CN109948577A - A kind of cloth recognition methods, device and storage medium - Google Patents
A kind of cloth recognition methods, device and storage medium Download PDFInfo
- Publication number
- CN109948577A CN109948577A CN201910239784.6A CN201910239784A CN109948577A CN 109948577 A CN109948577 A CN 109948577A CN 201910239784 A CN201910239784 A CN 201910239784A CN 109948577 A CN109948577 A CN 109948577A
- Authority
- CN
- China
- Prior art keywords
- cloth
- flower pattern
- image
- feature
- texture
- 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
Abstract
The present invention discloses a kind of cloth recognition methods, device and storage medium, this method comprises: passing through the color characteristic of color feature extracted model extraction cloth image;Pass through the flower pattern textural characteristics of cloth image described in flower pattern texture feature extraction model extraction;The feature of cloth is indicated using the color characteristic and flower pattern textural characteristics;The image of cloth to be identified is inputted, the identification to the fabric color feature and flower pattern textural characteristics to be identified is completed.The present invention combines flower pattern, texture and the color of cloth in identification, the color characteristic of cloth and flower pattern textural characteristics are separately handled, cloth substantially flower pattern and grain details in order to balance, Feature Selection Model Zuo Liangge branch, one classification for substantially flower pattern, one is used for the identification of the thinner feature of cloth, and recognition speed is fast, and recognition effect is good.
Description
Technical field
The present invention relates to cloth identification technology field more particularly to a kind of cloth recognition methods, device and storage medium.
Background technique
Cloth is the necessity of human lives.The flower pattern texture and color of cloth are all kinds of, and are not easy to describe in words that this makes
At the manufacturer of textile cloth, retailer and look for cloth personnel extremely difficult when looking for certain cloth.If by scanning or clapping to cloth
According to mode the cloth in matching cloth library is gone to cloth identification, the time for looking for cloth can be greatlyd save, and can be matched to and target
Cloth flower pattern and the similar cloth of color.If cloth figure is associated with the production technology of upper cloth, the whole of available cloth production
Information can reduce the cloth production cycle, improve company's benefit.Compared with other object identifications, cloth identification has its uniqueness, cloth
Picture structure feature is not strong, and feature is not easy to extract, and cloth has big flower pattern again and have a small texture, and when identification should identify flower pattern again
Take into account texture and color.Cloth flower pattern multiplicity, texture and orderly texture and random texture.Existing technology is to taking into account flower pattern line
The recognition effect of reason and color is also undesirable.
Summary of the invention
It is an object of the invention to by a kind of cloth recognition methods, device and storage medium, to solve background above skill
The problem of art part is mentioned.
To achieve this purpose, the present invention adopts the following technical scheme:
A kind of cloth recognition methods, this method comprises the following steps:
S101, pass through the color characteristic of color feature extracted model extraction cloth image;Pass through flower pattern texture feature extraction
The flower pattern textural characteristics of cloth image described in model extraction;
S102, the feature that cloth is indicated using the color characteristic and flower pattern textural characteristics;
The image of S103, input cloth to be identified, are completed to the fabric color feature to be identified and flower pattern textural characteristics
Identification.
Particularly, pass through the color characteristic of color feature extracted model extraction cloth image in the step S101, comprising:
Fixed size size image is cut out from cloth figure, counts its color histogram, indicates the colouring information of cloth,
Extract the color characteristic of cloth image.
Particularly, pass through the flower pattern line of cloth image described in flower pattern texture feature extraction model extraction in the step S101
Manage feature, comprising:
By the flower pattern and texture information of flower pattern texture feature extraction model extraction cloth image, using deep learning network
Extract its flower pattern textural characteristics.
Particularly, the flower pattern and texture information by flower pattern texture feature extraction model extraction cloth image uses
Deep learning network extracts its flower pattern textural characteristics, before further include:
One, cloth image classification: to cloth label, being divided into N number of classification for cloth according to the difference of cloth flower pattern and texture,
Wherein, each classification includes m group;
Two, cloth image preprocessing: each cloth image is pre-processed, it is several that a cloth image, which is expanded,
Open image, comprising: random cropping is carried out to a cloth image, it is random to add by the Image Adjusting after cutting to fixed size
RGB figure is turned grayscale image, generates several images by illumination and contrast.
