CN106993851A - A kind of shoe tree parameter automatic prediction method and prediction meanss based on shoes image and foot type image - Google Patents
A kind of shoe tree parameter automatic prediction method and prediction meanss based on shoes image and foot type image Download PDFInfo
- Publication number
- CN106993851A CN106993851A CN201710351500.3A CN201710351500A CN106993851A CN 106993851 A CN106993851 A CN 106993851A CN 201710351500 A CN201710351500 A CN 201710351500A CN 106993851 A CN106993851 A CN 106993851A
- Authority
- CN
- China
- Prior art keywords
- image
- shoes
- foot type
- shoe tree
- tree parameter
- 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.)
- Granted
Links
Classifications
-
- A—HUMAN NECESSITIES
- A43—FOOTWEAR
- A43D—MACHINES, TOOLS, EQUIPMENT OR METHODS FOR MANUFACTURING OR REPAIRING FOOTWEAR
- A43D1/00—Foot or last measuring devices; Measuring devices for shoe parts
- A43D1/04—Last-measuring devices
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/049—Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Molecular Biology (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Footwear And Its Accessory, Manufacturing Method And Apparatuses (AREA)
- Image Analysis (AREA)
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
Abstract
The present invention relates to a kind of shoe tree parameter automatic prediction method and prediction meanss based on shoes image and foot type image, comprise the following steps:(S1) collection shoes image and foot type image;(S2) training production confrontation network;(S3) existing shoes view data and foot type view data are expanded;(S4) shoes image and corresponding foot type image are combined into six channel images, the shoe tree parameter model based on shoes image and foot type image is trained using recurrent neural network;(S5) shoe tree parameter model automatic Prediction shoe tree parameter is utilized.Shoe tree parameter automatic prediction method and prediction meanss of the invention based on shoes image and foot type image, can be automatic to expand shoes image and foot type view data, automatic Prediction shoe tree parameter by gathering user's shoes image and foot type image.
Description
Technical field
It is more particularly to a kind of to be based on shoes image the present invention relates to a kind of shoe tree parameter automatic prediction method and prediction meanss
With the shoe tree parameter automatic prediction method and prediction meanss of foot type image.
Background technology
Shoe tree is the core of shoes production technology, is also the key factor for restricting shoes production.Conventional shoe tree measurement, is needed
Want experienced skilled worker to carry out complicated measurement conversion, can just obtain shoe tree parameter, not only waste time and energy, also result in enterprise
Productivity limited by technical staff's human efficiency remained high so that customizing shoes service prices, efficiency can not be improved, into
For the key factor restricted market development.The present invention provides one kind according to shoes image and foot type image automatic Prediction shoe tree
The method and apparatus of parameter, consumer can require to shoot several shoes photos and foot type photograph by using mobile phone according to certain
Piece, just can fast and accurately obtain shoe tree parameter, and then customized suitable shoes, with very cheap price, experience just
Prompt, accurate shoes customize service.
The content of the invention
In order to solve the deficiencies in the prior art, can quick, convenient, accurately it be based on it is an object of the invention to provide one kind
The shoe tree parameter automatic prediction method and prediction meanss of shoes image and foot type image.
To achieve these goals, the invention provides a kind of shoe tree parameter based on shoes image and foot type image is automatic
Forecasting Methodology, comprises the following steps:
S1, collection shoes image and foot type image;
S2, training production confrontation network;
S3, expands existing shoes view data and foot type view data;
S4, six channel images are combined into by shoes image and corresponding foot type image, and base is trained using recurrent neural network
In shoes image and the shoe tree parameter model of foot type image;
S5, utilizes shoe tree parameter model automatic Prediction shoe tree parameter.
Preferably, the step S1 further comprises following steps:
S101, according to shoe tree parameter, collection shoes image and foot type image, each shoe tree parameter acquisition it is corresponding 8 not
With angle shoes image and foot type image;
One shoe tree parameter corresponding 8 different angle shoes images and foot type image are randomly divided into four groups, often by S102
Two angular images of group a, angular image includes shoes image and foot type image.
