CN108851833A - Smart pier glass and its training method based on machine learning training algorithm - Google Patents
Smart pier glass and its training method based on machine learning training algorithm Download PDFInfo
- Publication number
- CN108851833A CN108851833A CN201810556928.6A CN201810556928A CN108851833A CN 108851833 A CN108851833 A CN 108851833A CN 201810556928 A CN201810556928 A CN 201810556928A CN 108851833 A CN108851833 A CN 108851833A
- Authority
- CN
- China
- Prior art keywords
- data
- microprocessor
- training
- face
- 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
-
- A—HUMAN NECESSITIES
- A47—FURNITURE; DOMESTIC ARTICLES OR APPLIANCES; COFFEE MILLS; SPICE MILLS; SUCTION CLEANERS IN GENERAL
- A47G—HOUSEHOLD OR TABLE EQUIPMENT
- A47G1/00—Mirrors; Picture frames or the like, e.g. provided with heating, lighting or ventilating means
- A47G1/02—Mirrors used as equipment
Landscapes
- Image Processing (AREA)
Abstract
The present invention provides a kind of smart pier glass and its training method based on machine learning training algorithm, belong to computer application technology, it include microprocessor, image collection module, RF proximity sensor switch, GPS positioning system, GPRS wireless transparent transmission module and remote server including pier glass ontology, touch LED display, hiding scale and intelligent measurement and control unit, intelligent measurement and control unit.This method generates confrontation algorithm for training network using machine learning and builds frame, collect throughout the year the wearing picture in the case of different weather as raw data set, weather data situation when being taken pictures using user is criterion, by generating network, differentiating network iteration repeatedly, judgement, rapidly, it accurately exports under current weather conditions, what user should dress meets the data of wearing of current weather, and output comparison picture in the display system.Equally, according to this algorithm, in the case of different weather may be implemented, the facial skin quality examining report of user is exported.
Description
Technical field
The invention belongs to computer application technologies, and in particular to a kind of intelligence based on machine learning training algorithm is worn
Clothing mirror and its training method.
Background technique
From in May, 2012, Japanese Seraku company is proposed in primary scientific and technological exhibition a based on android system
Intelligent mirror, hereafter many companies are devoted to the research of Intelligent mirror, and the function of Intelligent mirror is also various, application
Also it is not quite similar.
Although the sub- function of existing smart pier glass on the market is more, it is all based on the Intelligent mirror of android system,
Device is equivalent to the big tablet computer with touch function, and size is bigger, and cost is higher, and price is more expensive, therefore general
Property is poor.
Summary of the invention
In order to overcome the shortcomings of the prior art described above, the present invention provides a kind of based on machine learning training algorithm
Smart pier glass and measurement method.
To achieve the goals above, the present invention provides the following technical solutions:
Based on the smart pier glass of machine learning training algorithm, including pier glass ontology, touch LED display, hide
Scale and intelligent measurement and control unit, the intelligent measurement and control unit include microprocessor, image collection module, RF proximity sensor switch,
GPS positioning system, GPRS wireless transparent transmission module and remote server;
The pier glass ontology includes pedestal and the mirror surface that the base top is arranged in, the touch LED display
The upper right corner of the mirror surface is set, and the hiding scale is the electronic scale with wifi function, and the hiding scale is arranged described
On pan frontward end face, the remote server is arranged in the pedestal, the microprocessor, RF proximity sensor switch, GPS
After positioning system, GPRS wireless transparent transmission module are arranged at the touch LED display, described image obtains module setting and exists
The touch LED display surface, the mirror surface side wall are provided with USB interface, the USB interface and microprocessor electricity
Connection;
The touch LED display, image collection module, RF proximity sensor switch, GPS positioning system, GPRS without
Line transparent transmission module and USB interface are electrically connected with the microprocessor, the GPRS wireless transparent transmission module respectively with hiding scale and
Remote server is wirelessly connected;
The touch LED display, for inputting memo information, and output people facial skin quality examining report, wear
Wear skin care advisory information, user's weight and memo information, current weather conditions;
Described image obtains module, is sent to the micro process for obtaining the image information of user, and by image information
Device;
The GPS positioning system for positioning the specific geographic information of user's device therefor, and geographic information data is sent out
Give the microprocessor;
The RF proximity sensor switch, is sent to the microprocessor for detecting human body signal, and by human body signal;
The hiding scale passes through the GPRS wireless transparent transmission module for measuring human body weight data, and by weight data
It is sent to the microprocessor;
The microprocessor is for receiving the memo information, image information, geography information, human body signal and weight number
According to, and data interaction is carried out by the GPRS wireless transparent transmission module and the remote server, the microprocessor is also
For the facial skin quality examining report to the touch LED display sender, wearing skin care advisory information, user's weight
And memo information, current weather conditions;
The remote server, for storing the pattern data library of the corresponding dress of different weather, face difference skin quality
Pattern data library;Confrontation network G AN training algorithm is generated according to machine learning, data training is carried out, forms the facial skin quality of people
Examining report, wearing skin care advisory information, carry out data interaction by the GPRS wireless transparent transmission module and the microprocessor
Transmission.