Particularly, the flower pattern and texture information by flower pattern texture feature extraction model extraction cloth image uses
Deep learning network extracts its flower pattern textural characteristics, comprising:
Cloth image is sent into depth characteristic and extracts network, and depth characteristic is extracted after network uses feature L2 regularization and done
Feature after embedding, embedding is divided to two branch process: softmax loss is done after connecing full articulamentum by a branch,
For the classification to cloth classification;Three sons loss (Tripletloss) is done by second branch, for distinguishing cloth group.
Particularly, the step S103 includes: the image for inputting cloth to be identified, using Pasteur's distance metric color characteristic
Similarity;Cloth to be identified is indicated from embedding layers of extraction feature vector by trained flower pattern texture recognition model
The flower pattern and textural characteristics of material, the distance for calculating feature vector indicate the similarity of cloth.
Based on above-mentioned cloth recognition methods, the invention also discloses a kind of cloth identification device, which includes:
Color feature extracted unit, for passing through the color characteristic of color feature extracted model extraction cloth image;
Flower pattern texture feature extraction unit, for the flower by cloth image described in flower pattern texture feature extraction model extraction
Type textural characteristics;
Cloth feature unit, for indicating the feature of cloth using the color characteristic and flower pattern textural characteristics;
Cloth recognition unit, it is special to the fabric color to be identified for completing according to the image for inputting cloth to be identified
The identification of flower pattern of seeking peace textural characteristics.
Particularly, the color feature extracted unit is specifically used for: fixed size size image is cut out from cloth figure,
Its color histogram is counted, indicates the colouring information of cloth, extracts the color characteristic of cloth image;
The flower pattern texture feature extraction unit is specifically used for: passing through flower pattern texture feature extraction model extraction cloth image
Flower pattern and texture information, its flower pattern textural characteristics is extracted using deep learning network;
Wherein, the flower pattern and texture information by flower pattern texture feature extraction model extraction cloth image, using depth
Degree learning network extracts its flower pattern textural characteristics, before further include: one, cloth image classification: to cloth label, according to cloth flower
Cloth is divided into N number of classification by the difference of type and texture, wherein each classification includes m group;Two, cloth image preprocessing: right
Each cloth image is pre-processed, and a cloth image is expanded as several images;It is described to pass through flower pattern textural characteristics
The flower pattern and texture information for extracting model extraction cloth image, extract its flower pattern textural characteristics using deep learning network, comprising:
Cloth image is sent into depth characteristic and extracts network, and depth characteristic is extracted after network uses feature L2 regularization and is embedding,
Feature after embedding is divided to two branch process: softmax loss is done after connecing full articulamentum by a branch, for cloth
The classification of classification;Three sons loss (Triplet loss) is done by second branch, for distinguishing cloth group.
Particularly, the cloth recognition unit is specifically used for: the image of cloth to be identified is inputted, using Pasteur's distance metric
The similarity of color characteristic;It is indicated by trained flower pattern texture recognition model from embedding layers of extraction feature vector
The flower pattern and textural characteristics of cloth to be identified, the distance for calculating feature vector indicate the similarity of cloth.
The present invention also provides a kind of computer readable storage mediums, store computer program instructions thereon, when the meter
Such as method of any of claims 1-6 is realized when calculation machine program instruction is executed by processor.
Cloth recognition methods, device and storage medium proposed by the present invention combine flower pattern, the line of cloth in identification
Reason and color, the color characteristic of cloth and flower pattern textural characteristics are separately handled, cloth substantially flower pattern and texture is thin in order to balance
Section, Feature Selection Model Zuo Liangge branch, a classification for substantially flower pattern, one is used for the identification of the thinner feature of cloth,
Recognition speed is fast, and recognition effect is good.
Detailed description of the invention
Fig. 1 is cloth recognition methods flow diagram provided in an embodiment of the present invention;
Fig. 2 is cloth classification provided in an embodiment of the present invention and group classification schematic diagram;
Fig. 3 is that cloth image provided in an embodiment of the present invention carries out pretreatment process schematic diagram;
Fig. 4 is feature extraction network overall architecture schematic diagram provided in an embodiment of the present invention;
Fig. 5 is provided in an embodiment of the present invention three sub- learning process schematic diagrames.