Preferably, step S2 further comprises following steps:
S201, data amplification, including flip horizontal, rotation transformation and random sanction are carried out to shoes image and foot type image
Cut, the amplification operation of image is identical in group;
S202, training production confrontation network, including generation network and differentiation network, conversion life is carried out by image in group
Into;
In S203, training process, group in image alternately as input and output be trained, alternately training generation model and
Discrimination model, and with back-propagation algorithm update two networks parameter.
Advantageously, step S202 includes:
Generation network is one and first encodes the process decoded afterwards, and input picture is that 3 channel resolutions are 256*256, first will
Input picture is parsed into 1*1*512 vector, one ReLU active coating of every layer of convolutional layer heel by 8 layers of convolutional layer;Then will
Vector after the parsing is by 8 layers of warp lamination, and up-sampling turns into the image that 3 channel resolutions are 256*256, every layer of convolutional layer
One ReLU active coating of heel;Finally code and decode between corresponding layer, there is corresponding super passage from coding layer to decoding layer
It is direct-connected;
It is the process of a coding to differentiate network, inputs the image for another image in group and generation network G generation
The figure of 6 passages of composition, including 5 convolutional layers, each one ReLU active coating of convolutional layer heel finally give 8*8's
Confidence level figure, for differentiating whether the image of generation network generation is true picture;
Training objective is:
Wherein, x, y are the true shoes image pattern and foot type image pattern of collection, and y makes a living into the shoes of network G generation
Image pattern and foot type image pattern.
Preferably, the step S3 is to utilize the existing shoes view data of generation network expansion and foot type image trained
Data, so that training sample is expanded into one times.
Preferably, the step S4 further comprises following steps:
S401, six channel images are combined into by shoes image and foot type image, using convolutional neural networks as carrying
Take the network of combination image feature;
The weight of S402, initialization convolutional neural networks and recurrent neural network, convolutional Neural net is inputted by combination image
In network, the corresponding shoe tree parameter of the combination image is sequentially input in recurrent neural network, forward pass calculating is done in neutral net,
The shoe tree parameter predicted, and with real shoe tree parameter calculation error, reverse propagated error, at the same update convolutional Neural net
Network and recurrent neural network parameter.
Preferably, the step S5 further comprises following steps:
S501, by shoes image to be predicted and foot type image input generation network, the new shoes image of generation and foot type
Image;
S502, the combination image of real combination image and generation is inputted in shoe tree parameter model respectively, to meter before doing
Calculation obtains two groups of shoe tree parameters;
S503, using the average value of this two groups of shoe tree parameters as final result, prediction obtains shoe tree parameter.
Present invention also offers a kind of shoe tree parameter automatic Prediction device based on shoes image and foot type image, including footwear
Subgraph acquisition module, for gathering shoes image;Foot type image capture module, for gathering foot type image;Image expands mould
Block, for expanding shoes image and foot type image;Training module, for training production to resist network and recurrent neural network;
Memory module, for preserving shoes image, foot type image, shoe tree parameter and production confrontation network model, recurrent neural network
Model;Automatic Prediction module, for automatic Prediction shoe tree parameter.
Advantageously, the automatic Prediction module predicts true combination image shoe tree ginseng corresponding with generation combination image respectively
Number, then averages, obtains final shoe tree parameter prediction value.
For current shoe tree parameter measurement inefficiency, subjectivity by force, restriction productivity dynamics and production prices are reduced
Problem, designs shoe tree parameter automatic prediction method and device based on shoes image and foot type image, can be according to client's
Shoes image and the automatic expanding data of foot type image simultaneously predict corresponding shoe tree parameter.
Brief description of the drawings
Fig. 1 is the automatic prediction method flow chart that the present invention is provided;
Fig. 2 is the generation network that the present invention is provided and the expanded view for differentiating network;
Fig. 3 is the expanded view of convolutional neural networks provided by the present invention and recurrent neural network;
Fig. 4 is the structural representation for the automatic Prediction device preferred embodiment that the present invention is provided.