Preferably, it is camera that described image, which obtains module,.
Preferably, the microprocessor is ARM microprocessor, model STM32F103;The RF proximity sensor switch
For 86 type TCL Regrand human induction switch.
Preferably, the height of the pier glass ontology is 220cm, width 80cm.
Another object of the present invention is to provide a kind of training sides of smart pier glass based on machine learning training algorithm
Method carries out source data collection and frame is built before training:The people for collecting somewhere are throughout the year in the case of different weather
Wearing picture be stored in the remote server as raw data set, raw data set format is referring to CelebA data set
Production is put up with PyCharm and generates network G and differentiation network D;
The training method includes the following steps:
Step 1 obtains training data:User stands in face of the pier glass ontology, at least with the pier glass ontology
The distance of 50cm is kept, described image obtains module photograph user and wears image data, and is sent to image data is worn
The microprocessor, the GPS positioning system obtain the physical data in user location, and physical data are sent to described
Microprocessor, it is described to wear image data and physical data composition wear input data, described in the microprocessor passes through again
GPRS wireless transparent transmission module will wear input data and be transmitted to the remote server;
Step 2, confrontation network parameter tuning:The remote server puts up the input data feeding of wearing received
Generation network G in be trained, obtain dress and generate data, then dress is generated into data and raw data set and is input to jointly
Distinguish that network D is trained;
Step 3 differentiates that dress generates whether data are true:Distinguish that network D judgement dress generates data and raw data set
When having difference, 2 re -training of return step;When distinguishing that network D judgement dress generates data and raw data set indifference, carry out
Step 4;
Step 4:Finally obtained dress generates image data after output training, and provide dress generate image data with
User wears the matching degree of image data;
Step 5, user face obtain module close to described image, and described image obtains module and individually shoots user's face
Portion's whole characteristic, and face's whole characteristic is sent to the microprocessor, the microprocessor passes through described again
Face's whole characteristic is transmitted to the remote server by GPRS wireless transparent transmission module, and the remote server will receive
Face's whole characteristic be sent into generate network G in be trained, obtain face generate data, then by face generate data with
The pattern data library of the face difference skin quality of remote server storage constitute face's input data be sent into distinguish network D into
Row repetition training, final discriminator reach balance, can not differentiate when face generates data and the data in pattern data library and stop instructing
Practice, exports user's face skin quality examining report;
Step 6:The remote server stores the output of the step 4 and step 5 as a result, and sending out the output result
Touch LED display is sent to be shown.
Preferably, the physical condition data include longitude, latitude, temperature, intensity of illumination and the purple in user location
Outside line intensity.
Preferably, user's face skin quality examining report, including face's skin quality characteristic information, qualitative user's skin quality, mention
There are problems, the cosmetics information for instructing care method and suggesting smearing for facial skin out.