Specific embodiment
Present invention will be further explained below with reference to the attached drawings and examples.It is understood that tool described herein
Body embodiment is used only for explaining the present invention rather than limiting the invention.It also should be noted that for the ease of retouching
It states, only some but not all contents related to the present invention are shown in the drawings, it is unless otherwise defined, used herein all
Technical and scientific term has the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.It is used herein
Term be intended merely to description specific embodiment, it is not intended that in limitation the present invention.
Embodiment one
It please refers to shown in Fig. 1, Fig. 1 is cloth recognition methods flow diagram provided in an embodiment of the present invention.
Cloth recognition methods specifically comprises the following steps: in the present embodiment
S101, pass through the color characteristic of color feature extracted model extraction cloth image;Pass through flower pattern texture feature extraction
The flower pattern textural characteristics of cloth image described in model extraction.
S102, the feature that cloth is indicated using the color characteristic and flower pattern textural characteristics.
The image of S103, input cloth to be identified, are completed to the fabric color feature to be identified and flower pattern textural characteristics
Identification.
Specifically, pass through the color characteristic of color feature extracted model extraction cloth image described in the present embodiment, packet
It includes: cutting out fixed size size image from cloth figure, count its color histogram, indicate the colouring information of cloth, extract
The color characteristic of cloth image.
It is described in the present embodiment special by the flower pattern texture of cloth image described in flower pattern texture feature extraction model extraction
Sign, comprising: by the flower pattern and texture information of flower pattern texture feature extraction model extraction cloth image, using deep learning network
Extract its flower pattern textural characteristics.Wherein, the flower pattern and texture by flower pattern texture feature extraction model extraction cloth image
Information extracts its flower pattern textural characteristics using deep learning network, before further include:
One, to cloth label, cloth cloth image classification: is divided by plain color, lattice according to the difference of cloth flower pattern and texture
N number of classifications such as son, striped, canine tooth, hidden item, wherein each classification includes m group.One group is a cloth or several
Extremely similar cloth.Wherein, a cloth or several extremely similar clothes refer to that the whole flower pattern of cloth is similar with texture,
For example be all plain color cloth and texture is all twill, it is all that the width of striped cloth and striped is close, textile texture is identical, and color can be with
It is different.Classification and group are classified as follows shown in Fig. 2.
Two, cloth image preprocessing: a cloth or several extremely similar clothes regard a group as, are used to cloth
Thinner identification is carried out, each cloth image is pre-processed, a cloth image is expanded as several images, specifically
Process is as shown in Figure 3: random cropping is carried out to a cloth image, the Image Adjusting (resize) after cutting is big to fixing
It is small, it is random to add illumination and contrast, RGB figure is turned into grayscale image, generates several images;In the present embodiment by a cloth
Image amplification is 30 images.
Specifically, passing through the flower pattern of flower pattern texture feature extraction model extraction cloth image and texture letter in the present embodiment
Breath, extracts its flower pattern textural characteristics using deep learning network.As shown in figure 4, cloth image, which is sent into depth characteristic, extracts network,
Depth characteristic extracts network using embedding is after feature L2 regularization, and the feature after embedding is divided to two bifurcations
Reason: softmax loss is done after connecing full articulamentum by a branch, for the classification to cloth classification;Wherein, embedding refers to
It is N-dimensional vector by the cloth Feature Conversion of extraction (size of N is settable);Softmax refers to Softmax layers of computation model cloth
Material is divided into the probability of each classification, selects the class of maximum probability as output classification;The loss of three sons is done by second branch
(Tripletloss), for distinguishing cloth group;Wherein, it is lost about three sons: loss function:
In formula,For target cloth,For positive sample cloth,For negative sample cloth, α is positive negative sample interval, and function f () will
Cloth figure switchs to feature vector.As shown in figure 5, constructing positive negative sample to each group of other cloth, a target cloth is selected, with
Target cloth is that same group of other cloth figure is considered as its positive sample, and difference organizes other cloth and is considered as its negative sample;Cloth image three
Join sub- selection method: selecting a cloth image as target cloth first, at it with one figure conduct of random selection in group
Its positive sample;Then a group is randomly choosed, and is different groups from target cloth, a figure is randomly choosed in the group
As negative sample.
