward distance l of left eye eyeball skeleton point, which is defined, with left side human body contour outline intersection point, defines left eye eyeball skeleton point
The upward distance l is B points with right side human body contour outline intersection point, and it is that front is wide to define A points to B point X-direction differences;Define left shoulder
A skeleton point left side angle [alpha] and left side human body contour outline intersection point E points obliquely, define the right shoulder skeleton point right side obliquely the angle [alpha] with
Right side human body contour outline intersection point F points, definition E points to F point X-direction differences are shoulder breadth;Left shoulder skeleton point is defined to left elbow skeleton point
Distance is upper left arm lengths EK, and the distance for defining right shoulder skeleton point to right elbow skeleton point is upper right arm lengths FM, defines left elbow bone
The distance of bone point to left wrist skeleton point is lower-left arm lengths KL, and the distance for defining right elbow skeleton point to right wrist skeleton point is lower-left arm
Length MN, it is (EK+FM)/2 to define upper arm lengths, and it is (KL+MN)/2 to define lower arm lengths;Define right and left shoulders bone central point with
The distance of left and right hip joint dot center point is that body is high, and defining the E points right side, the places a distance m are G points obliquely, and a definition F points left side is obliquely
It is H points at one distance m, definition G points to H points X-direction distance are positive chest breadth;Define the upward distance n of left hip joint and left side
Profile intersection point is I points, and defining right hip joint, the distance n and right lateral contours intersection point are J points upwards, define I points to J point X-directions
Distance is that front waist is wide.
Further, user's human body side image based on acquisition is calculated human body side measurement data and specifically includes:
Human body contour outline is extracted since cephalad apex, the first concave point near zone in definition left side is averagely worth to Q points, defines Q points and arrives
Cephalad apex Y-direction height is side head height;It is R points to define on the right side of Q points X-directions with intersection point on the right side of human body contour outline, define Q points and
The distance of R point X-directions is that side neck is wide;It is O points that the upward distance l ' of eyes skeleton point, which is defined, with left side human body contour outline intersection point, fixed
The distance l ' and right side human body contour outline intersection point are P points to ocular prosthesis eyeball skeleton point upwards, and it is side to define O points with P points X-direction distance
It is wide;It defines right and left shoulders bone central point and is side surface body height at a distance from left and right hip joint dot center point;Define right and left shoulders bone
Central point is S points to next distance m ' and left side human body contour outline intersection point, defines the right and left shoulders bone central point distance m ' downwards
It is T points with right side human body contour outline intersection point, it is side chest breadth to define S points with T points X-direction distance;Define left and right hip joint central point
Upward distance n ' is U points with left side human body contour outline intersection point, defines left and right hip joint central point the distance n ' and right side upwards
Human body contour outline intersection point is V points, and it is that side waist is wide to define U points with V points X-direction distance.
Further, according to the above-mentioned human body a front surface and a side surface measurement data being calculated, by oval calculation formula or
Person is more, and curve combining formula calculates body curve's measurement data.
Further, further include step 4:Export the user body curve measurement data.
Body curve's measuring system based on deep learning that the invention also provides a kind of, including:
Human body segmentation and skeleton point regression data library acquisition module, for obtaining human body segmentation and skeleton point regression data
Library;
User's human body image acquisition module, for receiving user's human body image;
Body curve's computing module, for the user body curve measurement data to be calculated.
Further, human body segmentation and skeleton point regression data library acquisition module further comprise training module and test mould
Block.
Further, training module is used for:Human body picture is chosen as training sample;Human body picture is pre-processed;
CNN and RPN training is carried out to the obtained picture of pretreatment, obtains target category score, target frame, human body segmentation's region unit and
Skeleton point regression model.
Further, test module is used for:Input picture frame;Read trained target category score, target frame, people
Body cut zone block and skeleton point Parameters in Regression Model;It calls a pictures frame to do propagated forward, obtains the mesh of input picture frame
Mark category score, target frame, human body segmentation's area data and bone point data.
Further, the image that user's human body image acquisition module receives includes using monocular cam or binocular camera
The user's human body direct picture and user's human body side image of acquisition.
Further, the user's human body image obtained using the monocular cam includes the user with object of reference
Human body direct picture and user's human body side image.
Further, body curve's computing module further comprises detecting demarcating module and human body contour outline and skeleton point extraction
Module;Wherein, detection demarcating module is used for, and calibration is detected to object of reference angle point;Human body contour outline and skeleton point extraction module
For, human body segmentation's area data is extracted, according to human body segmentation's area data, extracts human body segmentation's edges of regions coordinate,
By the edge coordinate from the beginning the open-top beginning from top to bottom, Coutinuous store from left to right, and according to the skeleton point regression model
Obtain bone point coordinates.
Further, when the object of reference is gridiron pattern, calibration is detected to object of reference angle point and is specifically included:Detection
Gridiron pattern in picture, according to the X-comers data detected, by each angular coordinate detected by from left to right, from
Top to bottm obtains sequential storage, calculates the length of X and each pixel expression of Y-direction, above-mentioned multiple length is averaged to obtain every
The physical length that a pixel indicates.
Further, body curve's computing module further comprises human region computing module, human region computing module
Including human body front surface region computing module and human body lateral side regions computing module.