Smart pier glass and its training method provided by the invention based on machine learning training algorithm has following
Beneficial effect:
(1) this smart pier glass can rapidly capture the face-image of people, remote server by image collection module
Using the algorithm of machine learning, the skin quality of people is fast and accurately determined, and combines intensity of illumination, uitraviolet intensity etc. comprehensive
Factor, and corresponding facial skin quality examining report is provided, precision is high, and time of measuring is short;
(2) this smart pier glass can capture rapidly the general image information of people by image collection module, including people's
Wearing, in conjunction with weather condition, remote server works as the day before yesterday by the way that whether the wearing that machine learning algorithm can export people in real time meets
Gas, and provide the information of suggestion wearing;
(3) upper right corner of mirror surface is arranged in the touch LED display of this smart pier glass, and area occupied is small, saves significantly
Cost is saved;
(4) this smart pier glass is based on machine learning algorithm, passes through dual training (input-training-output-repeatedly
Input), can the desired information of more accurate, rapid output user, more intimate measured body weight, memorandum prompting function
The practicability of this smart pier glass can more be added;Meanwhile the generation of machine learning fights algorithm for training network, itself can not
Disconnected self-teaching, self-perfection export better quality, identification capability is stronger, and speed is faster so that self performance be continuously improved
Image data.
Detailed description of the invention
Fig. 1 is the structural schematic diagram of the smart pier glass based on machine learning training algorithm of the embodiment of the present invention 1;
Fig. 2 is the structural block diagram of the smart pier glass based on machine learning training algorithm of the embodiment of the present invention 1;
Fig. 3 makes a living into confrontation network G AN flow chart;
Fig. 4 makes a living into confrontation network G AN circuit training flow chart.
Specific embodiment
With reference to the accompanying drawing, further description of the specific embodiments of the present invention.Following embodiment is only used for more
Technical solution of the present invention is clearly demonstrated, and not intended to limit the protection scope of the present invention.
Embodiment 1
A kind of smart pier glass based on machine learning training algorithm is present embodiments provided, specifically such as Fig. 1 and Fig. 2 institute
Show, including pier glass ontology, touch LED display 1, hiding scale 2 and intelligent measurement and control unit, intelligent measurement and control unit include micro- place
Device 3, image collection module 4, RF proximity sensor switch 5, GPS positioning system 6, GPRS wireless transparent transmission module 7 and distal end is managed to take
Business device 8;
Pier glass ontology includes pedestal 9 and the mirror surface 10 that 9 top of pedestal is arranged in, and touch LED display 1 is arranged in mirror
The upper right corner in face 10, hiding scale 2 is the electronic scale with wifi function, hides scale 2 and is arranged on 9 front end face of pedestal, when people stands
When on scale, the weight data of measurement is sent to by microprocessor 3 by GPRS wireless transparent transmission module 7, microprocessor 3 will count again
According to the cortex-A9 device control unit for being sent to control display screen, to be shown on touch LED display 1.Distal end takes
The setting of device 8 be engaged in pedestal 9, microprocessor 3, RF proximity sensor switch 5, GPS positioning system 6, GPRS wireless transparent transmission module 7
After being arranged at touch LED display 1, image collection module 4 is arranged in touch 1 surface of LED display, 10 side wall of mirror surface
It is provided with usb 11, usb 11 is electrically connected with microprocessor 3;
Touch LED display 1, image collection module 4, RF proximity sensor switch 5, GPS positioning system 6, GPRS without
Line transparent transmission module 7 and usb 11 are electrically connected with microprocessor 3, GPRS wireless transparent transmission module 7 respectively with hiding scale 2 and remote
Server 8 is held to be wirelessly connected;
Touch LED display 1, for inputting memo information, and facial skin quality examining report, the wearing shield of output people
Skin advisory information, user's weight and memo information, current weather conditions;
Image collection module 4 is sent to microprocessor 3 for obtaining the image information of user, and by image information;
GPS positioning system 6 for positioning the specific geographic information of user's device therefor, and geographic information data is sent
To microprocessor 3;
RF proximity sensor switch 5 is sent to microprocessor 3 for detecting human body signal, and by human body signal;
Scale 2 is hidden, is sent to for measuring human body weight data, and by weight data by GPRS wireless transparent transmission module 7
Microprocessor 3;
Microprocessor 3 passes through for receiving memo information, image information, geography information, human body signal and weight data
GPRS wireless transparent transmission module 7 and remote server 8 carry out data interaction, and microprocessor 3 is also used to show to touch LED
Shield facial skin quality examining report, wearing skin care advisory information, user's weight and the memo information of 1 sender, work as the day before yesterday
Gas situation;
Remote server 8, for storing the pattern data library of the corresponding dress of different weather, the pattern of face difference skin quality
Database;Confrontation network G AN training algorithm is generated according to machine learning, carries out data training, forms the facial skin quality detection of people
Report, wearing skin care advisory information, carry out data interaction by GPRS wireless transparent transmission module 7 and microprocessor 3.