Embodiment
As shown in Figure 1, Figure 2 and Figure 3, the invention provides a kind of shoe tree parameter based on shoes image and foot type image certainly
Dynamic Forecasting Methodology, comprises the following steps:
S1, collection shoes image and foot type image;
S2, training production confrontation network;
S3, expands existing shoes view data and foot type view data;
S4, six channel images are combined into by shoes image and corresponding foot type image, and base is trained using recurrent neural network
In shoes image and the shoe tree parameter model of foot type image;
S5, utilizes shoe tree parameter model automatic Prediction shoe tree parameter.
Preferably, the step S1 further comprises following steps:
S101, according to shoe tree parameter, collection shoes image and foot type image, each shoe tree parameter acquisition it is corresponding 8 not
With angle shoes image and foot type image;
One shoe tree parameter corresponding 8 different angle shoes images and foot type image are randomly divided into four groups, often by S102
Two angular images of group a, angular image includes shoes image and foot type image.
Preferably, step S2 further comprises following steps:
S201, data amplification, including flip horizontal, rotation transformation and random sanction are carried out to shoes image and foot type image
Cut, the amplification operation of image is identical in group;
S202, training production confrontation network, including generation network and differentiation network, conversion life is carried out by image in group
Into;
In S203, training process, group in image alternately as input and output be trained, alternately training generation model and
Discrimination model, and with back-propagation algorithm update two networks parameter.
In the step S101, shoe tree parameter, specifically include last carving length, pin plantar enclose G2, pin plantar enclose G3, pin pocket with enclose G4,
Pin is wide, the palm is wide, wide, the 5th plantar toe profile in wide, little toe evagination dot profile outer width, the first plantar toe profile in thumb evagination dot profile
Outer width, waist curve outer width, heel heart profile overall with, thumb heights, foot type G2 are high, foot type G3 is high, foot type G4 is high, pin projection bottom surface
Product, G2 areas, G3 areas, G4 areas, the preceding middle section areas of G4, pin front volume, whole pin volume.
In the step S201, shoes image and foot type image are pre-processed, including carry out flip horizontal, rotation
Conversion and random cropping, for increasing sample size.
In the step S202, generation network is one and first encodes the process decoded afterwards, and input picture is 3 channel resolutions
Rate is 256*256, and input picture is first parsed into 1*1*512 vector, every layer of convolutional layer heel one by 8 layers of convolutional layer
ReLU active coatings;The subsequent vector by this after parsing is by 8 layers of warp lamination, and it is 256*256 that up-sampling, which turns into 3 channel resolutions,
Image, one ReLU active coating of every layer of convolutional layer heel;Finally code and decode between corresponding layer, from coding layer to decoding
Layer has corresponding super passage direct-connected;It is the process of a coding to differentiate network, is inputted as another image in group and generation net
The figure of 6 passages of the image composition of network G generations, including 5 convolutional layers, each one ReLU active coating of convolutional layer heel, finally
8*8 confidence level figure is obtained, for differentiating whether the image of generation network generation is true picture;Training objective is:
Wherein, x, y are the true shoes image pattern and foot type image pattern of collection, and y makes a living into the shoes of network G generation
Image pattern and foot type image pattern.
In the step S203, network is resisted using error back propagation (BP) Algorithm for Training production, according to differentiation
The error update network parameter predicted the outcome with actual sample of network, first fixed generation network, adjusts and differentiates network, when sentencing
It is fixed to differentiate network when the loss of other network no longer declines, adjustment generation network, so alternately, until two networks all
When reaching stable, deconditioning.
Preferably, the step S3 is to utilize the existing shoes view data of generation network expansion and foot type image trained
Data, so that training sample is expanded into one times.
Preferably, the step S4 further comprises following steps:
S401, six channel images are combined into by shoes image and foot type image, using convolutional neural networks as carrying
Take the network of combination image feature;
The weight of S402, initialization convolutional neural networks and recurrent neural network, convolutional Neural net is inputted by combination image
In network, the corresponding shoe tree parameter of the combination image is sequentially input in recurrent neural network, forward pass calculating is done in neutral net,
The shoe tree parameter predicted, and with real shoe tree parameter calculation error, reverse propagated error, at the same update convolutional Neural net
Network and recurrent neural network parameter.
