CN110298393A - A kind of hair scalp health conditions detection method based on deep learning - Google Patents
A kind of hair scalp health conditions detection method based on deep learning Download PDFInfo
- Publication number
- CN110298393A CN110298393A CN201910517659.7A CN201910517659A CN110298393A CN 110298393 A CN110298393 A CN 110298393A CN 201910517659 A CN201910517659 A CN 201910517659A CN 110298393 A CN110298393 A CN 110298393A
- Authority
- CN
- China
- Prior art keywords
- hair
- model
- scalp
- picture
- detector
- 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
- 210000004209 hair Anatomy 0.000 title claims abstract description 167
- 210000004761 scalp Anatomy 0.000 title claims abstract description 77
- 238000001514 detection method Methods 0.000 title claims abstract description 49
- 230000036541 health Effects 0.000 title claims abstract description 30
- 238000013135 deep learning Methods 0.000 title claims abstract description 29
- 210000003780 hair follicle Anatomy 0.000 claims abstract description 70
- 208000001840 Dandruff Diseases 0.000 claims abstract description 29
- 206010061218 Inflammation Diseases 0.000 claims abstract description 28
- 230000004054 inflammatory process Effects 0.000 claims abstract description 28
- 230000000903 blocking effect Effects 0.000 claims abstract description 25
- 230000003862 health status Effects 0.000 claims abstract description 12
- 230000004069 differentiation Effects 0.000 claims abstract description 10
- 238000012549 training Methods 0.000 claims description 28
- 238000013527 convolutional neural network Methods 0.000 claims description 25
- 239000002775 capsule Substances 0.000 claims description 5
- 238000007689 inspection Methods 0.000 claims description 4
- 238000009472 formulation Methods 0.000 claims description 3
- 239000000203 mixture Substances 0.000 claims description 3
- 238000012545 processing Methods 0.000 claims description 3
- 230000001629 suppression Effects 0.000 claims description 3
- 238000012360 testing method Methods 0.000 abstract description 9
- 238000010586 diagram Methods 0.000 description 6
- 210000003128 head Anatomy 0.000 description 4
- 238000000034 method Methods 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 201000004384 Alopecia Diseases 0.000 description 1
- 206010000496 acne Diseases 0.000 description 1
- 231100000360 alopecia Toxicity 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 201000010099 disease Diseases 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000003646 hair health Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 230000033764 rhythmic process Effects 0.000 description 1
- 210000003491 skin Anatomy 0.000 description 1
- 238000003786 synthesis reaction Methods 0.000 description 1
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/0033—Features or image-related aspects of imaging apparatus classified in A61B5/00, e.g. for MRI, optical tomography or impedance tomography apparatus; arrangements of imaging apparatus in a room
- A61B5/004—Features or image-related aspects of imaging apparatus classified in A61B5/00, e.g. for MRI, optical tomography or impedance tomography apparatus; arrangements of imaging apparatus in a room adapted for image acquisition of a particular organ or body part
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/0059—Measuring for diagnostic purposes; Identification of persons using light, e.g. diagnosis by transillumination, diascopy, fluorescence
- A61B5/0077—Devices for viewing the surface of the body, e.g. camera, magnifying lens
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/44—Detecting, measuring or recording for evaluating the integumentary system, e.g. skin, hair or nails
- A61B5/448—Hair evaluation, e.g. for hair disorder diagnosis
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7235—Details of waveform analysis
- A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems
- A61B5/7267—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems involving training the classification device
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
Abstract
The hair scalp health conditions detection method based on deep learning that the invention discloses a kind of, comprising: picture to be detected is filtered, the picture comprising hair, scalp information is obtained;Picture comprising hair, scalp information is detected, hair root area information, dandruff quantity and inflammation quantity are obtained;Hair root portion area information is detected, number of hairs and hair follicle quantity is calculated;Blocked state differentiation is carried out to the hair follicle in hair root portion area information, and calculates the quantity of blocking hair follicle;Dandruff quantity, inflammation quantity, number of hairs, hair follicle quantity, blocking hair follicle quantity carry out comprehensive assessment, obtain the health status of scalp, hair.The present invention directly carries out the detection of hair scalp health conditions to the picture comprising hair, scalp information by 1 model of classifier, 1 model of detector, 2 model of detector, 2 model of classifier, reduce the complexity of detection, testing cost is also reduced simultaneously, there is stronger adaptability.
Description
Technical field
The invention belongs to human health detection technique field, especially a kind of hair scalp health shape based on deep learning
Condition detection method.