The step S103 includes: the image for inputting cloth to be identified in the present embodiment, using Pasteur's distance metric face
The similarity of color characteristic, distance is closer, then fabric color is more close;By trained flower pattern texture recognition model, from
Embedding layers of extraction feature vector, indicate the flower pattern and textural characteristics of cloth to be identified, and the distance for calculating feature vector indicates
The similarity of cloth, the distance between cloth feature vector two clothes of nearlyr expression are more similar.
Embodiment two
Based on above-mentioned cloth recognition methods, a kind of cloth identification device is present embodiments provided, which includes:
Color feature extracted unit, for passing through the color characteristic of color feature extracted model extraction cloth image.
Flower pattern texture feature extraction unit, for the flower by cloth image described in flower pattern texture feature extraction model extraction
Type textural characteristics.
Cloth feature unit, for indicating the feature of cloth using the color characteristic and flower pattern textural characteristics.
Cloth recognition unit, it is special to the fabric color to be identified for completing according to the image for inputting cloth to be identified
The identification of flower pattern of seeking peace textural characteristics.
Specifically, the color feature extracted unit is specifically used in the present embodiment: cutting out fixation from cloth figure
Size size image, counts its color histogram, indicates the colouring information of cloth, extracts the color characteristic of cloth image.It is described
Flower pattern texture feature extraction unit is specifically used for: passing through the flower pattern and texture of flower pattern texture feature extraction model extraction cloth image
Information extracts its flower pattern textural characteristics using deep learning network;Wherein, described to pass through flower pattern texture feature extraction model extraction
The flower pattern and texture information of cloth image extract its flower pattern textural characteristics using deep learning network, before further include: one, cloth
Expect image classification: to cloth label, cloth being divided by N number of classification according to the difference of cloth flower pattern and texture, wherein each classification
Include m group;Two, cloth image preprocessing: pre-processing each cloth image, is by a cloth image amplification
Several images;The flower pattern and texture information by flower pattern texture feature extraction model extraction cloth image, using depth
Learning network extracts its flower pattern textural characteristics, comprising: cloth image is sent into depth characteristic and extracts network, and depth characteristic extracts network
Using embedding is after feature L2 regularization, the feature after embedding is divided to two branch process: a branch connects to be connected entirely
Softmax loss is done after connecing layer, for the classification to cloth classification;Three sons loss (Triplet is in second branch
Loss), for distinguishing cloth group.Positive negative sample is constructed to each group of other cloth, a target cloth is selected, with target cloth
It is that same group of other cloth figure is considered as its positive sample, difference organizes other cloth and is considered as its negative sample;The choosing of three son of cloth image
Selection method: selecting a cloth image as target cloth first, is used as its positive sample with a figure is randomly choosed in group at it
This;Then a group is randomly choosed, and is different groups from target cloth, a figure is randomly choosed in the group as negative
Sample.
The cloth recognition unit is specifically used in the present embodiment: input the image of cloth to be identified, using Pasteur away from
Similarity from measurement color characteristic, distance is closer, then fabric color is more close;Pass through trained flower pattern texture recognition mould
Type indicates the flower pattern and textural characteristics of cloth to be identified from embedding layers of extraction feature vector, calculate feature vector away from
From the similarity for indicating cloth, the distance between cloth feature vector two clothes of nearlyr expression are more similar.
Embodiment three
A kind of computer readable storage medium is present embodiments provided, computer program is stored on the storage medium and refers to
It enables, the cloth recognition methods as described in above-described embodiment one is realized when the computer program instructions are executed by processor.
Technical solution of the present invention combines flower pattern, texture and the color of cloth in identification, and the color of cloth is special
Flower pattern of seeking peace textural characteristics are separately handled, and cloth substantially flower pattern and grain details, Feature Selection Model does two points in order to balance
Branch, a classification for substantially flower pattern, one is used for the identification of the thinner feature of cloth, and recognition speed is fast, and recognition effect is good.
Those of ordinary skill in the art will appreciate that realizing that all or part in above-described embodiment is can to pass through calculating
Machine program is completed to instruct relevant hardware, and the program can be stored in a computer-readable storage medium, the journey
Sequence is when being executed, it may include such as the process of the embodiment of above-mentioned each method.Wherein, the storage medium can for magnetic disk, CD,
Read-only memory or random access memory etc..