In the present embodiment, image collection module 4 is camera;Microprocessor 3 is ARM microprocessor, model
STM32F103;RF proximity sensor switch 5 is 86 type TCL Regrand human induction switch.
In order to meet the demand of wearing the clothes, the height of pier glass ontology is 220cm, width 80cm, is suitble to public use.
Based on the above smart pier glass, the present embodiment additionally provides a kind of intelligence based on machine learning training algorithm and wears the clothes
The training method of mirror carries out source data collection and frame is built before training:The people for collecting somewhere are different throughout the year
Wearing picture under weather condition is stored in remote server 8 as raw data set, raw data set format reference
The production of CelebA data set is put up with PyCharm and generates network G and differentiation network D;
First to such as generating network G and differentiating that network D is illustrated, as shown in Figure 3 and Figure 4, it is set to touch LED
The image collection module 4 on 1 surface of display screen dresses situation for shooting user, which passes through GPRS data transparent transmission mould
Block 7 is transferred to remote server 8, and the weather data obtained with server 8 collectively forms input data, is input to generation network G
(Generator Network) is trained, and is generated network and is also referred to as generator, its main purpose is to generate to distinguish network D
It is thought as the data of truthful data, we are known as the data after training to generate data.
(confrontation network is being generated by the image data of generation data and server memory storage after generation network G training
In be known as initial data) be input to jointly distinguish network D (discriminator) be trained, distinguish network also referred to as distinguish
Device, its main purpose are to distinguish to generate whether data are truthful datas, and what it was carried out is two discriminant classifications.
Training method includes specific following steps:
Step 1 obtains training data:User stands in face of pier glass ontology, at least keeps 50cm with pier glass ontology
Distance, image collection module 4 shoots user and wears image data, and will wear image data and be sent to microprocessor 3,
GPS positioning system 6 obtains the physical data in user location, and physical data is sent to microprocessor 3, wears picture number
Wear input data according to physical data composition, microprocessor 3 again by GPRS wireless transparent transmission module 7 will wear input data biography
Transport to remote server 8;Here physical data includes the longitude in user location, latitude, temperature, intensity of illumination and ultraviolet
Line intensity;
Step 2, confrontation network parameter tuning:Remote server 8 wears what input data send what is received into was put up
It generates and is trained in network G, obtain dress and generate data, then dress generation data and raw data set are input to jointly and are distinguished
Other network D is trained;
Specially:It generates network G and corresponding picture is generated according to given condition C, it is raw then will to generate network G
At picture be sent into the conditional differentiation network D that puts up, differentiate network D for differentiating that picture is true picture or generation
The picture that network G generates;Wherein, condition C refers to label corresponding to image content, that is, remote server 8 obtains when taken a picture
Weather data;
Step 3 differentiates that dress generates whether data are true:Distinguish that network D judgement dress generates data and raw data set
When having difference, 2 re -training of return step;When distinguishing that network D judgement dress generates data and raw data set indifference, carry out
Step 4;
Specially:For truthful data, cost function is calculated using label=1 to train arbiter, for generator,
Cost function is calculated using label=0 to train arbiter, it is larger with truthful data gap that when original state generates data, distinguishes
Other device is easy to differentiate;By training, model modification is generated to the loss guidance whether really judgement obtains, gap is reduced;More than
Training is reciprocal to be carried out, and discriminator constantly enhances itself resolving ability, and generator exports more and more true output data to take advantage of
Deceive discriminator, ultimately generate data it is similar with truthful data when, discriminator can not identify true and false, that is, reach " Nash Equilibrium ";
Step 4:Finally obtained dress generates image data after output training, and provide dress generate image data with
User wears the matching degree of image data;