Background technique
With the development of the society, the accelerating rhythm of life, people's lives pressure is gradually increased, and hair problem shows one
The trend that kind grows in intensity, the various problems of alopecia, dandruff, inflammation etc. emerge one after another, and the age of hair problem crowd occur
Span section is also increasing.Because hair root is a kind of relatively small test object of target, present people are difficult benefit
The detection of the health status of scalp, hair is carried out with conventional equipment.
For present people in the health status detection for carrying out scalp, hair, the mode taken is usually that hair care shop please be special
The Senior Nurse of industry detects, and in the detection process Senior Nurse by large-scale detecting instrument to the healthy shape of scalp, hair
Condition is detected and analyzed, and the defect of this mode is expensive, and people cannot obtain scalp, the head of itself in real time
The health status of hair.
Summary of the invention
In view of the above technical problems, the present invention provides a kind of hair scalp health conditions detection side based on deep learning
Method, comprising:
1 model of calling classification device is filtered picture to be detected, obtains the picture comprising hair and/or scalp information;
It calls 1 model of detector to detect the picture comprising hair and/or scalp information, obtains hair root area
Information, dandruff quantity and inflammation quantity;
It calls 2 model of detector to detect hair root portion area information, number of hairs and hair follicle quantity is calculated;
2 model of calling classification device carries out blocked state differentiation to the hair follicle in hair root portion area information, and calculates blocking
The quantity of hair follicle;
Dandruff quantity, inflammation quantity, number of hairs, hair follicle quantity, blocking hair follicle quantity carry out comprehensive assessment, obtain
The health status of scalp, hair.
Optionally, picture to be detected is filtered in 1 model of calling classification device, obtains believing comprising hair and/or scalp
Before the picture of breath, further includes:
1 model of classifier, 1 model of detector, 2 model of detector, 2 model of classifier are trained respectively.
Optionally, 1 model of classifier, 1 model of detector, 2 model of detector, 2 model of classifier are trained respectively
Specifically:
Obtain training sample;
Construct depth convolutional neural networks;
Training sample is input in depth convolutional neural networks and is trained, 1 model of classifier, detector are respectively obtained
1 model, 2 model of detector, 2 model of classifier.
Optionally, training sample is input in depth convolutional neural networks and is trained specifically:
1 model of classifier, 1 model of detector, 2 mould of detector are carried out respectively using deep learning frame tensorflow
The training of type, 2 model of classifier.
Optionally, 1 model of calling classification device is filtered picture to be detected, obtains comprising hair and/or scalp information
Picture specifically:
Picture to be detected is adjusted to the size suitable for 1 mode input of classifier;
Picture to be detected adjusted is input in depth convolutional neural networks and is filtered, obtain comprising hair and/
Or the picture of scalp information.
Optionally, it calls 1 model of detector to detect the picture comprising hair and/or scalp information, obtains hair
Root area information, dandruff quantity and inflammation quantity specifically:
Picture to be detected is adjusted to the size suitable for 1 mode input of detector;
Picture to be detected adjusted is input in depth convolutional neural networks and is calculated, hair root area is obtained
Information, dandruff quantity and inflammation quantity.
Optionally, it calls 2 model of detector to detect hair root portion area information, number of hairs and hair is calculated
Capsule quantity specifically:
Calculate the centre coordinate of hair root area information;
The point centered on centre coordinate, on picture to be detected cut 200*300 range region as number of hairs with
The assessment area of hair follicle quantity;
Assessment area is adjusted to the size suitable for 2 mode input of detector;
Assessment area adjusted is input in depth convolutional neural networks and is calculated, number of hairs and hair follicle are obtained
Quantity.
Optionally, assessment area adjusted is input in depth convolutional neural networks and is calculated, obtain hair number
Amount and hair follicle quantity specifically:
Obtain the location information and classification information of hair root in assessment area;
Non-maxima suppression processing is carried out to location information and classification information, obtains hair follicle quantity, wherein hair follicle quantity is logical
It crosses detection block to be shown, the quantity of detection block is number of hairs multiplied by the summation of corresponding classification coefficient.
Optionally, 2 model of calling classification device carries out blocked state differentiation to the hair follicle in hair root portion area information, and counts
Calculate the quantity of blocking hair follicle specifically:
It will test frame to be adjusted to be suitable for the size of 2 mode input of classifier;
Detection block after adjustment is input in depth convolutional neural networks and is differentiated, and calculates the quantity of blocking hair follicle.
Optionally, dandruff quantity, inflammation quantity, number of hairs, hair follicle quantity, blocking hair follicle quantity carry out comprehensive comment
Estimate, obtain the health status of scalp, hair specifically:
According to the marking of formulation rule, dandruff quantity, inflammation quantity, number of hairs, hair follicle quantity, blocking hair follicle number
Amount, which assign, divides, and score weighted sum obtains the final score of picture to be detected, judges hair scalp health by final score
Situation.