Preferably, the step S5 further comprises following steps:
S501, by shoes image to be predicted and foot type image input generation network, the new shoes image of generation and foot type
Image;
S502, the combination image of real combination image and generation is inputted in shoe tree parameter model respectively, to meter before doing
Calculation obtains two groups of shoe tree parameters;
S503, using the average value of this two groups of shoe tree parameters as final result, prediction obtains shoe tree parameter.
As shown in figure 4, present invention also offers a kind of shoe tree parameter automatic Prediction based on shoes image and foot type image
Device, including shoes image capture module, for gathering shoes image;Foot type image capture module, for gathering foot type image;
Image enlargement module, for expanding shoes image and foot type image;Training module, for training production to resist network and recurrence
Neutral net;Memory module, for preserving shoes image, shoe tree parameter and production confrontation network model, recurrent neural network
Model;Automatic Prediction module, for automatic Prediction shoe tree parameter.
Preferably, the automatic Prediction module predicts true combination image shoe tree ginseng corresponding with generation combination image respectively
Number, then averages, obtains final shoe tree parameter prediction value.
Skilled worker is excessively relied on for current shoe tree parameter, the pass declined with cost is improved as restriction shoemaking productivity
The problem of key factor, design shoe tree parameter automatic prediction method and prediction meanss based on shoes image and foot type image, energy
The shoes image and foot type image by shooting consumer are reached, it is quick, convenient, accurate to obtain corresponding shoe tree parameter.
Preferred specific embodiment of the invention described in detail above.It should be appreciated that one of ordinary skill in the art
Just many modifications and variations can be made according to the design concept of the present invention without creative work.Therefore, all the art
Middle technical staff's design concept under this invention passes through logical analysis, reasoning, or a limited experiment on the basis of existing technology
Available technical scheme, all should be within the scope of the present invention and/or in the protection domain that is defined in the patent claims.
Claims (9)
1. a kind of shoe tree parameter automatic prediction method based on shoes image and foot type image, it is characterised in that including following step
Suddenly:
(S1) collection shoes image and foot type image;
(S2) training production confrontation network;
(S3) existing shoes view data and foot type view data are expanded;
(S4) shoes image and corresponding foot type image are combined into six channel images, footwear is based on using recurrent neural network training
The shoe tree parameter model of subgraph and foot type image;
(S5) shoe tree parameter model automatic Prediction shoe tree parameter is utilized.
2. a kind of shoe tree parameter automatic prediction method based on shoes image and foot type image according to claim 1, its
It is characterised by, the step (S1) comprises the following steps:
(S101) according to shoe tree parameter, collection shoes image and foot type image, the corresponding 8 different angles of each shoe tree parameter acquisition
Spend shoes image and foot type image;
(S102) a shoe tree parameter corresponding 8 different angle shoes images and foot type image are randomly divided into four groups, every group
Two angular images a, angular image includes shoes image and foot type image.
3. a kind of shoe tree parameter automatic prediction method based on shoes image and foot type image according to claim 2, its
It is characterised by, the step (S2) comprises the following steps:
(S201) data amplification, including flip horizontal, rotation transformation and random cropping, group are carried out to shoes image and foot type image
The amplification operation of interior image is identical;
(S202) training production confrontation network, including generation network and differentiation network, conversion generation is carried out by image in group;
(S203) in training process, image is trained alternately as input and output in group, is alternately trained generation model and is sentenced
Other model, and with back-propagation algorithm update two networks parameter.
4. a kind of shoe tree parameter automatic prediction method based on shoes image and foot type image according to claim 3, its
It is characterised by, the step (S202) includes:
Generation network is one and first encodes the process decoded afterwards, and input picture is that 3 channel resolutions are 256*256, first will input
Image is parsed into 1*1*512 vector, one ReLU active coating of every layer of convolutional layer heel by 8 layers of convolutional layer;Then this is solved
Vector after analysis is by 8 layers of warp lamination, and up-sampling turns into the image that 3 channel resolutions are 256*256, every layer of convolutional layer heel
One ReLU active coating;Finally code and decode between corresponding layer, have corresponding super passage direct-connected from coding layer to decoding layer;
It is the process of a coding to differentiate network, inputs the image composition for another image in group and generation network G generation
6 passages figure, including 5 convolutional layers, each one ReLU active coating of convolutional layer heel finally give 8*8 confidence
Degree figure, for differentiating whether the image of generation network generation is true picture;
Training objective is:
Wherein, x, y are the true shoes image pattern and foot type image pattern of collection, and y makes a living into the shoes image of network G generation
Sample and foot type image pattern.