Note that the above is only a better embodiment of the present invention and the applied technical principle.It will be appreciated by those skilled in the art that
The invention is not limited to the specific embodiments described herein, be able to carry out for a person skilled in the art it is various it is apparent variation,
It readjusts and substitutes without departing from protection scope of the present invention.Therefore, although being carried out by above embodiments to the present invention
It is described in further detail, but the present invention is not limited to the above embodiments only, without departing from the inventive concept, also
It may include more other equivalent embodiments, and the scope of the invention is determined by the scope of the appended claims.
Claims (10)
1. a kind of cloth recognition methods, which comprises the steps of:
S101, pass through the color characteristic of color feature extracted model extraction cloth image;Pass through flower pattern texture feature extraction model
Extract the flower pattern textural characteristics of the cloth image;
S102, the feature that cloth is indicated using the color characteristic and flower pattern textural characteristics;
The image of S103, input cloth to be identified, complete the knowledge to the fabric color feature and flower pattern textural characteristics to be identified
Not.
2. cloth recognition methods according to claim 1, which is characterized in that mentioned in the step S101 by color characteristic
Take the color characteristic of model extraction cloth image, comprising:
Fixed size size image is cut out from cloth figure, counts its color histogram, indicates the colouring information of cloth, is extracted
The color characteristic of cloth image.
3. cloth recognition methods according to claim 1, which is characterized in that special by flower pattern texture in the step S101
Sign extracts the flower pattern textural characteristics of cloth image described in model extraction, comprising:
By the flower pattern and texture information of flower pattern texture feature extraction model extraction cloth image, extracted using deep learning network
Its flower pattern textural characteristics.
4. cloth recognition methods according to claim 3, which is characterized in that described to pass through flower pattern texture feature extraction model
The flower pattern and texture information for extracting cloth image, extract its flower pattern textural characteristics using deep learning network, before further include:
One, cloth image classification: to cloth label, being divided into N number of classification for cloth according to the difference of cloth flower pattern and texture,
In, each classification includes m group;
Two, cloth image preprocessing: each cloth image is pre-processed, a cloth image is expanded as several figures
Picture, comprising: random cropping is carried out to a cloth image, the Image Adjusting after cutting to fixed size is added into illumination at random
And contrast, RGB figure is turned into grayscale image, generates several images.
5. cloth recognition methods according to claim 4, which is characterized in that described to pass through flower pattern texture feature extraction model
The flower pattern and texture information for extracting cloth image, extract its flower pattern textural characteristics using deep learning network, comprising:
Cloth image is sent into depth characteristic and extracts network, and depth characteristic is extracted after network uses feature L2 regularization and done
Feature after embedding, embedding is divided to two branch process: softmax loss is done after connecing full articulamentum by a branch,
For the classification to cloth classification;The loss of three sons is done by second branch, for distinguishing cloth group.
6. cloth recognition methods according to claim 5, which is characterized in that the step S103 includes: that input is to be identified
The image of cloth, using the similarity of Pasteur's distance metric color characteristic;By trained flower pattern texture recognition model, from
Embedding layers of extraction feature vector, indicate the flower pattern and textural characteristics of cloth to be identified, and the distance for calculating feature vector indicates
The similarity of cloth.
7. a kind of cloth identification device, which is characterized in that the device includes:
Color feature extracted unit, for passing through the color characteristic of color feature extracted model extraction cloth image;
Flower pattern texture feature extraction unit, for the flower pattern line by cloth image described in flower pattern texture feature extraction model extraction
Manage feature;
Cloth feature unit, for indicating the feature of cloth using the color characteristic and flower pattern textural characteristics;
Cloth recognition unit, for according to the image for inputting cloth to be identified, completing to the fabric color feature to be identified and
The identification of flower pattern textural characteristics.