The beneficial effects of the present invention are:
The hair scalp health conditions detection method based on deep learning that the invention discloses a kind of, comprising: calling classification
1 model of device is filtered picture to be detected, obtains the picture comprising hair and/or scalp information;Call 1 model pair of detector
Picture comprising hair and/or scalp information is detected, and hair root area information, dandruff quantity and inflammation number are obtained
Amount;It calls 2 model of detector to detect hair root portion area information, number of hairs and hair follicle quantity is calculated;It calls
2 model of classifier carries out blocked state differentiation to the hair follicle in hair root portion area information, and calculates the quantity of blocking hair follicle;It is right
Dandruff quantity, inflammation quantity, number of hairs, hair follicle quantity, blocking hair follicle quantity carry out comprehensive assessment, obtain scalp, hair
Health status.The present invention passes through direct using 1 model of classifier, 1 model of detector, 2 model of detector, 2 model of classifier
The detection that the related index parameter of hair scalp health conditions is carried out to the picture comprising hair, scalp information, reduces detection
Complexity, while also reducing testing cost, there is stronger adaptability.
Detailed description of the invention
It, below will be to specific in order to illustrate more clearly of the specific embodiment of the invention or technical solution in the prior art
Embodiment or attached drawing needed to be used in the description of the prior art be briefly described, it should be apparent that, it is described below
Attached drawing is some embodiments of the present invention, for those skilled in the art, without creative efforts,
It is also possible to obtain other drawings based on these drawings.
Fig. 1 is a kind of realization stream of the hair scalp health conditions detection method based on deep learning in the embodiment of the present invention
Cheng Tu;
Fig. 2 is picture schematic diagram to be detected in the embodiment of the present invention;
Fig. 3 is testing result schematic diagram in the embodiment of the present invention;
Fig. 4 is the implementation flow chart based on depth convolutional neural networks model training in the embodiment of the present invention;
Fig. 5 is the structural block diagram of depth convolutional neural networks in the embodiment of the present invention.
Specific embodiment
Technical solution of the present invention is clearly and completely described below in conjunction with attached drawing, it is clear that described implementation
Example is a part of the embodiment of the present invention, instead of all the embodiments.Based on the embodiments of the present invention, ordinary skill
Personnel's every other embodiment obtained without making creative work, shall fall within the protection scope of the present invention.
The hair scalp health conditions detection method based on deep learning that the present invention provides a kind of, to solve existing head
Sending out scalp health conditions detection method needs the Senior Nurse of profession could be to the healthy shape of scalp, hair by large-scale detecting instrument
Condition is detected and analyzed, and people are difficult with the problem of conventional equipment carries out the health status detection of scalp, hair.For this purpose,
The hair scalp health conditions detection method based on deep learning that the embodiment of the invention discloses a kind of:
1 model of calling classification device is filtered picture to be detected, obtains the picture comprising hair and/or scalp information;
It calls 1 model of detector to detect the picture comprising hair and/or scalp information, obtains hair root area information, head
Consider quantity and inflammation quantity to be worth doing;It calls 2 model of detector to detect hair root portion area information, number of hairs is calculated
With hair follicle quantity;2 model of calling classification device carries out blocked state differentiation to the hair follicle in hair root portion area information, and calculates stifled
Fill in the quantity of hair follicle;Dandruff quantity, inflammation quantity, number of hairs, hair follicle quantity, blocking hair follicle quantity carry out synthesis and comment
Estimate, obtains the health status of scalp, hair.It will be described in more detail below:
Referring to FIG. 1, for a kind of hair scalp health conditions detection method based on deep learning in the embodiment of the present invention
Implementation flow chart, should hair scalp health conditions detection method based on deep learning include:
S1,1 model of calling classification device are filtered picture to be detected, obtain the figure comprising hair and/or scalp information
Piece, referring to FIG. 2, Fig. 2 is picture schematic diagram to be detected in the embodiment of the present invention.
Picture to be detected is adjusted to be suitble to the input size of 1 model of classifier, 1 model of calling classification device is to figure to be detected
Piece carries out classification judgement: if it is determined that the picture comprising hair and/or scalp information, operating into next step.
It should be noted that if it is determined that the picture not comprising hair and/or scalp information, then directly export picture not
Comprising hair and/or scalp information as a result, being no longer further processed.
S2 calls 1 model of detector to detect the picture comprising hair and/or scalp information, obtains hair root
Area information, dandruff quantity and inflammation quantity.