5. a kind of shoe tree parameter automatic prediction method based on shoes image and foot type image according to claim 4, its
It is characterised by, the step (S3) is to utilize the existing shoes view data of generation network expansion and foot type picture number trained
According to so that training sample is expanded into one times.
6. a kind of shoe tree parameter automatic prediction method based on shoes image and foot type image according to claim 5, its
It is characterised by, the step (S4) comprises the following steps:
(S401) shoes image and foot type image are combined into six channel images, extraction group is used as using convolutional neural networks
Close the network of characteristics of image;
(S402) weight of initialization convolutional neural networks and recurrent neural network, convolutional neural networks are inputted by combination image
In, the corresponding shoe tree parameter of the combination image is sequentially input in recurrent neural network, forward pass calculating is done in neutral net, is obtained
To the shoe tree parameter of prediction, and with real shoe tree parameter calculation error, reverse propagated error, at the same update convolutional neural networks
With recurrent neural network parameter.
7. a kind of shoe tree parameter automatic prediction method based on shoes image and foot type image according to claim 6, its
It is characterised by, the step (S5) comprises the following steps:
(S501) by shoes image to be predicted and foot type image input generation network, new shoes image and foot type figure is generated
Picture;
(S502) combination image of real combination image and generation is inputted in shoe tree parameter model respectively, does forward calculation and obtain
To two groups of shoe tree parameters;
(S503) using the average value of this two groups of shoe tree parameters as final result, prediction obtains shoe tree parameter.
8. a kind of shoe tree parameter automatic Prediction device based on shoes image and foot type image, it is characterised in that including:
Shoes image capture module, for gathering shoes image;
Foot type image capture module, for gathering foot type image;
Image enlargement module, for expanding shoes image and foot type image;
Training module, for training production to resist network and recurrent neural network;
Memory module, for preserving shoes image, foot type image, shoe tree parameter and production confrontation network model, recurrent neural
Network model;
Automatic Prediction module, for automatic Prediction shoe tree parameter.
9. a kind of shoe tree parameter automatic Prediction device based on shoes image and foot type image according to claim 8, its
It is characterised by:The automatic Prediction module predicts true combination image shoe tree parameter corresponding with generation combination image respectively, so
After average, obtain final shoe tree parameter prediction value.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710351500.3A CN106993851B (en) | 2017-05-18 | 2017-05-18 | Shoe tree parameter automatic prediction method and prediction device based on shoe image and foot shape image |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710351500.3A CN106993851B (en) | 2017-05-18 | 2017-05-18 | Shoe tree parameter automatic prediction method and prediction device based on shoe image and foot shape image |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106993851A true CN106993851A (en) | 2017-08-01 |
CN106993851B CN106993851B (en) | 2022-02-18 |
Family
ID=59434426
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710351500.3A Active CN106993851B (en) | 2017-05-18 | 2017-05-18 | Shoe tree parameter automatic prediction method and prediction device based on shoe image and foot shape image |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106993851B (en) |
Cited By (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108364087A (en) * | 2018-01-11 | 2018-08-03 | 安徽优思天成智能科技有限公司 | A kind of spatio-temporal prediction method of urban mobile tail gas concentration |
CN108710896A (en) * | 2018-04-24 | 2018-10-26 | 浙江工业大学 | The field learning method of learning network is fought based on production |