Specially:Image data after the multiple training of output, while filming apparatus original image initially shot is exported, and
The matching degree for providing two width picture wearable garments is exported by way of voice or text;
Step 5, user face are close to image collection module 4, and individually shooting user face is whole for image collection module 4
Characteristic, and face's whole characteristic is sent to microprocessor 3, microprocessor 3 passes through GPRS wireless transparent transmission module 7 again
Face's whole characteristic is transmitted to remote server 8, the face's whole characteristic received is sent by remote server 8
It generates and is trained in network G, obtain face and generate data, then face is generated into the face that data and remote server 8 store
The pattern data library of portion's difference skin quality constitutes face's input data and is sent into discrimination network D progress repetition training, and final discriminator reaches
To balance, deconditioning when face generates data and the data in pattern data library, output user's face skin quality inspection can not be differentiated
Observe and predict announcement;User's face skin quality examining report specifically includes face's skin quality characteristic information, qualitative user's skin quality, proposes face
There are problems, the cosmetics information for instructing care method and suggesting smearing for skin;User's skin quality includes dry skin, neutral skin
Matter, Combination skin quality, oily skin, sensitive skin, user's facial skin include such as whether pore is coarse, color there are problem
The symptoms such as spot, acne, skin quality be dark and gloomy description;
Step 6:The output of 8 storing step 4 of remote server and step 5 is as a result, and send output result to touch
LED display 1 is shown.
This smart pier glass fights algorithm for training network by the generation of machine learning, can rapidly export and work as the day before yesterday
In the case of gas, whether the wearing situation of user meets current weather, and exports corresponding suggestion wearing picture.Work as user
When close to mirror, photographic device captures human face image information and is simultaneously transferred to server, can be with by the training algorithm of machine learning
The facial skin quality examining report for exporting people, for skin quality and current weather, the advisory information of cosmetics should be smeared by exporting user,
The smart pier glass can also real-time display current weather conditions, room temperature, indoor humidity, user's weight and standby simultaneously
Forget the function that record is reminded.RF proximity sensor is installed in mirror, whether detection user contacts mirror, and user touches without practical
Mirror, mirror can also incude to obtain, and this device of intelligent opening is to show corresponding information.
The design captures human body and face information with the photographic device in mirror, is placed on engineering as input data
It is trained in the generation confrontation network algorithm of habit, the quality for capturing image determines the matter of trained number and output image
Amount and speed, it is higher to the image input quality requirement of face when being trained test to people's skin quality, the environment of surrounding and
The variation of light has a certain impact to image taking, is further improved.
Smart pier glass provided in this embodiment can be applicable to following occasion:
1. the intelligent apparatus can be placed at home, gets dressed when going out, stand before mirror, this device can export in real time
The matching degree of dress and current weather, by analyzing user face skin quality, it is proposed that the cosmetics that same day weather should be smeared.Out
The device, some great things that memorandum recording system can remind today or the following several days users and household that should do are opened in front of the door
?.
2. the design device can also be placed in beauty parlor, the customer to come in can apply this device to test skin skin
Matter can be surveyed using this device after beauty again after system provides suggestion according to progress cosmetic treatments are suggested
Examination carries out front and back comparison, not only can rapidly test out oneself skin of face skin quality characteristic and there are the problem of, but also can be with
Effect after observation treatment.
Embodiment described above is merely preferred embodiments of the present invention, and the scope of protection of the present invention is not limited to this,
Anyone skilled in the art within the technical scope of the present disclosure, the technical solution that can be become apparent to
Simple change or equivalence replacement, all belong to the scope of protection of the present invention.
Claims (7)
1. the smart pier glass based on machine learning training algorithm, which is characterized in that aobvious including pier glass ontology, touch LED
Display screen (1) hides scale (2) and intelligent measurement and control unit, and the intelligent measurement and control unit includes microprocessor (3), image collection module
(4), RF proximity sensor switch (5), GPS positioning system (6), GPRS wireless transparent transmission module (7) and remote server (8);
The pier glass ontology includes the mirror surface (10) of pedestal (9) and setting at the top of the pedestal (9), the touch LED
In the upper right corner of the mirror surface (10), the hiding scale (2) is the electronic scale with wifi function, described for display screen (1) setting
It hides scale (2) to be arranged on the pedestal (9) front end face, remote server (8) setting is described in the pedestal (9)
Microprocessor (3), RF proximity sensor switch (5), GPS positioning system (6), GPRS wireless transparent transmission module (7) are arranged at institute
After stating touch LED display (1), described image obtains module (4) and is arranged in touch LED display (1) surface, institute
It states mirror surface (10) side wall to be provided with USB interface (11), the USB interface (11) is electrically connected with the microprocessor (3);
The touch LED display (1), image collection module (4), RF proximity sensor switch (5), GPS positioning system
(6), GPRS wireless transparent transmission module (7) and USB interface (11) are electrically connected with the microprocessor (3), and the GPRS is wirelessly saturating
Transmission module (7) is wirelessly connected with hiding scale (2) and remote server (8) respectively;
The touch LED display (1), for inputting memo information, and facial skin quality examining report, the wearing of output people
Skin care advisory information, user's weight and memo information, current weather conditions;
Described image obtains module (4), is sent to the microprocessor for obtaining the image information of user, and by image information
(3);
The GPS positioning system (6), for positioning the specific geographic information of user's device therefor, and geographic information data is sent out
Give the microprocessor (3);
The RF proximity sensor switchs (5), for detecting human body close signal, and human body close signal is sent to described micro-
Processor (3);
The hiding scale (2) passes through the GPRS wireless transparent transmission module for measuring human body weight data, and by weight data
(7) microprocessor (3) are sent to;
The microprocessor (3) is for receiving the memo information, image information, geography information, human body close signal and weight
Data, and data interaction is carried out by the GPRS wireless transparent transmission module (7) and the remote server (8), it is described micro-
Processor (3) is also used to the facial skin quality examining report to touch LED display (1) sender, wearing skin care suggestion
Information, user's weight and memo information, current weather conditions;
The remote server (8), for storing the pattern data library of the corresponding dress of different weather, the figure of face difference skin quality
Sample database;Confrontation network G AN training algorithm is generated according to machine learning, carries out data training, forms the facial skin quality inspection of people
Announcement, wearing skin care advisory information are observed and predicted, data are carried out by the GPRS wireless transparent transmission module (7) and the microprocessor (3)
Alternating transmission.
2. the smart pier glass according to claim 1 based on machine learning training algorithm, which is characterized in that described image
Obtaining module (4) is camera.
3. the smart pier glass according to claim 1 based on machine learning training algorithm, which is characterized in that micro- place
Managing device (3) is ARM microprocessor, model STM32F103;The RF proximity sensor switch (5) is 86 type TCL Regrand people
Body-sensing inductive switch.
4. the smart pier glass according to claim 1 based on machine learning training algorithm, which is characterized in that described to wear the clothes
The height of mirror ontology is 220cm, width 80cm.
5. a kind of training method for the smart pier glass described in claim 1 based on machine learning training algorithm, special
Sign is, before training, carries out source data collection and frame is built:Collect people's different weather feelings throughout the year in somewhere
Wearing picture under condition is stored in the remote server (8) as raw data set, raw data set format reference
The production of CelebA data set is put up with PyCharm and generates network G and differentiation network D;
The training method includes the following steps:
Step 1 obtains training data:Described image obtains module (4) shooting user and wears image data, and will wear picture
Data are sent to the microprocessor (3), and the GPS positioning system (6) obtains the physical data in user location, and by object
Reason data are sent to the microprocessor (3), it is described wear image data and physical data and constitutes wear input data, it is described micro-
Processor (3) will wear input data by the GPRS wireless transparent transmission module (7) again and be transmitted to the remote server (8);
Step 2, confrontation network parameter tuning:The remote server (8) puts up the input data feeding of wearing received
Generation network G in be trained, obtain dress and generate data, then dress is generated into data and raw data set and is input to jointly
Distinguish that network D is trained;
Step 3 differentiates that dress generates whether data are true:Distinguish that network D judgement dress generates data and raw data set and has difference
When other, 2 re -training of return step;When distinguishing that network D judgement dress generates data and raw data set indifference, step is carried out
4;
Step 4:Finally obtained dress generates image data after output training, and provides dress and generate image data and use
Person wears the matching degree of image data;
Step 5, described image obtain module (4) individually shooting user face whole characteristic, and by face's whole feature
Data are sent to the microprocessor (3), and the microprocessor (3) is again by the GPRS wireless transparent transmission module (7) by face
Whole characteristics are transmitted to the remote server (8), face's whole characteristic that the remote server (8) will receive
It is trained according to being sent into generate in network G, obtains face and generate data, then face is generated into data and the remote server
(8) the pattern data library of the face difference skin quality stored constitutes face's input data and is sent into discrimination network D progress repetition training,
Final discriminator reaches balance, can not differentiate deconditioning when face generates data and the data in pattern data library, and output uses
Person's face skin quality examining report;
Step 6:The remote server (8) stores the output of the step 4 and step 5 as a result, and sending out the output result
Touch LED display (1) is sent to be shown.
6. training method according to claim 5, which is characterized in that the physical condition data include user location
Longitude, latitude, temperature, intensity of illumination and uitraviolet intensity.
7. training method according to claim 5, which is characterized in that user's face skin quality examining report, including face
Skin quality characteristic information, qualitative user's skin quality, proposing facial skin, there are problems, the makeup instructed care method and suggest smearing
Product information.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810556928.6A CN108851833A (en) | 2018-06-01 | 2018-06-01 | Smart pier glass and its training method based on machine learning training algorithm |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810556928.6A CN108851833A (en) | 2018-06-01 | 2018-06-01 | Smart pier glass and its training method based on machine learning training algorithm |
Publications (1)
Publication Number | Publication Date |
---|---|
CN108851833A true CN108851833A (en) | 2018-11-23 |
Family
ID=64336458
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810556928.6A Pending CN108851833A (en) | 2018-06-01 | 2018-06-01 | Smart pier glass and its training method based on machine learning training algorithm |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108851833A (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110276657A (en) * | 2019-06-10 | 2019-09-24 | 秒针信息技术有限公司 | Determination method, apparatus, storage medium and the electronic device of target object |
CN111210438A (en) * | 2019-12-26 | 2020-05-29 | 新绎健康科技有限公司 | Mirror |
CN111835807A (en) * | 2019-04-23 | 2020-10-27 | 青岛海尔洗衣机有限公司 | Control method of clothes treatment equipment |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102907970A (en) * | 2012-10-29 | 2013-02-06 | 吕林璐 | Multi-functional mirror |
CN105550777A (en) * | 2015-12-16 | 2016-05-04 | 美的集团股份有限公司 | Method and apparatus for recommending clothing match of user and dresser mirror |