For being determined as the picture comprising hair and/or scalp information, by the picture tune comprising hair and/or scalp information
It is whole to call 1 model of detector to be detected to the input size for being suitble to 1 model of detector, obtain in picture the quantity of inflammation and
The quantity of dandruff and the information of hair root area.
It should be noted that if the information of hair root area is not detected in picture to be detected, it is regular according to marking,
The assessment marking of two dimensions directly is carried out to the dandruff quantity, the inflammation quantity that detect, is obtained after score weighted sum to be detected
Picture final score, is no longer further processed.
S3 calls 2 model of detector to detect hair root portion area information, number of hairs and hair follicle number is calculated
Amount.
For the picture with hair root area information, the centre coordinate of the hair root area information is calculated, and with
Point centered on centre coordinate, assessment area of the crop width 200*300 as number of hairs and hair follicle quantity, assessment area is adjusted
To the size for being suitable for 2 mode input of detector, 2 model of detector is called to carry out number of hairs and hair to the assessment area
The detection of capsule quantity obtains the location information and classification information of hair root in assessment area, to location information and classification information
Non-maxima suppression processing is carried out, hair follicle quantity is obtained, wherein the hair follicle quantity shown by detection block, the inspection
The quantity of survey frame is number of hairs multiplied by the summation of corresponding classification coefficient.
It should be noted that the object type based on mark, an object box represents a hair follicle, then test object frame number
Amount is hair follicle quantity;Based on the number of hairs in each object type follicle group, by object box quantity of all categories multiplied by correspondence
The summation of classification coefficient is number of hairs, referring to FIG. 3, Fig. 3 is testing result schematic diagram in the embodiment of the present invention, 1 in Fig. 3
It is object box for assessment area, 2.
S4,2 model of calling classification device carry out blocked state differentiation to the hair follicle in hair root portion area information, and calculate stifled
Fill in the quantity of hair follicle.
The detection block detected in third step is adjusted to the input size suitable for 2 model of classifier, calling classification device 2
The blocked state of hair root is differentiated, to obtain the quantity for blocking hair follicle in assessment area ROI.
S5, dandruff quantity, inflammation quantity, number of hairs, hair follicle quantity, blocking hair follicle quantity carry out comprehensive assessment,
Obtain the health status of scalp, hair.
According to the marking of formulation rule, dandruff quantity, inflammation quantity, number of hairs, hair follicle quantity, blocking hair follicle number
Amount, which assign, divides, and score weighted sum obtains the final score of the picture to be detected, judges hair by the final score
Scalp health conditions.
It should be noted that according to testing result, each number of dimensions segmented is assigned and is divided, for example, dandruff quantity,
Inflammation quantity, blocking hair follicle quantity can be by following marking rules: quantity 0 is denoted as 100 points, and quantity 1 to 2 is denoted as 85 points, quantity 3
70 points are denoted as to 4, quantity 5 to 6 is denoted as 55 points, and 7 or more are denoted as 40 points;Number of hairs, hair follicle quantity are by following marking rule: number
Amount 0 is denoted as 40 points, and quantity 1 to 5 is denoted as 55 points, and quantity 5 to 10 is denoted as 70 points, and quantity 10 to 15 is denoted as 85 points, and 15 or more are denoted as
100 points.
It should be noted that the score of each dimension is weighted summation and is obtained most according to the weight coefficient of each dimension
Whole hundred-mark system score, the weight coefficient of each dimension are as follows: 1/ number of dimensions.For example, it is assumed that be 5 dimensions, the then power of each dimension
Weight coefficient is 1/5.
The optional embodiment of the embodiment of the present invention is described in detail in conjunction with attached drawing above, still, the embodiment of the present invention is simultaneously
The detail being not limited in above embodiment can be to of the invention real in the range of the technology design of the embodiment of the present invention
The technical solution for applying example carries out a variety of simple variants, these simple variants belong to the protection scope of the embodiment of the present invention.
Fig. 4 and Fig. 5 are please referred to, Fig. 4 is the realization based on depth convolutional neural networks model training in the embodiment of the present invention
Flow chart, Fig. 5 are the structural block diagram of depth convolutional neural networks in the embodiment of the present invention.
Picture to be detected is filtered in 1 model of step S1 calling classification device, is obtained comprising hair and/or scalp information
Picture before, further includes:
1 model of classifier, 1 model of detector, 2 model of detector, 2 model of classifier are trained respectively.
Wherein, tool is trained to 1 model of classifier, 1 model of detector, 2 model of detector, 2 model of classifier respectively
Body is as follows:
S10 obtains training sample.
S20 constructs depth convolutional neural networks.