Further, people is calculated for user's human body direct picture based on acquisition in human body front surface region computing module
Body front measurement data, specially:Human body contour outline is extracted since cephalad apex, first, right side of definition concave point near zone is flat
D points are worth to, the first concave point near zone in definition left side is averagely worth to C points, defines C points to D point X-direction differences as just
Face neck is wide, and definition C points to cephalad apex are head height;Define the upward distance l of left eye eyeball skeleton point and left side human body contour outline intersection point
For A points, defining left eye eyeball skeleton point, the distance l and right side human body contour outline intersection point are B points upwards, define A points to B point X-directions
Difference is that front is wide;A left shoulder skeleton point left side angle [alpha] and left side human body contour outline intersection point E points obliquely are defined, right shoulder bone is defined
The angle [alpha] and right side human body contour outline intersection point F points, definition E points to F point X-direction differences are shoulder breadth obliquely on the bone point right side;Definition is left
The distance of shoulder skeleton point to left elbow skeleton point is upper left arm lengths EK, and the distance for defining right shoulder skeleton point to right elbow skeleton point is the right side
Upper arm lengths FM, the distance for defining left elbow skeleton point to left wrist skeleton point is lower-left arm lengths KL, defines right elbow skeleton point to the right side
The distance of wrist skeleton point is lower-left arm lengths MN, and it is (EK+FM)/2 to define upper arm lengths, and it is (KL+MN)/2 to define lower arm lengths;
Define right and left shoulders bone central point with it is high for body at a distance from left and right hip joint dot center point, the definition E points right side is obliquely at a distance m
For G points, defining a F points left side, the places the distance m are H points obliquely, define G points to H points X-direction distance be front chest breadth;Definition is left
The upward distance n of hip joint is I points with left side profile intersection point, defines right hip joint the distance n and right lateral contours intersection point upwards
For J points, it is wide for front waist to J points X-direction distance to define I points.
Further, people is calculated for user's human body side image based on acquisition in human body lateral side regions computing module
Body side surface measurement data, specially:Human body contour outline is extracted since cephalad apex, first, left side of definition concave point near zone is flat
Q points are worth to, definition Q points to cephalad apex Y-direction height are side head height;It defines right with human body contour outline on the right side of Q points X-direction
Top-cross point is R points, and the distance for defining Q points and R point X-directions is that side neck is wide;Define the upward distance l ' of eyes skeleton point and a left side
Side human body contour outline intersection point is O points, and defining eyes skeleton point, the distance l ' and right side human body contour outline intersection point are P points upwards, define O
Point is that side is wide with P points X-direction distance;It defines right and left shoulders bone central point and is side at a distance from left and right hip joint dot center point
Face body is high;It is S points that right and left shoulders bone central point, which is defined, to next distance m ' and left side human body contour outline intersection point, defines right and left shoulders bone
The distance m ' and right side human body contour outline intersection point are T points to central point downwards, and it is side chest breadth to define S points with T points X-direction distance;
It is U points that the upward distance n ' of left and right hip joint central point, which is defined, with left side human body contour outline intersection point, defines left and right hip joint central point
The upward distance n ' is V points with right side human body contour outline intersection point, and it is that side waist is wide to define U points with V points X-direction distance.
Further, human region computing module leads to according to the above-mentioned human body a front surface and a side surface measurement data being calculated
It crosses oval calculation formula or more curve combining formula calculates body curve's measurement data.The additional aspect and advantage of the present invention
It will be set forth in part in the description, partly will become apparent from the description below, or practice understanding through the invention
It arrives.
Description of the drawings
The above-mentioned and/or additional aspect and advantage of the present invention will become in the description from combination following accompanying drawings to embodiment
Obviously and it is readily appreciated that, wherein:
Fig. 1 is a kind of flow chart of human body back curve acquisition methods based on camera of the embodiment of the present invention.
Fig. 2 is the training stage of the embodiment of the present invention and the flow chart of test phase.
Fig. 3 is the training detail flowchart of the embodiment of the present invention.
Fig. 4 is the human body direct picture extraction schematic diagram of the embodiment of the present invention.
Fig. 5 is the human body side image extraction schematic diagram of the embodiment of the present invention.
Fig. 6 is gridiron pattern schematic diagram used in the embodiment of the present invention.
Fig. 7 is a kind of composition figure of body curve's measuring system based on camera of the embodiment of the present invention.
Specific implementation mode
The embodiment of the present invention is described below in detail, examples of the embodiments are shown in the accompanying drawings, wherein from beginning to end
Same or similar label indicates same or similar element or element with the same or similar functions.Below with reference to attached
The embodiment of figure description is exemplary, it is intended to for explaining the present invention, and is not considered as limiting the invention.
In the description of this specification, reference term " one embodiment ", " some embodiments ", " example ", " specifically show
The description of example " or " some examples " etc. means specific features, structure, material or spy described in conjunction with this embodiment or example
Point is included at least one embodiment or example of the invention.In the present specification, schematic expression of the above terms are not
It must be directed to identical embodiment or example.Moreover, particular features, structures, materials, or characteristics described can be in office
What can be combined in any suitable manner in one or more embodiments or example.In addition, those skilled in the art can say this
Different embodiments or examples described in bright book are engaged and are combined.
Below in conjunction with the accompanying drawings and with reference to the specific embodiment description present invention.It is based on as shown in Figure 1, the present invention proposes one kind
The human body back curve acquisition methods of deep learning, include the following steps:Step 1:It obtains human body segmentation and skeleton point returns number
According to library;Step 2:Input user's human body image;Step 3:Using human body segmentation and skeleton point regression data library to the figure of input
As being detected calibration and extraction, user's body curve's measurement data is calculated.The specific of each step is introduced separately below
Flow:
Step 1:Human body segmentation and skeleton point regression data library are obtained, is specifically included:Training stage and test phase.Such as
Shown in Fig. 2,
For the training stage, choose a large amount of human body contour outline pictures as training sample, picture pre-processed, so as into
Row postorder is trained.RPN training is carried out to carrying out pretreated picture, multiple human body Probability Areas is obtained, carries out CNN instructions later
Practice, obtains final human body segmentation's database.