8. cloth identification device according to claim 7, which is characterized in that the color feature extracted unit is specifically used
In: fixed size size image is cut out from cloth figure, counts its color histogram, indicates the colouring information of cloth, is extracted
The color characteristic of cloth image;
The flower pattern texture feature extraction unit is specifically used for: passing through the flower of flower pattern texture feature extraction model extraction cloth image
Type and texture information extract its flower pattern textural characteristics using deep learning network;
Wherein, the flower pattern and texture information by flower pattern texture feature extraction model extraction cloth image, using depth
Practise network and extract its flower pattern textural characteristics, before further include: one, cloth image classification: to cloth label, according to cloth flower pattern and
Cloth is divided into N number of classification by the difference of texture, wherein each classification includes m group;Two, cloth image preprocessing: to each
It opens cloth image to be pre-processed, a cloth image is expanded as several images;It is described to pass through flower pattern texture feature extraction
The flower pattern and texture information of model extraction cloth image extract its flower pattern textural characteristics using deep learning network, comprising: cloth
Image is sent into depth characteristic and extracts network, and depth characteristic is extracted after network uses feature L2 regularization and is embedding,
Feature after embedding is divided to two branch process: softmax loss is done after connecing full articulamentum by a branch, for cloth
The classification of classification;The loss of three sons is done by second branch, for distinguishing cloth group.
9. cloth identification device according to claim 8, which is characterized in that the cloth recognition unit is specifically used for: defeated
The image for entering cloth to be identified, using the similarity of Pasteur's distance metric color characteristic;Pass through trained flower pattern texture recognition
Model indicates the flower pattern and textural characteristics of cloth to be identified, calculates feature vector from embedding layers of extraction feature vector
Distance indicates the similarity of cloth.
10. a kind of computer readable storage medium, stores computer program instructions thereon, which is characterized in that when the computer
It realizes when program instruction is executed by processor such as any one of claim 1-6 the method.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910239784.6A CN109948577B (en) | 2019-03-27 | 2019-03-27 | Cloth identification method and device and storage medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910239784.6A CN109948577B (en) | 2019-03-27 | 2019-03-27 | Cloth identification method and device and storage medium |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109948577A true CN109948577A (en) | 2019-06-28 |
CN109948577B CN109948577B (en) | 2020-08-04 |
Family
ID=67012019
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910239784.6A Active CN109948577B (en) | 2019-03-27 | 2019-03-27 | Cloth identification method and device and storage medium |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109948577B (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110321850A (en) * | 2019-07-05 | 2019-10-11 | 杭州时趣信息技术有限公司 | Garment material automatic identifying method, device, system, equipment and storage medium |
CN114253199A (en) * | 2022-03-01 | 2022-03-29 | 江苏苏美达纺织有限公司 | Weaving control method and system |
Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104331447A (en) * | 2014-10-29 | 2015-02-04 | 邱桃荣 | Cloth color card image retrieval method |
CN107256246A (en) * | 2017-06-06 | 2017-10-17 | 西安工程大学 | PRINTED FABRIC image search method based on convolutional neural networks |
EP3404582A1 (en) * | 2017-05-15 | 2018-11-21 | Siemens Aktiengesellschaft | Training an rgb-d classifier with only depth data and privileged information |
WO2018213108A1 (en) * | 2017-05-15 | 2018-11-22 | Siemens Aktiengesellschaft | Domain adaptation and fusion using weakly supervised target irrelevant data |
CN108875827A (en) * | 2018-06-15 | 2018-11-23 | 广州深域信息科技有限公司 | A kind of method and system of fine granularity image classification |
CN108960265A (en) * | 2017-05-22 | 2018-12-07 | 阿里巴巴集团控股有限公司 | Optimization method, image classification method, the apparatus and system of image classification process |
CN108960080A (en) * | 2018-06-14 | 2018-12-07 | 浙江工业大学 | Based on Initiative Defense image to the face identification method of attack resistance |
CN109117862A (en) * | 2018-06-29 | 2019-01-01 | 北京达佳互联信息技术有限公司 | Image tag recognition methods, device and server |