Training sample is input in depth convolutional neural networks and is trained by S30, respectively obtains 1 model of classifier, inspection
Survey 1 model of device, 2 model of detector, 2 model of classifier.
Again, training sample is input in depth convolutional neural networks and is trained specifically:
1 model of classifier, 1 model of detector, 2 mould of detector are carried out respectively using deep learning frame tensorflow
The training of type, 2 model of classifier.
It should be noted that the training to 1 model of classifier specifically:
Training dataset of several pictures as 1 model of classifier is acquired, depth convolutional neural networks is built and extracts training
The feature of all pictures in collecting carries out 1 model training of classifier using deep learning frame tensorflow, to be trained
Whether good 1 model of classifier can be distinguished in image to be detected after 1 model training of classifier is good comprising hair and/or scalp
Information, according to differentiation as a result, the picture not comprising hair and/or scalp information can be filtered out.
It should be noted that the training to 1 model of detector specifically:
Sample mark is carried out using marking software LabelImg to the collected picture comprising hair and/or scalp information
Note, mark classification includes: ROI (hair root area information), and Lump_of_dandruff (dandruff), Pimple (inflammation) take
It builds depth convolutional neural networks mark object is positioned and identified, be detected using deep learning frame tensorflow
1 model training of device, to obtain trained 1 model of detector, 1 model of detector is able to detect that in picture to be detected scorching
Disease, dandruff and hair root area information.
It should be noted that the training to 2 model of detector specifically:
The hair root area information of 1 model of detector mark is saved, and is carried out again using marking software LabelImg
Sample mark, marking classification again includes: Hair1 (1), Hair1_1 (intersection), Hair2 (2), Hair3 (3),
Hair4 (4), Hair5 (the mixed and disorderly block of half picture height), Hair8 (the mixed and disorderly block of picture height), build depth convolution again
Neural network is positioned and is identified to mark object again, carries out 2 mould of detector using deep learning frame tensorflow
Type training, to obtain trained 2 model of detector, 2 model of detector is able to detect that the different heads in picture to be detected
Root of hair portion state.In addition, the object type based on mark, a detection block represents a hair follicle, therefore the detection block number detected
Amount is hair follicle quantity, based on the number of hairs in each object type follicle group, by detection block quantity of all categories multiplied by correspondence
The summation of classification coefficient is number of hairs, to obtain the calculation amount of number of hairs and hair follicle quantity.
It should be noted that the training to 2 model of classifier specifically:
The detection block that 2 model inspection of detector arrives is saved to roll up as the training dataset of 2 model of classifier using depth
Product neural network carries out feature extraction to the training dataset of 2 model of classifier, and accesses softmax layers of 2 category classification of progress:
The non-blocking picture of 0-, 1- block picture, and the training of disaggregated model is then carried out using deep learning frame tensorflow, thus
Trained 2 model of classifier is obtained, 2 model of classifier can distinguish the blocking of hair root and non-plugged condition.
In conclusion the invention discloses a kind of hair scalp health conditions detection method based on deep learning, comprising:
1 model of calling classification device is filtered picture to be detected, obtains the picture comprising hair and/or scalp information;Call detection
1 model of device detects the picture comprising hair and/or scalp information, obtain hair root area information, dandruff quantity with
And inflammation quantity;It calls 2 model of detector to detect hair root portion area information, number of hairs and hair follicle number is calculated
Amount;2 model of calling classification device carries out blocked state differentiation to the hair follicle in hair root portion area information, and calculates blocking hair follicle
Quantity;Dandruff quantity, inflammation quantity, number of hairs, hair follicle quantity, blocking hair follicle quantity carry out comprehensive assessment, obtain to the end
The health status of skin, hair.The present invention is by utilizing 1 model of classifier, 1 model of detector, 2 model of detector, 2 mould of classifier
Type directly carries out the detection of the related index parameter of hair scalp health conditions to the picture comprising hair, scalp information, reduces
The complexity of detection, while testing cost is also reduced, there is stronger adaptability.
What has been described above is only an embodiment of the present invention, and the common sense such as well known specific structure and characteristic are not made herein in scheme
Excessive description.It, without departing from the structure of the invention, can be with it should be pointed out that for those skilled in the art
Make several modifications and improvements.These also should be considered as protection scope of the present invention, these all will not influence what the present invention was implemented
Effect and patent practicability.The scope of protection required by this application should be based on the content of the claims, in specification
The records such as specific embodiment can be used for explaining the content of claim.