CN107437073A (en) * | 2017-07-19 | 2017-12-05 | 竹间智能科技(上海)有限公司 | Face skin quality analysis method and system based on deep learning with generation confrontation networking |
CN107644093A (en) * | 2017-09-27 | 2018-01-30 | 上海斐讯数据通信技术有限公司 | Method and system are recommended in wearing the clothes for intelligent terminal |
CN107772931A (en) * | 2016-08-24 | 2018-03-09 | 南京乐朋电子科技有限公司 | A kind of multifunctional mirror based on intelligence system |
KR20180026133A (en) * | 2016-09-02 | 2018-03-12 | 주식회사 디젠 | Dressing table, System apparatus for recommendation of fashion information |
CN107853942A (en) * | 2016-09-22 | 2018-03-30 | 苏州润桐专利运营有限公司 | A kind of new type auto display time and the multifunctional mirror of weather |
-
2018
- 2018-06-01 CN CN201810556928.6A patent/CN108851833A/en active Pending
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102907970A (en) * | 2012-10-29 | 2013-02-06 | 吕林璐 | Multi-functional mirror |
CN105550777A (en) * | 2015-12-16 | 2016-05-04 | 美的集团股份有限公司 | Method and apparatus for recommending clothing match of user and dresser mirror |
CN107772931A (en) * | 2016-08-24 | 2018-03-09 | 南京乐朋电子科技有限公司 | A kind of multifunctional mirror based on intelligence system |
KR20180026133A (en) * | 2016-09-02 | 2018-03-12 | 주식회사 디젠 | Dressing table, System apparatus for recommendation of fashion information |
CN107853942A (en) * | 2016-09-22 | 2018-03-30 | 苏州润桐专利运营有限公司 | A kind of new type auto display time and the multifunctional mirror of weather |
CN107437073A (en) * | 2017-07-19 | 2017-12-05 | 竹间智能科技(上海)有限公司 | Face skin quality analysis method and system based on deep learning with generation confrontation networking |
CN107644093A (en) * | 2017-09-27 | 2018-01-30 | 上海斐讯数据通信技术有限公司 | Method and system are recommended in wearing the clothes for intelligent terminal |
Non-Patent Citations (1)
Title |
---|
张营营: "生成对抗网络模型综述", 《电子设计工程》 * |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111835807A (en) * | 2019-04-23 | 2020-10-27 | 青岛海尔洗衣机有限公司 | Control method of clothes treatment equipment |
CN110276657A (en) * | 2019-06-10 | 2019-09-24 | 秒针信息技术有限公司 | Determination method, apparatus, storage medium and the electronic device of target object |
CN111210438A (en) * | 2019-12-26 | 2020-05-29 | 新绎健康科技有限公司 | Mirror |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Gadekallu et al. | Identification of malnutrition and prediction of BMI from facial images using real-time image processing and machine learning | |
US10062163B2 (en) | Health information service system | |
CN108765708A (en) | A kind of unmanned automatic vending machine | |
CN107397530B (en) | Traditional Chinese medicine constitution and emotion distinguishing and conditioning system | |
US10098545B2 (en) | Lens information management system for surface condition measurement and analysis and information management method for surface condition measurement and analysis | |
CN108851833A (en) | Smart pier glass and its training method based on machine learning training algorithm | |
CN110264299A (en) | Clothes recommended method, device and computer equipment based on recognition of face | |
CN104586362A (en) | Mobile internet terminal-oriented skin or hair detection and care system | |
CN109035629A (en) | A kind of shopping settlement method and device based on open automatic vending machine | |
CN104809463B (en) | A kind of high-precision fire disaster flame detection method for converting dictionary learning based on intensive scale invariant feature | |
CN110036402A (en) | The data processing method of prediction for media content performance | |
CN111310692B (en) | Detection object management method, system, machine readable medium and equipment | |
TWI452998B (en) | System and method for establishing and analyzing skin parameters using digital image multi-area analysis | |
CN209029110U (en) | Chinese medicine facial diagnosis is health management system arranged | |
CN109934047A (en) | Face identification system and its face identification method based on deep learning | |
CN108710874A (en) | A kind of automatic vending machine based on image collecting device | |
CN106951448A (en) | A kind of personalization, which is worn, takes recommendation method and system | |
CN101458817A (en) | Color analysis system and method | |
CN109997199A (en) | Tuberculosis inspection method based on deep learning | |
CN111297337A (en) | Detection object judgment method, system, machine readable medium and equipment | |
CN113159876B (en) | Clothing collocation recommendation device, method and storage medium | |
CN108287161A (en) | A kind of Intelligent tea organoleptic evaluation method and system | |
CN208796351U (en) | A kind of unmanned automatic vending machine | |
CN112863027A (en) | Security place access & exit video image analysis processing system | |
CN209103314U (en) | A kind of automatic vending machine based on image collecting 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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20181123 |