CN109344845A (en) * | 2018-09-21 | 2019-02-15 | 哈尔滨工业大学 | A kind of feature matching method based on Triplet deep neural network structure |
-
2019
- 2019-03-27 CN CN201910239784.6A patent/CN109948577B/en active Active
Patent Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104331447A (en) * | 2014-10-29 | 2015-02-04 | 邱桃荣 | Cloth color card image retrieval method |
EP3404582A1 (en) * | 2017-05-15 | 2018-11-21 | Siemens Aktiengesellschaft | Training an rgb-d classifier with only depth data and privileged information |
WO2018213108A1 (en) * | 2017-05-15 | 2018-11-22 | Siemens Aktiengesellschaft | Domain adaptation and fusion using weakly supervised target irrelevant data |
CN108960265A (en) * | 2017-05-22 | 2018-12-07 | 阿里巴巴集团控股有限公司 | Optimization method, image classification method, the apparatus and system of image classification process |
CN107256246A (en) * | 2017-06-06 | 2017-10-17 | 西安工程大学 | PRINTED FABRIC image search method based on convolutional neural networks |
CN108960080A (en) * | 2018-06-14 | 2018-12-07 | 浙江工业大学 | Based on Initiative Defense image to the face identification method of attack resistance |
CN108875827A (en) * | 2018-06-15 | 2018-11-23 | 广州深域信息科技有限公司 | A kind of method and system of fine granularity image classification |
CN109117862A (en) * | 2018-06-29 | 2019-01-01 | 北京达佳互联信息技术有限公司 | Image tag recognition methods, device and server |
CN109344845A (en) * | 2018-09-21 | 2019-02-15 | 哈尔滨工业大学 | A kind of feature matching method based on Triplet deep neural network structure |
Non-Patent Citations (1)
Title |
---|
刘嘉唯,等: "融合颜色与l王P纹理特征的布料色卡图像检索", 《软件导刊》 * |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110321850A (en) * | 2019-07-05 | 2019-10-11 | 杭州时趣信息技术有限公司 | Garment material automatic identifying method, device, system, equipment and storage medium |
CN114253199A (en) * | 2022-03-01 | 2022-03-29 | 江苏苏美达纺织有限公司 | Weaving control method and system |
CN114253199B (en) * | 2022-03-01 | 2022-05-06 | 江苏苏美达纺织有限公司 | Weaving control method and system |
Also Published As
Publication number | Publication date |
---|---|
CN109948577B (en) | 2020-08-04 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110348319B (en) | Face anti-counterfeiting method based on face depth information and edge image fusion | |
US10019655B2 (en) | Deep-learning network architecture for object detection | |
Rachmadi et al. | Vehicle color recognition using convolutional neural network | |
CN105069400B (en) | Facial image gender identifying system based on the sparse own coding of stack | |
US20120155766A1 (en) | Patch description and modeling for image subscene recognition | |
US9025864B2 (en) | Image clustering using a personal clothing model | |
CN109657595A (en) | Based on the key feature Region Matching face identification method for stacking hourglass network | |
CN104091341B (en) | A kind of image fuzzy detection method based on conspicuousness detection | |
CN109214420A (en) | The high texture image classification method and system of view-based access control model conspicuousness detection | |
CN109635676A (en) | A method of positioning source of sound from video | |
CN111832443B (en) | Construction method and application of construction violation detection model | |
CN111027377B (en) | Double-flow neural network time sequence action positioning method | |
CN109740572A (en) | A kind of human face in-vivo detection method based on partial color textural characteristics | |
US20220076117A1 (en) | Methods for generating a deep neural net and for localising an object in an input image, deep neural net, computer program product, and computer-readable storage medium | |
CN106408037A (en) | Image recognition method and apparatus | |
Qin et al. | A fast and robust text spotter | |
CN110728302A (en) | Method for identifying color textile fabric tissue based on HSV (hue, saturation, value) and Lab (Lab) color spaces | |
CN110059677A (en) | Digital table recognition methods and equipment based on deep learning | |
CN109948577A (en) | A kind of cloth recognition methods, device and storage medium | |
WO2019229524A2 (en) | Neural network calculation method and system, and corresponding dual neural network implementation | |
CN104851183A (en) | Paper currency face and orientation recognition method and device | |
CN110223310A (en) | A kind of line-structured light center line and cabinet edge detection method based on deep learning | |
CN108595558A (en) | A kind of image labeling method of data balancing strategy and multiple features fusion | |
Li et al. | Recognizing human actions by BP-AdaBoost algorithm under a hierarchical recognition framework | |
CN107274425B (en) | A kind of color image segmentation method and device based on Pulse Coupled Neural Network |
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 |