Claims (10)
1. a kind of hair scalp health conditions detection method based on deep learning characterized by comprising
1 model of calling classification device is filtered picture to be detected, obtains the picture comprising hair and/or scalp information;
It calls 1 model of detector to detect the picture comprising hair and/or scalp information, obtains hair root area
Information, dandruff quantity and inflammation quantity;
It calls 2 model of detector to detect hair root area information, number of hairs and hair follicle quantity is calculated;
2 model of calling classification device carries out blocked state differentiation to the hair follicle in the area information of the hair root, and calculates blocking
The quantity of hair follicle;
To the dandruff quantity, the inflammation quantity, the number of hairs, the hair follicle quantity, the blocking hair follicle quantity
Comprehensive assessment is carried out, the health status of scalp, hair is obtained.
2. the hair scalp health conditions detection method according to claim 1 based on deep learning, which is characterized in that
1 model of calling classification device is filtered picture to be detected, before obtaining the picture comprising hair and/or scalp information, also wraps
It includes:
1 model of classifier, 1 model of the detector, 2 model of the detector, 2 model of the classifier are carried out respectively
Training.
3. the hair scalp health conditions detection method according to claim 2 based on deep learning, which is characterized in that institute
It states and 1 model of classifier, 1 model of the detector, 2 model of the detector, 2 model of the classifier is instructed respectively
Practice specifically:
Obtain training sample;
Construct depth convolutional neural networks;
The training sample is input in the depth convolutional neural networks and is trained, 1 mould of classifier is respectively obtained
Type, 1 model of the detector, 2 model of the detector, 2 model of the classifier.
4. the hair scalp health conditions detection method according to claim 3 based on deep learning, which is characterized in that institute
It states for the training sample to be input in the depth convolutional neural networks and be trained specifically:
1 model of classifier, 1 model of the detector, the inspection are carried out respectively using deep learning frame tensorflow
Survey the training of 2 model of device, 2 model of the classifier.
5. the hair scalp health conditions detection method according to claim 1 based on deep learning, which is characterized in that institute
It states 1 model of calling classification device to be filtered picture to be detected, obtains the picture comprising hair and/or scalp information specifically:
The picture to be detected is adjusted to the size suitable for 1 mode input of classifier;
The picture to be detected adjusted is input in the depth convolutional neural networks and is filtered, is obtained comprising hair
And/or the picture of scalp information.
6. the hair scalp health conditions detection method according to claim 1 based on deep learning, which is characterized in that institute
It states and 1 model of detector is called to detect the picture comprising hair and/or scalp information, obtain hair root area letter
Breath, dandruff quantity and inflammation quantity specifically:
The picture to be detected is adjusted to the size suitable for 1 mode input of detector;
The picture to be detected adjusted is input in the depth convolutional neural networks and is calculated, hair root is obtained
Area information, dandruff quantity and inflammation quantity.
7. the hair scalp health conditions detection method according to claim 1 based on deep learning, which is characterized in that institute
It states and 2 model of detector is called to detect hair root area information, number of hairs and hair follicle quantity tool is calculated
Body are as follows:
Calculate the centre coordinate of hair root area information;
The point centered on the centre coordinate cuts the region of 200*300 range as hair number on the picture to be detected
The assessment area of amount and hair follicle quantity;
The assessment area is adjusted to the size suitable for 2 mode input of detector;
The assessment area adjusted is input in the depth convolutional neural networks and is calculated, obtain number of hairs and
Hair follicle quantity.
8. the hair scalp health conditions detection method according to claim 7 based on deep learning, which is characterized in that institute
It states for the assessment area adjusted to be input in the depth convolutional neural networks and calculate, obtain number of hairs and hair
Capsule quantity specifically:
Obtain the location information and classification information of hair root in the assessment area;
Non-maxima suppression processing is carried out to the location information and the classification information, obtains hair follicle quantity, wherein the hair
Capsule quantity is shown that the quantity of the detection block is number of hairs multiplied by the summation of corresponding classification coefficient by detection block.
9. the hair scalp health conditions detection method according to claim 1 based on deep learning, which is characterized in that institute
It states 2 model of calling classification device and blocked state differentiation is carried out to the hair follicle in the area information of the hair root, and calculate blocking hair
The quantity of capsule specifically:
The detection block is adjusted to the size suitable for 2 mode input of classifier;
Detection block described after adjustment is input in the depth convolutional neural networks and is differentiated, and calculates the number of blocking hair follicle
Amount.
10. the hair scalp health conditions detection method according to claim 1 based on deep learning, which is characterized in that
The dandruff quantity, the inflammation quantity, the number of hairs, the hair follicle quantity, the blocking hair follicle quantity are carried out
Comprehensive assessment obtains the health status of scalp, hair specifically:
According to the marking of formulation rule, to the dandruff quantity, the inflammation quantity, the number of hairs, the hair follicle number
Amount, the blocking hair follicle quantity, which assign, divides, and score weighted sum obtains the final score of the picture to be detected, by described
Final score judges hair scalp health conditions.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910517659.7A CN110298393A (en) | 2019-06-14 | 2019-06-14 | A kind of hair scalp health conditions detection method based on deep learning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910517659.7A CN110298393A (en) | 2019-06-14 | 2019-06-14 | A kind of hair scalp health conditions detection method based on deep learning |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110298393A true CN110298393A (en) | 2019-10-01 |
Family
ID=68028037
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910517659.7A Pending CN110298393A (en) | 2019-06-14 | 2019-06-14 | A kind of hair scalp health conditions detection method based on deep learning |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110298393A (en) |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111275703A (en) * | 2020-02-27 | 2020-06-12 | 腾讯科技(深圳)有限公司 | Image detection method, image detection device, computer equipment and storage medium |
CN111428655A (en) * | 2020-03-27 | 2020-07-17 | 厦门大学 | Scalp detection method based on deep learning |
CN111489823A (en) * | 2020-04-08 | 2020-08-04 | Oppo广东移动通信有限公司 | Hair health condition evaluation method and device, mobile terminal and storage medium |
CN112084965A (en) * | 2020-09-11 | 2020-12-15 | 义乌市悦美科技有限公司 | Scalp hair detection device and system |
CN113627425A (en) * | 2021-07-16 | 2021-11-09 | 汕头大学 | Hair follicle identification and extraction method and system based on neural network model |
WO2022000238A1 (en) * | 2020-06-30 | 2022-01-06 | Henkel Ag & Co. Kgaa | Method for determining a scalp type of a user |
CN114758748A (en) * | 2022-04-07 | 2022-07-15 | 南京新生医疗科技有限公司 | Intelligent analysis processing method and system for hair planting environment |
TWI823143B (en) * | 2021-09-13 | 2023-11-21 | 美科實業股份有限公司 | Intelligent dandruff detection system and method |
Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103649750A (en) * | 2011-07-12 | 2014-03-19 | 宝洁公司 | Method for assessing condition of skin and/or scalp |
CN104299011A (en) * | 2014-10-13 | 2015-01-21 | 吴亮 | Skin type and skin problem identification and detection method based on facial image identification |
CN106469302A (en) * | 2016-09-07 | 2017-03-01 | 成都知识视觉科技有限公司 | A kind of face skin quality detection method based on artificial neural network |
CN107169960A (en) * | 2017-05-15 | 2017-09-15 | 精诚工坊电子集成技术(北京)有限公司 | A kind of skin surface pore size detection method based on color of image space |
CN107679507A (en) * | 2017-10-17 | 2018-02-09 | 北京大学第三医院 | Facial pores detecting system and method |
CN108280397A (en) * | 2017-12-25 | 2018-07-13 | 西安电子科技大学 | Human body image hair detection method based on depth convolutional neural networks |
CN108810050A (en) * | 2017-05-03 | 2018-11-13 | 杭州美界科技有限公司 | A kind of beauty information supplying system based on artificial intelligence |
CN109359527A (en) * | 2018-09-11 | 2019-02-19 | 杭州格像科技有限公司 | Hair zones extracting method and system neural network based |
CN208573183U (en) * | 2018-02-09 | 2019-03-05 | 武汉技兴科技有限公司 | A kind of human hair information collecting device and intelligent clipping device |
CN109512384A (en) * | 2017-09-18 | 2019-03-26 | 南台学校财团法人南台科技大学 | Scalp detection device |
-
2019
- 2019-06-14 CN CN201910517659.7A patent/CN110298393A/en active Pending
Patent Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103649750A (en) * | 2011-07-12 | 2014-03-19 | 宝洁公司 | Method for assessing condition of skin and/or scalp |
CN104299011A (en) * | 2014-10-13 | 2015-01-21 | 吴亮 | Skin type and skin problem identification and detection method based on facial image identification |
CN106469302A (en) * | 2016-09-07 | 2017-03-01 | 成都知识视觉科技有限公司 | A kind of face skin quality detection method based on artificial neural network |
CN108810050A (en) * | 2017-05-03 | 2018-11-13 | 杭州美界科技有限公司 | A kind of beauty information supplying system based on artificial intelligence |
CN107169960A (en) * | 2017-05-15 | 2017-09-15 | 精诚工坊电子集成技术(北京)有限公司 | A kind of skin surface pore size detection method based on color of image space |
CN109512384A (en) * | 2017-09-18 | 2019-03-26 | 南台学校财团法人南台科技大学 | Scalp detection device |
CN107679507A (en) * | 2017-10-17 | 2018-02-09 | 北京大学第三医院 | Facial pores detecting system and method |
CN108280397A (en) * | 2017-12-25 | 2018-07-13 | 西安电子科技大学 | Human body image hair detection method based on depth convolutional neural networks |
CN208573183U (en) * | 2018-02-09 | 2019-03-05 | 武汉技兴科技有限公司 | A kind of human hair information collecting device and intelligent clipping device |
CN109359527A (en) * | 2018-09-11 | 2019-02-19 | 杭州格像科技有限公司 | Hair zones extracting method and system neural network based |
Non-Patent Citations (1)
Title |
---|
WEN-SHIUNG HUANG ET AL: "A Cloud-based Intelligent Skin and Scalp Analysis System", 《2018 IEEE VISUAL COMMUNICATIONS AND IMAGE PROCESSING (VCIP)》 * |
Cited By (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111275703A (en) * | 2020-02-27 | 2020-06-12 | 腾讯科技(深圳)有限公司 | Image detection method, image detection device, computer equipment and storage medium |
CN111275703B (en) * | 2020-02-27 | 2023-10-27 | 腾讯科技(深圳)有限公司 | Image detection method, device, computer equipment and storage medium |
CN111428655A (en) * | 2020-03-27 | 2020-07-17 | 厦门大学 | Scalp detection method based on deep learning |
CN111489823A (en) * | 2020-04-08 | 2020-08-04 | Oppo广东移动通信有限公司 | Hair health condition evaluation method and device, mobile terminal and storage medium |
WO2022000238A1 (en) * | 2020-06-30 | 2022-01-06 | Henkel Ag & Co. Kgaa | Method for determining a scalp type of a user |
CN112084965A (en) * | 2020-09-11 | 2020-12-15 | 义乌市悦美科技有限公司 | Scalp hair detection device and system |
CN113627425A (en) * | 2021-07-16 | 2021-11-09 | 汕头大学 | Hair follicle identification and extraction method and system based on neural network model |
CN113627425B (en) * | 2021-07-16 | 2023-09-05 | 汕头大学 | Hair follicle identification and extraction method and system based on neural network model |
TWI823143B (en) * | 2021-09-13 | 2023-11-21 | 美科實業股份有限公司 | Intelligent dandruff detection system and method |
CN114758748A (en) * | 2022-04-07 | 2022-07-15 | 南京新生医疗科技有限公司 | Intelligent analysis processing method and system for hair planting environment |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110298393A (en) | A kind of hair scalp health conditions detection method based on deep learning | |
JP4401079B2 (en) | Subject behavior analysis | |
CN105022929B (en) | A kind of cognition accuracy analysis method of personal traits value test | |
CN106022378B (en) | Sitting posture judgment method and based on camera and pressure sensor cervical spondylosis identifying system | |
CN112370018B (en) | Computer application software for predicting difficult airway and airway management data system | |
CN109389041A (en) | A kind of fall detection method based on joint point feature | |
KR20130024065A (en) | Apparatus and method for detecting complexion, apparatus and method for determinig health using complexion, apparatus and method for generating health sort function | |
JP7254868B2 (en) | Skin strain measurement method | |
TWI430776B (en) | Smart video skin test system and method of the same | |
CN113197558B (en) | Heart rate and respiratory rate detection method and system and computer storage medium | |
CN114565957A (en) | Consciousness assessment method and system based on micro expression recognition | |
EP3529764A1 (en) | Device for determining features of a person | |
CN115916045A (en) | Method for determining the scalp type of a user | |
CN111803031A (en) | Non-contact type drug addict relapse monitoring method and system | |
CN116386845A (en) | Schizophrenia diagnosis system based on convolutional neural network and facial dynamic video | |
CN116343302A (en) | Micro-expression classification and identification system based on machine vision | |
CN116115239A (en) | Embarrassing working gesture recognition method for construction workers based on multi-mode data fusion | |
CN113360838B (en) | Nozzle casting monitoring method and system based on image processing | |
CN114742090A (en) | Cockpit man-machine interaction system based on mental fatigue monitoring | |
CN112288735A (en) | Method for automatically detecting dental fracture by utilizing tooth cone beam CT (computed tomography) image based on neural network | |
JP7112080B2 (en) | Sway analysis device and method | |
CN113570545A (en) | Visual identification pain grading assessment method | |
CN106725364B (en) | Controller fatigue detection method and system based on probability statistical method | |
CN110349587A (en) | Target individual is snored differentiating method under a kind of two people's scenes | |
CN110175989A (en) | Video data processing method and its device |
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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20191001 |
|
RJ01 | Rejection of invention patent application after publication |