Fig. 3 shows the detail flowchart of training stage, and the training stage is by Body R-CNN networks and RPN network two parts
Composition.Shared convolutional layer is based on ResNext-101 networks and extracts feature, and candidate target is extracted using RPN (region suggestion network)
Regional frame, and be mapped in artwork, different size of characteristic area is aligned to fixed size characteristic pattern.Training stage is fixed
Adopted multitask loss function L=Lcls+Lbox+Lmask+Lkeypoint, LclsIt is Classification Loss function, obtains target classification score, Lbox
It is target frame loss function, obtains target frame, LmaskIt is human body segmentation's loss function, obtains human body segmentation's data, Lkeypoint
It is skeleton point loss function, obtains skeleton point regression model, four branch's joint cross-trainings improves whole robust
Property.
Concrete example illustrates training process below:
First, input picture size is 224 × 224 × 3 (3 are three channels, that is, tri- kinds of RGB).Then, first layer
Convolution kernel dimension be 7 × 7 × 3 × 96, so that conv1 is obtained the result is that 110 × 110 × 96 (110 come from (224-7+
Pad)/2+1, pad are fillings, i.e., pixel is supplemented around picture, the purpose for the arrangement is that in order to divide exactly divided by 2 are
Because 2 be the stride in figure).It is exactly to do a pond later, obtains pool1, the size of the core in pond is 3 × 3, so pond
The dimension of picture is 55 × 55 × 96 ((110-3+pad)/2+1=55) after change;It is again convolution with that followed by, specifically
The dimension of convolution kernel be 5 × 5 × 96 × 256, obtain conv2:26×26×256;It is exactly similar process below, finally takes
The output of conv5, that is, 13 × 13 × 256 give RPN networks.
The sliding window size of RPN networks is 3 × 3, it is only necessary to the convolution of 4 dimensions as one 3 × 3 × 256 × 256
Core, so that it may with the vector for tieing up each 3 × 3 sliding window convolution at one 256.Cls layer have 18 output nodes, that
One 1 × 1 × 256 × 18 convolution kernel is used between 256-d and cls layer, so that it may to obtain cls layer, this
A 1 × 1 × 256 × 18 convolution kernel is exactly full connection.The output of reg layer is 36, thus corresponding convolution kernel be 1 ×
1 × 256 × 36, it can be obtained by outputing for reg layer in this way.Then it can all be connect behind cls layer and reg layer
Onto the loss function of oneself, the value of loss function is provided, while can be according to derivation as a result, providing the data of backpropagation.
For test phase, trained model parameter is read, calls a pictures to do propagated forward and obtains target classification
Score, target frame, human body segmentation's area data and bone point data.
Step 2:Input user's human body image.
Monocular cam or binocular camera acquisition can be used in incoming picture.It is required that user is shone towards on the right side of picture and front
Each one of piece, and clap and take whole upper part of the body image, as shown in Figure 4 and Figure 5.To the direct picture of user and the number of side image
When according to acquisition, it is ensured that the alignment of coordinate.The acquisition of data carries out under certain coordinate system, ensures direct picture and side
The co-ordinate zero point of image data acquiring is identical, such as all extracts waistline from hip joint a certain distance, the extraction of front waistline is with side
The waistline in face extracts on sustained height, ensures that the extraction line of this two waistlines is all unified in direct picture and side image
, mentioned above principle is also deferred to for the extraction at other positions.When acquiring user's human body image using monocular cam, user side
Face and a gridiron pattern is arranged below as object of reference, it includes tessellated user side and direct picture to need acquisition.
Further, gridiron pattern is not limited to for the object of reference of monocular cam, arbitrarily there are fixed known dimensions
Object of reference, such as identity card, A4 paper, mobile phone.
Step 3:Calibration is detected to the image of input using human body segmentation and skeleton point regression data library, is calculated
To user's body curve's measurement data.
Step 3.1:Calibration is detected to image;
For the user images obtained by monocular cam, need to be demarcated using object of reference such as gridiron pattern.Such as Fig. 6
It is shown, detect 4 × 8 gridiron patterns in picture, 20 millimeters of each square physical length of gridiron pattern, according to the X-comers detected
Data, each angular coordinate detected from left to right, sequential storage from top to bottom.It is possible thereby to calculate X and each picture of Y-direction
The length that element indicates averages above-mentioned multiple length to obtain the physical length of each pixel expression.
Preferably, gridiron pattern longitudinally includes at least 2 row, and gridiron pattern laterally includes at least 6 rows, it is furthermore preferred that gridiron pattern is vertical
It is arranged to comprising 2-6, gridiron pattern includes laterally 6-12 rows, and more preferably, for gridiron pattern longitudinally comprising 4 row, gridiron pattern includes laterally 8
Row.Color can be that arbitrary two kinds of colors are intersected in gridiron pattern, it is preferred that color is intersected for black-and-white two color in gridiron pattern.
Further, each grid length of side of gridiron pattern can be that any limit is long, as long as can be identified from camera.It is preferred that
, each grid length of side of the gridiron pattern is 10-50 millimeters, it is furthermore preferred that each grid length of side of the gridiron pattern is 15-
25 millimeters, it is furthermore preferred that each grid length of side of the gridiron pattern is 20 millimeters, when the gridiron pattern length of side of the present invention is
At 10-50 millimeters, human body back curve acquisition methods measurement accuracy highest.
Further, gridiron pattern is not limited to for the calibration of monocular cam, the arbitrary ginseng with fixed known dimensions
Above-mentioned detection staking-out work is may be incorporated for according to object.Such as identity card, A4 paper, mobile phone.User claps and takes towards picture right side
Complete upper part of the body image is placed the objects of reference such as the identity card, A4 paper, mobile phone as object of reference at its back, is imaged using monocular
Head acquisition is passed to picture after the image with above-mentioned object of reference.The boundary of identity card, A4 paper, mobile phone in detection picture etc.,
According to the vertex data detected, by each apex coordinate from left to right, sequential storage from top to bottom.It is possible thereby to calculate each
The physical length on side.
User images can also be obtained by binocular camera, for the user images obtained by binocular camera, no
It needs to carry out gridiron pattern calibration.Using binocular camera photographed scene image, the number of people point of human body target in scene image is obtained
The image coordinate of point.Before using binocular camera photographed scene image, mounting height and the installation of binocular camera are coped with
Angle is demarcated, and setting angle includes pitch angle and the inclination angle of binocular camera.Firstly, it is necessary to measure scene using graduated scale
Vertical range mounting height as binocular camera of the ground to camera;Secondly, known altitude is captured using binocular camera
The image of target multiple location points in the scene;Again, multigroup different location point target that binocular camera captures image is obtained
Depth information.Minimization problem, the pitch angle of one group of binocular camera of solution and inclination angle are constructed, makes utilization that should be based on binocular vision
The variance of the actual height for the object height and target that the human body back curve measurement method of feel technology acquires is minimum, finally obtains
The pitch angle and inclination angle of one group of optimal binocular camera pitch angle and inclination angle as binocular camera.
Step 3.2, human body contour outline and skeleton point are extracted.It specifically includes:Human body segmentation's area data is extracted, according to
Human body segmentation's area data extracts human body segmentation's edges of regions coordinate, by edge coordinate from the beginning the open-top beginning from top to bottom, by it is left extremely
Fight continuity stores, and obtains bone point coordinates according to the skeleton point regression model.
Step 3.3, the user's human body direct picture and user's human body side image for being based respectively on acquisition calculate separately to obtain
Measurement data.
With reference to Fig. 4 elaborate the present invention is based on user's human body direct picture of acquisition be calculated human body front survey
The preferred embodiment for measuring data, specifically includes:Human body contour outline is extracted since cephalad apex, near first concave point in definition right side
Region averages obtain D points, and the first concave point near zone in definition left side is averagely worth to C points, it is poor to D point X-directions to define C points
Value is that front neck is wide, and definition C points to cephalad apex are head height;Define upward 1/5 head height of left eye eyeball skeleton point and left side human body wheel
Wide intersection point is A points, and it is B points to define upward 1/5 head height of left eye eyeball skeleton point with right side human body contour outline intersection point, defines A points to B points X
Direction difference is that front is wide;A left shoulder skeleton point left side 45 degree and left side human body contour outline intersection point E points obliquely are defined, right shoulder bone is defined
45 degree and right side human body contour outline intersection point F points, definition E points to F point X-direction differences are shoulder breadth obliquely on the bone point right side;Define left shoulder bone
The distance of bone point to left elbow skeleton point is upper left arm lengths EK, and the distance for defining right shoulder skeleton point to right elbow skeleton point is right upper arm
Length FM, the distance for defining left elbow skeleton point to left wrist skeleton point is lower-left arm lengths KL, defines right elbow skeleton point to right carpal bone
The distance of bone point is lower-left arm lengths MN, and it is (EK+FM)/2 to define upper arm lengths, and it is (KL+MN)/2 to define lower arm lengths;Definition
Right and left shoulders bone central point is high for body at a distance from left and right hip joint dot center point, defines the E points downward 1/5 body eminences of 20mm to the right
For G points, defining F points, the downward 1/5 body eminences of 20mm are H points to the left, and definition G points to H points X-direction distance are positive chest breadth;Definition
The left upward 1/5 body height of hip joint is I points with left side profile intersection point, defines the upward 1/5 body height of right hip joint and right lateral contours intersection point
For J points, it is wide for front waist to J points X-direction distance to define I points.
It continues with and illustrates that the present invention is based on user's human body side images of acquisition, and human body side is calculated in conjunction with attached drawing 5
The preferred embodiment of measurement data, specifically includes:Human body contour outline is extracted since cephalad apex, first concave point in definition left side is attached
Near field is averagely worth to Q points, and definition Q points to cephalad apex Y-direction height are side head height;Define Q points X-direction right side and people
Intersection point is R points on the right side of body profile, and the distance for defining Q points and R point X-directions is that side neck is wide;Define upward 1/5 side of eyes skeleton point
Face head height is O points with left side human body contour outline intersection point, defines the upward 1/5 side head height of eyes skeleton point and right side human body contour outline intersection point
For P points, it is that side is wide to define O points with P points X-direction distance;Define right and left shoulders bone central point and left and right hip joint dot center
The distance of point is that side surface body is high;It is S that the downward 1/5 side surface body height of right and left shoulders bone central point, which is defined, with left side human body contour outline intersection point
Point, it is T points to define the downward 1/5 side surface body height of right and left shoulders bone central point with right side human body contour outline intersection point, defines S points and the T points side X
It is side chest breadth to distance;It is U points that the upward 1/5 side surface body height of left and right hip joint central point, which is defined, with left side human body contour outline intersection point,
It is V points that the upward 1/5 side surface body height of left and right hip joint central point, which is defined, with right side human body contour outline intersection point, defines U points and V point X-directions
Distance is that side waist is wide.
Those skilled in the art could be aware that, the selection in above preferred embodiment about distance and angle, such as 1/5
Height, 1/5 body height, 20mm, 45 degree etc., can according to actual image zooming-out and measure situation be suitably adjusted, and
It is not limited to above-mentioned specific data.
Step 3.4, according to the above-mentioned human body a front surface and a side surface measurement data being calculated, by oval calculation formula or
More curve combining formula calculate body curve's measurement data.
Step 4 exports the user body curve measurement data.User's body curve's measurement data of above-mentioned output can
For user or need to refer to using the scene of the data.Such as when user carries out customed mattress or customed clothes
When, by using the above-mentioned body curve's measurement data being calculated, it can preferably select the mattress for being suitble to oneself or dress ornament.
The invention also provides a kind of, and the body curve based on deep learning obtains system, as shown in fig. 7, the system packet
It includes:Human body segmentation and skeleton point regression data library acquisition module, for obtaining human body segmentation and skeleton point regression data library;User
Human body image acquisition module, for receiving user's human body image.Body curve's computing module, for the user people to be calculated
Body curve measurement data.
Wherein, human body segmentation and skeleton point regression data library acquisition module further comprise training module and test module.
Training module is used for:Human body picture is chosen as training sample;Human body picture is pre-processed;To pre-processing
The picture arrived carries out CNN and RPN training, obtains target category score, and target frame, human body segmentation's region unit and skeleton point return
Model.
Test module is used for:Input picture frame;Read trained target category score, target frame, human body segmentation area
Domain block and skeleton point Parameters in Regression Model;A pictures frame is called to do propagated forward, the target category for obtaining input picture frame obtains
Point, target frame, human body segmentation's area data and bone point data.
Specific training and test process are as described above, details are not described herein again.
The image that user's human body image acquisition module receives includes the use obtained using monocular cam or binocular camera
Family human body direct picture and user's human body side image.
When obtaining user's human body back image using monocular cam, user's human body image includes the use with object of reference
Family human body direct picture and user's human body side image.As shown in Figure 3-4, when using monocular cam acquisition user's human body image
When, user side and a gridiron pattern is arranged below as object of reference, it includes tessellated user side and front elevation to need acquisition
Picture.
Further, gridiron pattern is not limited to for the object of reference of monocular cam, arbitrarily there are fixed known dimensions
Object of reference, such as identity card, A4 paper, mobile phone.
Body curve's computing module further comprises detecting demarcating module and human body contour outline and skeleton point extraction module;Its
In, detection demarcating module is used for, and calibration is detected to object of reference angle point;Human body contour outline and skeleton point extraction module are used for, and are carried
Human body segmentation's area data is taken, according to human body segmentation's area data, human body segmentation's edges of regions coordinate is extracted, by the side
Edge coordinate from the beginning the open-top beginning from top to bottom, Coutinuous store from left to right, and bone is obtained according to the skeleton point regression model
Point coordinates.
As shown in figure 3, when the object of reference is gridiron pattern, detection demarcating module is detected calibration to object of reference angle point
It specifically includes:Detect the gridiron pattern in picture, according to the X-comers data detected, each angular coordinate that will be detected
By from left to right, sequential storage is obtained from top to bottom, the length of X and each pixel expression of Y-direction is calculated, by above-mentioned multiple length
It averages to obtain the physical length that each pixel indicates.
Preferably, gridiron pattern longitudinally includes at least 2 row, and gridiron pattern laterally includes at least 6 rows, it is furthermore preferred that gridiron pattern is vertical
It is arranged to comprising 2-6, gridiron pattern includes laterally 6-12 rows, and more preferably, for gridiron pattern longitudinally comprising 4 row, gridiron pattern includes laterally 8
Row.Color can be that arbitrary two kinds of colors are intersected in gridiron pattern, it is preferred that color is intersected for black-and-white two color in gridiron pattern.
Further, each grid length of side of gridiron pattern can be that any limit is long, as long as can be identified from camera.It is preferred that
, each grid length of side of the gridiron pattern is 10-50 millimeters, it is furthermore preferred that each grid length of side of the gridiron pattern is 15-
25 millimeters, it is furthermore preferred that each grid length of side of the gridiron pattern is 20 millimeters, when the gridiron pattern length of side of the present invention is
At 10-50 millimeters, human body back curve acquisition methods measurement accuracy highest.
Further, gridiron pattern is not limited to for the detection of monocular cam calibration, it is arbitrary that there are fixed known dimensions
Object of reference may be incorporated for above-mentioned detection staking-out work.Such as identity card, A4 paper, mobile phone.User towards on the right side of picture, and
Bat takes whole upper part of the body image, the objects of reference such as the identity card, A4 paper, mobile phone as object of reference is placed at its back, using monocular
Camera acquisition is passed to picture after the image with above-mentioned object of reference.The side of identity card, A4 paper, mobile phone in detection picture etc.
Boundary, according to the vertex data detected, by each apex coordinate from left to right, sequential storage from top to bottom.It is possible thereby to calculate
The physical length on each side.
User images can also be obtained by binocular camera, for the user images obtained by binocular camera, no
It needs to carry out gridiron pattern calibration.Using binocular camera photographed scene image, the number of people point of human body target in scene image is obtained
The image coordinate of point.Before using binocular camera photographed scene image, mounting height and the installation of binocular camera are coped with
Angle is demarcated, and setting angle includes pitch angle and the inclination angle of binocular camera.Firstly, it is necessary to measure scene using graduated scale
Vertical range mounting height as binocular camera of the ground to camera;Secondly, known altitude is captured using binocular camera
The image of target multiple location points in the scene;Again, multigroup different location point target that binocular camera captures image is obtained
Depth information.Minimization problem, the pitch angle of one group of binocular camera of solution and inclination angle are constructed, makes utilization that should be based on binocular vision
The variance of the actual height for the object height and target that the human body back curve measurement method of feel technology acquires is minimum, finally obtains
The pitch angle and inclination angle of one group of optimal binocular camera pitch angle and inclination angle as binocular camera.
Body curve's computing module further comprises human region computing module, human region computing module include human body just
Face area calculation module and human body lateral side regions computing module.
Wherein, human body is being calculated just for user's human body direct picture based on acquisition in human body front surface region computing module
Planar survey data, specially:Human body contour outline is extracted since cephalad apex, the first concave point near zone average value in definition right side
D points are obtained, the first concave point near zone in definition left side is averagely worth to C points, and definition C points to D point X-direction differences are front neck
Width, definition C points to cephalad apex are head height;It is A that upward 1/5 head height of left eye eyeball skeleton point, which is defined, with left side human body contour outline intersection point
Point, it is B points to define upward 1/5 head height of left eye eyeball skeleton point with right side human body contour outline intersection point, and define A points is to B point X-direction differences
Front is wide;A left shoulder skeleton point left side 45 degree and left side human body contour outline intersection point E points obliquely are defined, it is oblique to define the right shoulder skeleton point right side
Upper 45 degree are shoulder breadth with right side human body contour outline intersection point F points, definition E points to F point X-direction differences;Left shoulder skeleton point is defined to left elbow
The distance of skeleton point is upper left arm lengths EK, and the distance for defining right shoulder skeleton point to right elbow skeleton point is upper right arm lengths FM, fixed
The distance of the left elbow skeleton point of justice to left wrist skeleton point is lower-left arm lengths KL, define right elbow skeleton point to right wrist skeleton point distance
For lower-left arm lengths MN, it is (EK+FM)/2 to define upper arm lengths, and it is (KL+MN)/2 to define lower arm lengths;Define right and left shoulders bone
Central point is high for body at a distance from left and right hip joint dot center point, and defining E points, the downward 1/5 body eminences of 20mm are G points, definition to the right
The downward 1/5 body eminences of 20mm are H points to F points to the left, and definition G points to H points X-direction distance are positive chest breadth;Define left hip joint to
Upper 1/5 body height is I points with left side profile intersection point, and it is J points to define the upward 1/5 body height of right hip joint with right lateral contours intersection point, defines I
Point is wide for front waist to J points X-direction distance.
Human body lateral side regions computing module is calculated human body side for user's human body side image based on acquisition and surveys
Data are measured, specially:Human body contour outline is extracted since cephalad apex, the first concave point near zone in definition left side is averagely worth to
Q points, definition Q points to cephalad apex Y-direction height are side head height;It defines on the right side of Q points X-direction and is with intersection point on the right side of human body contour outline
R points, the distance for defining Q points and R point X-directions is that side neck is wide;Define the upward 1/5 side head height of eyes skeleton point and left side human body
Profile intersection point is O points, and it is P points to define the upward 1/5 side head height of eyes skeleton point with right side human body contour outline intersection point, defines O points and P
Point X-direction distance is that side is wide;It defines right and left shoulders bone central point and is side surface body at a distance from left and right hip joint dot center point
It is high;It is S points that the downward 1/5 side surface body height of right and left shoulders bone central point, which is defined, with left side human body contour outline intersection point, defines right and left shoulders bone
The downward 1/5 side surface body height of central point is T points with right side human body contour outline intersection point, and it is side chest breadth to define S points with T points X-direction distance;
It is U points that the upward 1/5 side surface body height of left and right hip joint central point, which is defined, with left side human body contour outline intersection point, defines left and right hip joint center
The upward 1/5 side surface body height of point is V points with right side human body contour outline intersection point, and it is that side waist is wide to define U points with V points X-direction distance.
Those skilled in the art could be aware that, the selection in above-described embodiment about distance and angle, such as 1/5 head height, 1/5
Body height, 20mm, 45 degree etc., according to actual image zooming-out and situation can be measured be suitably adjusted, and be not limited to
Above-mentioned specific data.
Further, according to the above-mentioned human body a front surface and a side surface measurement data being calculated, by oval calculation formula or
Person is more, and curve combining formula calculates body curve's measurement data.
Finally, it further includes user body curve output that a kind of body curve based on deep learning of the invention, which obtains system,
Module is such as shown by the display screen of mobile phone or computer for exporting and showing the user body curve measurement data
Etc..
The present invention proposes a kind of body curve's measurement method and system based on deep learning, it is proposed that a kind of end-to-end people
Dividing method is surveyed in physical examination, and one picture with object of reference of input can be detected it by trained database
Calibration and extraction measure, and then finally obtain body curve's measurement data, and can be completed in the case where coordinating without personnel can
The measurement of multiple human body target curves in measurement range.
Although the embodiments of the present invention has been shown and described above, it is to be understood that above-described embodiment is example
Property, it is not considered as limiting the invention, those skilled in the art within the scope of the invention can be to above-mentioned
Embodiment is changed, changes, replacing and modification.
Claims (10)
1. a kind of body curve's measurement method based on deep learning, which is characterized in that including:
Step 1:Obtain human body segmentation and skeleton point regression data library;
Step 2:Input user's human body image;
Step 3:Calibration is detected to the image of the input using the human body segmentation and skeleton point regression data library and is carried
It takes, user's body curve's measurement data is calculated.
2. a kind of body curve's measurement method based on deep learning as described in claim 1, it is characterised in that:The step
Human body segmentation is obtained in one and skeleton point regression data library specifically includes:Training stage and test phase.
3. a kind of body curve's measurement method based on deep learning as claimed in claim 2, it is characterised in that:The training
Stage includes:Human body picture is chosen as training sample;The human body picture is pre-processed;The pretreatment is obtained
Picture carries out CNN and RPN training, obtains target category score, and target frame, human body segmentation's region unit and skeleton point return mould
Type.
4. a kind of body curve's measurement method based on deep learning as claimed in claim 3, it is characterised in that:The step
The user's human body image inputted in two include the user's human body direct picture obtained using monocular cam or binocular camera and
User's human body side image.
5. a kind of body curve's measurement method based on deep learning as claimed in claim 4, it is characterised in that:The use
User's human body image that monocular cam obtains includes user's human body direct picture and user's human body side with object of reference
Face image.
6. a kind of body curve's measurement method based on deep learning as claimed in claim 5, it is characterised in that:The step
It includes being detected calibration to object of reference angle point and being clicked through to human body contour outline and bone to be detected calibration to input picture in three
Row extraction;
Wherein, include human body segmentation's area data being extracted, according to the human body segmentation region to human body contour outline and skeleton point extraction
Data extract human body segmentation's edges of regions coordinate, and by the edge coordinate, from the beginning the open-top beginning continuously deposits from top to bottom, from left to right
Storage, and bone point coordinates is obtained according to the skeleton point regression model.
7. a kind of body curve's measurement method based on deep learning as claimed in claim 6, it is characterised in that:When the ginseng
According to object be gridiron pattern when, calibration is detected to object of reference angle point and is specifically included:Gridiron pattern in picture is detected, according to what is detected
X-comers data, by each angular coordinate detected by from left to right, sequential storage from top to bottom calculates X and Y
The length that each pixel in direction indicates averages above-mentioned multiple length to obtain the physical length of each pixel expression.
8. a kind of body curve's measurement method based on deep learning as claimed in claim 7, it is characterised in that:Based on acquisition
User's human body direct picture human body front measurement data be calculated specifically include:
Human body contour outline is extracted since cephalad apex, the first concave point near zone in definition right side is averagely worth to D points, and definition is left
The concave point near zone of side first is averagely worth to C points, and it is that front neck is wide to define C points to D point X-direction differences, defines C points to the end
Portion vertex is head height;
It is A points that the upward distance l of left eye eyeball skeleton point, which is defined, with left side human body contour outline intersection point, defines the upward institute of left eye eyeball skeleton point
It is B points that distance l, which is stated, with right side human body contour outline intersection point, and it is that front is wide to define A points to B point X-direction differences;
A left shoulder skeleton point left side angle [alpha] and left side human body contour outline intersection point E points obliquely are defined, defines the right shoulder skeleton point right side obliquely
The angle [alpha] and right side human body contour outline intersection point F points, definition E points to F point X-direction differences are shoulder breadth;
The distance for defining left shoulder skeleton point to left elbow skeleton point is upper left arm lengths EK, defines right shoulder skeleton point to right elbow skeleton point
Distance be upper right arm lengths FM, define left elbow skeleton point to left wrist skeleton point distance be lower-left arm lengths KL, define right elbow
The distance of skeleton point to right wrist skeleton point is lower-left arm lengths MN, and it is (EK+FM)/2 to define upper arm lengths, defines lower arm lengths and is
(KL+MN)/2;
Define right and left shoulders bone central point with it is high for body at a distance from left and right hip joint dot center point, the definition E points right side obliquely one away from
From being G points at m, defining a F points left sides, the places the distance m are H points obliquely, define G points to H points X-direction distance be front chest breadth;
It is I points that the upward distance n of left hip joint, which is defined, with left side profile intersection point, defines right hip joint the distance n and the right side upwards
Side profile intersection point is J points, and it is wide for front waist to J points X-direction distance to define I points.
9. a kind of body curve's measurement method based on deep learning as claimed in claim 8, it is characterised in that:Based on acquisition
User's human body side image is calculated human body side measurement data and specifically includes:Extraction human body contour outline is opened from cephalad apex
Begin,
The first concave point near zone in definition left side is averagely worth to Q points, and definition Q points to cephalad apex Y-direction height are side
Head height;
It defines on the right side of Q points X-direction and intersection point is R points on the right side of human body contour outline, the distance for defining Q points and R point X-directions is side neck
It is wide;
It is O points that the upward distance l ' of eyes skeleton point, which is defined, with left side human body contour outline intersection point, and it is described upwards to define eyes skeleton point
Distance l ' is P points with right side human body contour outline intersection point, and it is that side is wide to define O points with P points X-direction distance;
It defines right and left shoulders bone central point and is side surface body height at a distance from left and right hip joint dot center point;
It is S points that right and left shoulders bone central point, which is defined, to next distance m ' and left side human body contour outline intersection point, is defined in right and left shoulders bone
The distance m ' and right side human body contour outline intersection point are T points to heart point downwards, and it is side chest breadth to define S points with T points X-direction distance;
It is U points that the upward distance n ' of left and right hip joint central point, which is defined, with left side human body contour outline intersection point, is defined in the hip joint of left and right
The distance n ' and right side human body contour outline intersection point are V points to heart point upwards, and it is that side waist is wide to define U points with V points X-direction distance.