CN109090769A (en) * | 2018-06-13 | 2018-12-28 | 东莞时谛智能科技有限公司 | A kind of reconstruct shoe last model detection method and system |
CN109614480A (en) * | 2018-11-26 | 2019-04-12 | 武汉大学 | A kind of generation method and device of the autoabstract based on production confrontation network |
CN109978985A (en) * | 2017-12-28 | 2019-07-05 | 同方威视科技江苏有限公司 | Data processing method and device, storage medium, electronic equipment |
CN110179204A (en) * | 2018-02-22 | 2019-08-30 | 意礴科技有限公司 | Recommend the system and method for the system and method and prediction shoe tree comfort level of shoe tree |
CN112425865A (en) * | 2019-08-08 | 2021-03-02 | 意礴数字科技(深圳)有限公司 | Method, device and system for generating shoe tree and evaluating comfort degree of shoe tree |
CN113205140A (en) * | 2021-05-06 | 2021-08-03 | 中国人民解放军海军航空大学航空基础学院 | Semi-supervised specific radiation source individual identification method based on generative countermeasure network |
US11250329B2 (en) | 2017-10-26 | 2022-02-15 | Nvidia Corporation | Progressive modification of generative adversarial neural networks |
US11263525B2 (en) | 2017-10-26 | 2022-03-01 | Nvidia Corporation | Progressive modification of neural networks |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP3156304U (en) * | 2009-06-15 | 2009-12-24 | アマーシン ルシモンAmeersing Luximon | Shoe mold design equipment |
CN101855637A (en) * | 2007-09-18 | 2010-10-06 | 马加里有限责任公司 | Method for calculating the geometry of a last for customised footwear |
CN101887477A (en) * | 2010-06-25 | 2010-11-17 | 温州大学 | Method for customizing digitalized shoe trees according to a plurality of images of foot shapes |
CN102063533A (en) * | 2010-12-27 | 2011-05-18 | 温州市鹿艺鞋材有限公司 | Method for designing shoe tree according to pelma image analysis data |
CN104820735A (en) * | 2015-04-17 | 2015-08-05 | 大族激光科技产业集团股份有限公司 | Method and device for making shoes based on three-dimensional reconstruction of feet |
CN105469441A (en) * | 2015-11-23 | 2016-04-06 | 南京脚度健康科技有限公司 | Shoe making method by uploading two-dimensional plane photo by network and establishing three-dimensional foot model |
-
2017
- 2017-05-18 CN CN201710351500.3A patent/CN106993851B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101855637A (en) * | 2007-09-18 | 2010-10-06 | 马加里有限责任公司 | Method for calculating the geometry of a last for customised footwear |
JP3156304U (en) * | 2009-06-15 | 2009-12-24 | アマーシン ルシモンAmeersing Luximon | Shoe mold design equipment |
CN101887477A (en) * | 2010-06-25 | 2010-11-17 | 温州大学 | Method for customizing digitalized shoe trees according to a plurality of images of foot shapes |
CN102063533A (en) * | 2010-12-27 | 2011-05-18 | 温州市鹿艺鞋材有限公司 | Method for designing shoe tree according to pelma image analysis data |
CN104820735A (en) * | 2015-04-17 | 2015-08-05 | 大族激光科技产业集团股份有限公司 | Method and device for making shoes based on three-dimensional reconstruction of feet |
CN105469441A (en) * | 2015-11-23 | 2016-04-06 | 南京脚度健康科技有限公司 | Shoe making method by uploading two-dimensional plane photo by network and establishing three-dimensional foot model |
Cited By (15)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US11763168B2 (en) | 2017-10-26 | 2023-09-19 | Nvidia Corporation | Progressive modification of generative adversarial neural networks |
US11250329B2 (en) | 2017-10-26 | 2022-02-15 | Nvidia Corporation | Progressive modification of generative adversarial neural networks |
US11263525B2 (en) | 2017-10-26 | 2022-03-01 | Nvidia Corporation | Progressive modification of neural networks |
CN109978985A (en) * | 2017-12-28 | 2019-07-05 | 同方威视科技江苏有限公司 | Data processing method and device, storage medium, electronic equipment |
CN109978985B (en) * | 2017-12-28 | 2023-11-07 | 同方威视科技江苏有限公司 | Data processing method and device, storage medium and electronic equipment |
CN108364087A (en) * | 2018-01-11 | 2018-08-03 | 安徽优思天成智能科技有限公司 | A kind of spatio-temporal prediction method of urban mobile tail gas concentration |
CN110179204B (en) * | 2018-02-22 | 2022-06-24 | 意礴数字科技(深圳)有限公司 | System and method for recommending shoe tree and system and method for predicting comfort level of shoe tree |
CN110179204A (en) * | 2018-02-22 | 2019-08-30 | 意礴科技有限公司 | Recommend the system and method for the system and method and prediction shoe tree comfort level of shoe tree |
CN108710896A (en) * | 2018-04-24 | 2018-10-26 | 浙江工业大学 | The field learning method of learning network is fought based on production |
CN108710896B (en) * | 2018-04-24 | 2021-10-29 | 浙江工业大学 | Domain learning method based on generative confrontation learning network |
CN109090769A (en) * | 2018-06-13 | 2018-12-28 | 东莞时谛智能科技有限公司 | A kind of reconstruct shoe last model detection method and system |
CN109614480A (en) * | 2018-11-26 | 2019-04-12 | 武汉大学 | A kind of generation method and device of the autoabstract based on production confrontation network |
CN112425865A (en) * | 2019-08-08 | 2021-03-02 | 意礴数字科技(深圳)有限公司 | Method, device and system for generating shoe tree and evaluating comfort degree of shoe tree |
CN113205140B (en) * | 2021-05-06 | 2022-11-15 | 中国人民解放军海军航空大学 | Semi-supervised specific radiation source individual identification method based on generative countermeasure network |
CN113205140A (en) * | 2021-05-06 | 2021-08-03 | 中国人民解放军海军航空大学航空基础学院 | Semi-supervised specific radiation source individual identification method based on generative countermeasure network |
Also Published As
Publication number | Publication date |
---|---|
CN106993851B (en) | 2022-02-18 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106993851A (en) | A kind of shoe tree parameter automatic prediction method and prediction meanss based on shoes image and foot type image | |
CN106296669B (en) | A kind of image quality evaluating method and device | |
CN108234870A (en) | Image processing method, device, terminal and storage medium | |
CN109685116A (en) | Description information of image generation method and device and electronic device | |
CN109215028A (en) | A kind of multiple-objection optimization image quality measure method based on convolutional neural networks | |
CN109102341B (en) | Old man portrait drawing method for old man service | |
CN110147833B (en) | Portrait processing method, device, system and readable storage medium | |
CN107578771A (en) | Audio recognition method and device, storage medium, electronic equipment | |
CN104599084A (en) | Crowd calculation quality control method and device | |
CN109754317A (en) | Merge interpretation clothes recommended method, system, equipment and the medium of comment | |
CN109740737A (en) | Convolutional neural networks quantizing method, device and computer equipment | |
CN104063686A (en) | System and method for performing interactive diagnosis on crop leaf segment disease images | |
CN110246148A (en) | The conspicuousness detection method of multi-modal depth information fusion and attention study | |
CN109635714A (en) | The antidote and device of file scanned image | |
CN108616892A (en) | A kind of 4G base station selection methods based on hybrid artificial immune system | |
CN107392446A (en) | A kind of step power station scheduling scheme evaluation method based on sensitivity analysis | |
CN110222816A (en) | Method for building up, image processing method and the device of deep learning model | |
CN106937774B (en) | Shoe tree size prediction method and device based on deep learning | |
CN109344879A (en) | A kind of decomposition convolution method fighting network model based on text-image | |
CN105096304B (en) | The method of estimation and equipment of a kind of characteristics of image | |
CN107845418A (en) | Intelligent health scale, intelligent health management method, system and storage medium | |
CN109448039A (en) | A kind of monocular depth estimation method based on depth convolutional neural networks | |
CN110148212A (en) | A kind of action sequence generation method and device, electronic equipment and storage medium | |
CN105989509A (en) | Service recommendation method for intelligent terminal and device | |
CN109886317A (en) | General image aesthetics appraisal procedure, system and equipment based on attention mechanism |
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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant |