WO2019011342A1

WO2019011342A1 - Cloth identification method and device, electronic device and storage medium

Info

Publication number: WO2019011342A1
Application number: PCT/CN2018/095742
Authority: WO
Inventors: 斯科特·马修·罗伯特; 黄鼎隆; 傅恺
Original assignee: 深圳码隆科技有限公司
Priority date: 2017-07-14
Filing date: 2018-07-16
Publication date: 2019-01-17
Also published as: CN107358243A

Abstract

Proposed are a cloth identification method and device, an electronic device and a storage medium. The method comprises: acquiring an image of cloth to be identified; extracting a cloth feature corresponding to the cloth to be identified in the image; and identifying, based on a pre-generated cloth identification neural network, the cloth feature to generate an identification result corresponding to the cloth to be identified. Thereby, the embodiments of the present disclosure realize the identification of cloth in an automated way, thus greatly reducing the labour cost. A dedicated cloth identification neural network is used in an identification process, so that a large quantity of various cloth can be identified, and the accuracy of identification can be effectively ensured while the identification efficiency is improved.

Description

Method, device, electronic device and storage medium for fabric identification

Cross-reference to related applications

The present disclosure claims priority to Chinese Patent Application No. JP-A No. No. No. No. No. No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No

Technical field

The present disclosure relates to the field of image recognition and data processing, and in particular, to a method and device for fabric identification, an electronic device, and a storage medium.

Background technique

With the development of social economy, the types of fabrics are becoming more and more abundant. At present, the identification of fabrics is based on manual methods, but this method is time-consuming and laborious and inefficient. Moreover, due to the increasing labor costs and the increasing variety of fabrics, the manual method can't cope with this quantitative level of identification needs. Therefore, the existing cloth identification methods are not suitable for the current social economy. Development needs.

Summary of the invention

In view of the defects in the prior art, the present disclosure proposes a method and device for fabric identification, an apparatus, an electronic device and a storage medium, which are identified by a special cloth recognition neural network to effectively identify the cloth, thereby effectively improving the recognition efficiency.

Specifically, the present disclosure proposes the following specific embodiments:

The embodiment of the present disclosure provides a method for fabric identification, including:

Obtaining an image of the fabric to be identified;

Extracting a cloth feature corresponding to the cloth to be identified in the image;

The recognition result of the cloth to be identified is generated based on the fabric feature identified by the pre-generated cloth identification neural network.

In a specific embodiment, the fabric features include: a color of the cloth, and/or a pattern of the cloth, and/or a texture of the cloth; wherein the pattern and texture are obtained by contouring the image.

In a specific embodiment, the “recognizing the cloth feature based on the pre-generated cloth identification neural network to generate the recognition result corresponding to the cloth to be identified” includes:

Determining a data format corresponding to the pre-generated cloth identification neural network;

If the data format includes a binary data format, the cloth feature is converted into a binary data format and then input into the cloth recognition neural network to generate a recognition result corresponding to the cloth to be identified.

In a specific embodiment, the cloth identification neural network is trained by the cloth specimen data exceeding a certain threshold; wherein the cloth specimen data includes a cloth feature and a recognition result.

In a specific embodiment, the cloth specimen data includes triple data; wherein the triple data includes: source data, forward data belonging to the same class as the source data, and the source The data belongs to different types of reverse data;

The source data is data randomly acquired from the cloth sample data;

The forward data is data that is randomly acquired in the cloth sample data and is consistent with the recognition result of the source data;

The reverse data is data that is randomly acquired in the cloth sample data and is inconsistent with the recognition result of the source data.

An embodiment of the present disclosure further provides a device for fabric identification, including:

Obtaining a module configured to obtain an image of the fabric to be identified;

An extraction module configured to extract a cloth feature corresponding to the fabric to be identified in the image;

The identification module is configured to generate a recognition result corresponding to the cloth to be identified based on identifying the cloth feature by a pre-generated cloth identification neural network.

In a specific embodiment, the identification module is configured to:

The source data is data randomly acquired from the cloth sample data;

Therefore, the embodiment of the present disclosure provides a method and device for fabric identification, an apparatus, an electronic device, and a storage medium, wherein the method includes: acquiring an image of the cloth to be identified; and extracting a cloth corresponding to the cloth to be identified in the image. a feature; identifying the cloth feature based on the pre-generated cloth recognition neural network, and generating a recognition result corresponding to the cloth to be identified. With the embodiment of the present disclosure, the identification of the cloth is realized in an automated manner, which greatly reduces the labor cost. Since the neural network is identified by a dedicated cloth during the recognition process, it is possible to recognize not only a large variety of fabrics, but also to improve the recognition efficiency while ensuring the accuracy of recognition.

DRAWINGS

In order to more clearly illustrate the technical solutions of the embodiments of the present disclosure, the drawings to be used in the embodiments will be briefly described below. It should be understood that the following drawings show only certain embodiments of the present disclosure, and thus It should be seen as a limitation on the scope, and those skilled in the art can obtain other related drawings according to these drawings without any creative work.

FIG. 1 is a schematic diagram of a data processing device according to an embodiment of the present disclosure;

2 is a schematic flowchart of a malicious user identification method according to an embodiment of the present disclosure;

FIG. 3 is a schematic diagram of sub-steps of step S120 in the malicious user identification method according to an embodiment of the present disclosure;

FIG. 4 is a schematic diagram of a malicious user identification apparatus according to an embodiment of the present disclosure.

Icon: 100-data processing device; 110-malicious user identification device; 111-feature acquisition module; 112-similarity calculation module; 113-initialization module; 114-iteration calculation module; 115-identification module; 120-memory; processor.

Detailed ways

In the following, various embodiments of the present disclosure will be described more fully. The present disclosure can have various embodiments, and adjustments and changes can be made therein. It should be understood, however, that the present invention is not limited to the specific embodiments disclosed herein, but the invention is to be construed as being included within the spirit and scope of the various embodiments of the present disclosure. All adjustments, equivalents and/or alternatives.

In the following, the term "comprising" or "including" may be used in the various embodiments of the present disclosure to indicate the existence of the disclosed function, operation or element, and does not limit one or more functions, operations or elements. increase. In addition, the terms "comprising," "having," "," It should not be understood that the existence or addition of one or more features, numbers, steps, operations, components or components of one or more other features, numbers, steps, operations, components, components or combinations of the foregoing are excluded. Or the possibility of a combination of the foregoing.

In various embodiments of the present disclosure, the expression "or" or "at least one of A or / and B" includes any or all combinations of the simultaneously listed characters. For example, the expression "A or B" or "at least one of A or / and B" may include A, may include B, or may include both A and B.

Expressions used in various embodiments of the present disclosure (such as "first", "second", etc.) may modify various constituent elements in various embodiments, but the corresponding constituent elements may not be limited. For example, the above statements do not limit the order and/or importance of the elements. The above statements are only configured to distinguish one element from another. For example, the first user device and the second user device indicate different user devices, although both are user devices. For example, a first element could be termed a second element, and a second element could be termed a first element, without departing from the scope of the various embodiments disclosed herein.

It should be noted that if the description "connects" one constituent element to another constituent element, the first constituent element can be directly connected to the second constituent element and can be "connected" between the first constituent element and the second constituent element. "The third component. On the contrary, when a constituent element is "directly connected" to another constituent element, it is understood that there is no third constituent element between the first constituent element and the second constituent element.

The term "user" as used in various embodiments of the present disclosure may indicate a person using an electronic device or a device using an electronic device (eg, an artificial intelligence electronic device).

The terms used in the various embodiments of the present disclosure are only intended to describe the specific embodiments and are not intended to limit the various embodiments of the present disclosure. As used herein, the singular forms " All terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the various embodiments of the present disclosure pertain, unless otherwise defined. The term (such as a term defined in a commonly used dictionary) will be interpreted as having the same meaning as the contextual meaning in the related art and will not be interpreted as having an idealized meaning or an overly formal meaning, Unless clearly defined in the various embodiments of the present disclosure.

Example 1

Embodiment 1 of the present disclosure discloses a method for fabric identification, as shown in FIG. 1, comprising the following steps:

Step 101: Acquire an image of the cloth to be identified;

Specifically, an image of the existing fabric to be identified may be obtained from the network, or after a certain fabric is seen, it is desired to identify the fabric, and thus the image is captured to obtain the captured image, and may be acquired on other devices. The stored image containing the cloth may be obtained by simply obtaining the image corresponding to the cloth to be identified, and is not limited to the above specific acquisition methods.

Step 102: Extract a cloth feature corresponding to the fabric to be identified in the image;

Specifically, in addition to features such as color, pattern, texture, etc., other features may be selected as the cloth features according to the needs and characteristics of the cloth itself, for example, may also include shininess, etc., as long as the image can reflect the cloth. Features can be.

As for the manner of obtaining each feature, taking a pattern or a texture as an example, the image may be binarized, the image with the gray value is obtained by binarization, and the pattern is determined according to the contour of the image with the gray value. And the texture, specifically, determining the image contour based on the gray value comparison of each pixel of the image with the gray value, and further extracting the pattern and the texture based on the subsequently recognized image contour to obtain the fabric to be identified Cloth features. In addition to the gray-based binarization processing method, the image recognition can be performed by means of a binarized neural network to obtain the desired cloth feature of the fabric to be identified in the image.

In addition, the color of the cloth can be divided into various colors such as red, yellow, blue, and green. The specific color can be further subdivided, and further differentiated according to the difference in hue, brightness, and saturation. The finer the color distinction, the more accurate the subsequent recognition process.

Corresponding to other fabric features, such as patterns and textures, are also more detailed and more clear, the subsequent recognition is more accurate; in addition, the more features of the fabric, the corresponding subsequent recognition effect is better. The number of specific features and the accuracy of the features can be flexibly adjusted according to actual needs and the application scenarios of the fabric.

In a specific embodiment, the independent weighting coefficients may also be set according to different features. In the subsequent identification process, if the features are not fully satisfied, the weighting system may select a matching result with the highest matching degree as the matching result. The result of the recognition; of course, the weight coefficient of each feature is the same, and the specific can be flexibly set as needed.

Step 103: Generate a recognition result corresponding to the cloth to be identified based on identifying the cloth feature by using a pre-generated cloth identification neural network.

In a specific embodiment, the “recognizing the cloth feature by using a pre-generated cloth identifying neural network to generate a recognition result corresponding to the cloth to be identified” in step 103, as shown in FIG. 2 include:

Step 1031: Determine a data format corresponding to the pre-generated cloth identification neural network;

Step 1032: If the data format includes a binary data format, convert the cloth feature into a binary data format, and then input the cloth recognition neural network to generate a recognition result corresponding to the cloth to be identified.

Specifically, if the cloth recognition neural network can support the binary data format, the bad features are converted, and the specific data is replaced by a binary data format, which is recognized in the input neural network, since the machine language is binary, Through the binary data format, the recognition process can be speeded up, and no additional data conversion is required when the identification is performed, so that the recognition efficiency can be effectively improved.

Specifically, the cloth specimen data based on the known recognition result is trained to configure the cloth identification neural network for fabric identification. The more the cloth specimen data is, the better the data is. The more the data, the cloth-recognition nerve generated by the training. The stronger the recognition ability of the network, the specific cloth specimen data can be the data of the artificially labeled fabric, and the data of the cloth identification neural network itself can be identified. Specifically, for example, the cloth identification neural network identifies a certain identification fabric as polyester. The follow-up feedback is given to the professional users of the fabric. After obtaining the feedback from the professional users of the fabric, the data can be used as the cloth specimen data.

Specifically, the cloth specimen data includes a cloth feature and a recognition result corresponding to the cloth feature, and the recognition result may be a correct recognition result or an incorrect recognition result, and if it is a correct recognition result, the recognition result is also obtained. Marked as correct; as for the wrong recognition result, it is marked as an error.

In a specific embodiment, as shown in FIG. 3, the cloth specimen data includes triple data; wherein the triple data includes: source data, and the same type as the source data Reverse data to data, and to different classes from the source data;

The source data is data randomly acquired from the cloth sample data;

Here, the description will be made in a specific embodiment, for example, the first picture of the polyester photographed by the professional photographer in the cloth sample data, and the polyester photographed by the non-professionals in the cloth sample data. a second picture of the non-polyester (for example, silk) photographed in the second picture and the cloth sample data, the specific first picture is the source data with the highest matching degree, and the second picture is taken by a non-professional There is a gap between the definition and resolution of the first picture, but it is still a picture of polyester, which is forward data; as for the third picture, it is reverse data of reverse contrast during training, so as to pass the positive and negative In contrast, the recognition accuracy of the cloth recognition neural network is more effectively trained.

Example 2

Embodiment 2 of the present disclosure further discloses a device for fabric identification, as shown in FIG. 4, comprising:

The obtaining module 201 is configured to obtain an image of the cloth to be identified;

The extracting module 202 is configured to extract a cloth feature corresponding to the cloth to be identified in the image;

The identification module 203 is configured to generate a recognition result corresponding to the cloth to be identified based on identifying the cloth feature by a cloth recognition neural network generated in advance.

In a specific embodiment, the identification module 203 is configured to:

The source data is data randomly acquired from the cloth sample data;

A person skilled in the art can understand that the drawings are only a schematic diagram of a preferred implementation scenario, and the modules or processes in the drawings are not necessarily required to implement the disclosure.

A person skilled in the art may understand that the modules in the apparatus in the implementation scenario may be distributed in the apparatus for implementing the scenario according to the implementation scenario description, or may be correspondingly changed in one or more devices different from the implementation scenario. The modules of the above implementation scenarios may be combined into one module, or may be further split into multiple sub-modules.

The above-mentioned serial numbers are only for the description, and do not represent the advantages and disadvantages of the implementation scenario.

The above disclosure is only a few specific implementation scenarios of the present disclosure, but the disclosure is not limited thereto, and any changes that can be made by those skilled in the art should fall within the protection scope of the present disclosure.

Industrial applicability

The method and device for fabric identification provided by the embodiment of the present disclosure, based on the pre-generated cloth identification neural network, identify the cloth feature, and generate a recognition result corresponding to the cloth to be identified. Thus, the disclosed embodiment realizes the identification of the cloth in an automated manner, which greatly reduces the labor cost. Moreover, since the neural network is identified by a dedicated cloth in the recognition process, not only a large variety of fabrics can be identified, but also the recognition efficiency can be improved while the accuracy of the recognition can be effectively ensured.

Claims

A method for fabric identification, characterized in that it comprises:

Obtaining an image of the fabric to be identified;

Extracting a cloth feature corresponding to the cloth to be identified in the image;

The recognition result of the cloth to be identified is generated based on the fabric feature identified by the pre-generated cloth identification neural network.
A method of fabric identification according to claim 1 wherein said fabric features comprise: a color of the cloth, and/or a pattern of the cloth, and/or a texture of the cloth; wherein said pattern and texture are passed The image is obtained by contour recognition.
A method for fabric identification according to claim 1 or 2, wherein said "recognizing said cloth feature based on a pre-generated cloth identifying neural network to generate a recognition result corresponding to said cloth to be identified "include:

Determining a data format corresponding to the pre-generated cloth identification neural network;

If the data format includes a binary data format, the cloth feature is converted into a binary data format and then input into the cloth recognition neural network to generate a recognition result corresponding to the cloth to be identified.
A fabric identification method according to any one of claims 1 to 3, wherein the cloth recognition neural network is trained by the cloth specimen data exceeding a certain threshold; wherein the cloth is Specimen data includes cloth features and recognition results.
The method for fabric identification according to claim 4, wherein the cloth specimen data comprises triple data; wherein the triple data comprises: source data, belonging to the same class as the source data Forward data, and reverse data of a different class from the source data;

The source data is data randomly acquired from the cloth sample data;

The forward data is data that is randomly acquired in the cloth sample data and is consistent with the recognition result of the source data;

The reverse data is data that is randomly acquired in the cloth sample data and is inconsistent with the recognition result of the source data.
A fabric identification method according to any one of claims 1 to 5, wherein the extracting the cloth features corresponding to the fabric to be identified in the image comprises:

Performing a binarization process on the image to obtain the image with a gray value;

And determining an image contour according to the gray value of each pixel in the image with the gray value, and extracting the cloth feature of the fabric to be identified according to the image contour.
A fabric identification method according to any one of claims 1 to 5, wherein the extracting the cloth features corresponding to the fabric to be identified in the image further comprises:

The image is identified using a binarized neural network to obtain a cloth feature of the fabric to be identified in the image.
A cloth identification device, comprising:

Obtaining a module configured to obtain an image of the fabric to be identified;

An extraction module configured to extract a cloth feature corresponding to the fabric to be identified in the image;

The identification module is configured to generate a recognition result corresponding to the cloth to be identified based on identifying the cloth feature by a pre-generated cloth identification neural network.
A fabric identifying apparatus according to claim 8, wherein said fabric features comprise: a color of the cloth, and/or a pattern of the cloth, and/or a texture of the cloth; wherein said pattern and texture are passed The image is obtained by contour recognition.
The apparatus for fabric identification according to claim 8 or 9, wherein the identification module is configured to:

Determining a data format corresponding to the pre-generated cloth identification neural network;

If the data format includes a binary data format, the cloth feature is converted into a binary data format and then input into the cloth recognition neural network to generate a recognition result corresponding to the cloth to be identified.
A fabric identifying apparatus according to any one of claims 8 to 10, wherein the cloth identifying neural network is obtained by training cloth specimen data exceeding a certain threshold; wherein the cloth Specimen data includes cloth features and recognition results.
A fabric identifying apparatus according to claim 11, wherein:

The cloth specimen data includes triple data; wherein the triple data includes: source data, forward data belonging to the same class as the source data, and a reverse of a different class from the source data data;

The source data is data randomly acquired from the cloth sample data;

The forward data is data that is randomly acquired in the cloth sample data and is consistent with the recognition result of the source data;

The reverse data is data that is randomly acquired in the cloth sample data and is inconsistent with the recognition result of the source data.
A method of fabric identification according to any of claims 8-12, characterized in that

The extracting module is further configured to perform binarization processing on the image to obtain the image with gray value; and determine an image contour according to the gray value of each pixel in the image with the gray value. Extracting the cloth feature of the fabric to be identified according to the image contour.
A method of fabric identification according to any of claims 8-12, characterized in that

The extraction module is further configured to identify the image by using a binarized neural network to obtain a cloth feature of the cloth to be identified in the image.
An electronic device, comprising: a processor, a memory, a bus, and a communication module;

The processor, the communication module, and the memory are connected by the bus;

The memory is configured to store a program;

The processor is configured to perform the method of fabric identification as claimed in any of claims 1-7 by invoking a program stored in the memory.
A computer readable storage medium having processor-executable non-volatile program code, wherein the program code causes the processor to perform cloth identification as claimed in any of claims 1-7 Methods.