10. a kind of body curve's measuring system based on deep learning, which is characterized in that including:
Human body segmentation and skeleton point regression data library acquisition module, for obtaining human body segmentation and skeleton point regression data library;
User's human body image acquisition module, for receiving user's human body image;
Body curve's computing module, for the user body curve measurement data to be calculated.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810651248.2A CN108564586A (en) | 2018-06-22 | 2018-06-22 | A kind of body curve's measurement method and system based on deep learning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810651248.2A CN108564586A (en) | 2018-06-22 | 2018-06-22 | A kind of body curve's measurement method and system based on deep learning |
Publications (1)
Publication Number | Publication Date |
---|---|
CN108564586A true CN108564586A (en) | 2018-09-21 |
Family
ID=63554350
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810651248.2A Pending CN108564586A (en) | 2018-06-22 | 2018-06-22 | A kind of body curve's measurement method and system based on deep learning |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108564586A (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109815816A (en) * | 2018-12-24 | 2019-05-28 | 山东山大鸥玛软件股份有限公司 | A kind of examinee examination hall abnormal behaviour analysis method based on deep learning |
CN109938737A (en) * | 2019-03-01 | 2019-06-28 | 苏州博慧智能科技有限公司 | A kind of human body body type measurement method and device based on deep learning critical point detection |
CN110111381A (en) * | 2019-03-13 | 2019-08-09 | 中山易裁剪网络科技有限公司 | A kind of long-range determining suit length system and its determining method |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105787439A (en) * | 2016-02-04 | 2016-07-20 | 广州新节奏智能科技有限公司 | Depth image human body joint positioning method based on convolution nerve network |
US20170243042A1 (en) * | 2011-02-04 | 2017-08-24 | Gannon Technologies Group, Llc | Systems and methods for biometric identification |
CN107492121A (en) * | 2017-07-03 | 2017-12-19 | 广州新节奏智能科技股份有限公司 | A kind of two-dimension human body bone independent positioning method of monocular depth video |
CN107886069A (en) * | 2017-11-10 | 2018-04-06 | 东北大学 | A kind of multiple target human body 2D gesture real-time detection systems and detection method |
-
2018
- 2018-06-22 CN CN201810651248.2A patent/CN108564586A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20170243042A1 (en) * | 2011-02-04 | 2017-08-24 | Gannon Technologies Group, Llc | Systems and methods for biometric identification |
CN105787439A (en) * | 2016-02-04 | 2016-07-20 | 广州新节奏智能科技有限公司 | Depth image human body joint positioning method based on convolution nerve network |
CN107492121A (en) * | 2017-07-03 | 2017-12-19 | 广州新节奏智能科技股份有限公司 | A kind of two-dimension human body bone independent positioning method of monocular depth video |
CN107886069A (en) * | 2017-11-10 | 2018-04-06 | 东北大学 | A kind of multiple target human body 2D gesture real-time detection systems and detection method |
Non-Patent Citations (1)
Title |
---|
杨冬梅: "基于双目视觉的人体参数获取方法的研究", 中国优秀博硕士学位论文全文数据库(硕士)工程科技辑 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109815816A (en) * | 2018-12-24 | 2019-05-28 | 山东山大鸥玛软件股份有限公司 | A kind of examinee examination hall abnormal behaviour analysis method based on deep learning |
CN109815816B (en) * | 2018-12-24 | 2023-02-03 | 山东山大鸥玛软件股份有限公司 | Deep learning-based examinee examination room abnormal behavior analysis method |
CN109938737A (en) * | 2019-03-01 | 2019-06-28 | 苏州博慧智能科技有限公司 | A kind of human body body type measurement method and device based on deep learning critical point detection |
CN110111381A (en) * | 2019-03-13 | 2019-08-09 | 中山易裁剪网络科技有限公司 | A kind of long-range determining suit length system and its determining method |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN101322589B (en) | Non-contact type human body measuring method for clothing design | |
CN106022213B (en) | A kind of human motion recognition method based on three-dimensional bone information | |
US20100312143A1 (en) | Human body measurement system and information provision method using the same | |
US7561726B2 (en) | Automated landmark extraction from three-dimensional whole body scanned data | |
US20050182341A1 (en) | Posture diagnosis equipment and program therefor | |
CN107041585A (en) | The measuring method of human dimension | |
Markiewicz et al. | 3D anthropometric algorithms for the estimation of measurements required for specialized garment design | |
CN105608737B (en) | A kind of human foot three-dimensional rebuilding method based on machine learning | |
CN108564586A (en) | A kind of body curve's measurement method and system based on deep learning | |
CN109938737A (en) | A kind of human body body type measurement method and device based on deep learning critical point detection | |
Chi et al. | Body scanning of dynamic posture | |
US11357423B2 (en) | Systems and methods to estimate human length | |
CN108629319A (en) | Image detecting method and system | |
CN110074788A (en) | A kind of body data acquisition methods and device based on machine learning | |
CN106952335A (en) | Set up the method and its system in manikin storehouse | |
CN114712769A (en) | Standing long jump intelligent distance measuring method and system based on computer vision | |
CN108363964A (en) | A kind of pretreated wrinkle of skin appraisal procedure and system | |
CN108324247A (en) | A kind of designated position wrinkle of skin appraisal procedure and system | |
CN105999655A (en) | A counting method and system for push-up tests | |
CN110477921B (en) | Height measurement method based on skeleton broken line Ridge regression | |
Chiu et al. | Automated body volume acquisitions from 3D structured-light scanning | |
CN103577800A (en) | Method for measuring human hand morphological parameters based on color images | |
CN108665471A (en) | A kind of human body back curve acquisition methods and system based on camera | |
JP2003049307A (en) | Clothing body size measuring method and measuring system | |
US20220198696A1 (en) | System for determining body measurement from images |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |