CN113158063A - Textile fabric personalized recommendation method based on image understanding - Google Patents

Textile fabric personalized recommendation method based on image understanding Download PDF

Info

Publication number
CN113158063A
CN113158063A CN202110510068.4A CN202110510068A CN113158063A CN 113158063 A CN113158063 A CN 113158063A CN 202110510068 A CN202110510068 A CN 202110510068A CN 113158063 A CN113158063 A CN 113158063A
Authority
CN
China
Prior art keywords
fabric
user
image
data
features
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202110510068.4A
Other languages
Chinese (zh)
Inventor
潘如如
何真真
张宁
向军
周建
王蕾
高卫东
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Jiangnan University
Original Assignee
Jiangnan University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Jiangnan University filed Critical Jiangnan University
Priority to CN202110510068.4A priority Critical patent/CN113158063A/en
Publication of CN113158063A publication Critical patent/CN113158063A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90—Details of database functions independent of the retrieved data types
    • G06F16/95—Retrieval from the web
    • G06F16/953—Querying, e.g. by the use of web search engines
    • G06F16/9535—Search customisation based on user profiles and personalisation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques
    • 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
    • G06N3/045—Combinations of networks
    • 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
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/40—Analysis of texture
    • G06T7/41—Analysis of texture based on statistical description of texture
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/90—Determination of colour characteristics
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/56—Extraction of image or video features relating to colour
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10024—Color image
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20081—Training; Learning
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30108—Industrial image inspection
    • G06T2207/30124—Fabrics; Textile; Paper

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • General Engineering & Computer Science (AREA)
  • Artificial Intelligence (AREA)
  • Databases & Information Systems (AREA)
  • Evolutionary Computation (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Computational Linguistics (AREA)
  • Molecular Biology (AREA)
  • Health & Medical Sciences (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Multimedia (AREA)
  • General Health & Medical Sciences (AREA)
  • Software Systems (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • Probability & Statistics with Applications (AREA)
  • Image Analysis (AREA)

Abstract

The invention discloses a method for personalized recommendation of textile fabrics based on image understanding, which comprises the following steps: describing a fabric text, and mainly considering three aspects of texture, color distribution and pattern shape; characterizing the fabric image in the existing data set; text features, image manual features and deep network neural features extracted from multi-source heterogeneous data are compressed, and the processing speed is reduced; the method comprises the steps of completing collection of user data by adopting a relational heterogeneous database, and completing construction of a user label system of a fabric recommendation system; using a multi-level user portrait model to complete the group subdivision of the user; mining rich conversion-type features in the database and generating accurate potential user vector representations; the exploration cyclic neural network is used for excavating the sequence relation of the session context and dynamically blending the session context into a propulsion system; and capturing the display or implicit evaluation of the user on the recommendation system through an interactive means, and finishing the continuous optimization of the recommendation system.

Description

Textile fabric personalized recommendation method based on image understanding
Technical Field
The invention belongs to the field of personalized recommendation of textile fabrics, and relates to a personalized recommendation method of textile fabrics based on image understanding.
Background
The application of the personalized textile fabric recommendation system can realize the recommendation of fabrics which are not expressed by a user and can optimize the selection range of the user. The textile garment fabric forms unique textures through fabric tissues, yarn types, yarn colors and patterns, and the fabric has a changeable appearance effect due to the fact that the fabric tissues, the yarn colors and the like are different. In the product design field, designers often describe products with words of some perceptual intentions, such as modern times, rhythm, lightness and the like, most of consumers mainly concentrate on visual effects in product cognition, and describe the products by directly adopting colors, patterns and the like, and a recommendation system can establish a bridge for communication between the designers and the consumers, so that the designers can understand correct requirements of consumption conveniently, and the products meeting the requirements of the consumers at present are designed. In addition, at present, the consumers are aware of the perceptual consumption, some consumers have clear product preferences when purchasing the clothing fabric, keywords can be directly searched when retrieving the products, but many consumers cannot accurately describe the product preferences, a fabric personalized recommendation system is established, the product preferences implicit by the consumers can be extracted according to the past browsing information of the consumers, the characteristics of the products are analyzed, and the personalized products meeting the consumer preferences are recommended according to the product characteristics.
Disclosure of Invention
Based on the above, the invention aims to provide a convenient, quick and accurate textile fabric personalized recommendation method based on image understanding.
Based on the above purpose, the invention provides an image understanding-based textile fabric personalized recommendation method, which comprises the following steps:
step 1: describing a fabric text, and mainly considering three aspects of texture, color distribution and pattern shape;
step 2: representing the fabric image in the existing data set, and representing the fabric by using the marking information of the fabric style, the pattern, the organizational structure, the pattern forming mode and other visual angles;
and step 3: compressing text features, image manual features and deep network neural features extracted from multi-source heterogeneous data, and reducing the processing speed;
and 4, step 4: the method comprises the steps of completing collection of user data by adopting a relational heterogeneous database, realizing and playing a user portrait technology, and completing construction of a user tag system of a material recommendation system;
and 5: using a multi-level user portrait model to complete the group subdivision of the user;
step 6: mining rich conversion-type features in the database and generating accurate potential user vector representations;
and 7: the exploration cyclic neural network is used for excavating the sequence relation of the session context and dynamically blending the session context into a propulsion system;
and 8: and capturing the display or implicit evaluation of the user on the recommendation system through an interactive means, and finishing the continuous optimization of the recommendation system.
Preferably, the low-order features such as series colors, textures and the like and the medium-order features extracted from a deep network are used as series features for representing the fabric;
preferably, the fabric is understood and characterized by the content of the depth exploration combined image and the semantic label together;
preferably, the deep learning model is integrated with extensive multi-element heterogeneous data, so that the fabric is represented more abstractly and more densely in a deep layer;
preferably, an interactive mode is adopted, and basic information of a user is combined;
preferably, the background analyzes the behavior information of the user in real time, effectively portrays the user portrait and carries out dynamic updating, and mines the motivation and preference of the user;
preferably, the session context information is analyzed from a plurality of visual angles and a personalized recommendation mechanism is dynamically merged;
preferably, the automatic adjustment of the weights of different factors is realized by matching with a graph attention model;
preferably, deep learning technologies such as reinforcement learning and transfer learning are adopted, nonlinear structural features between a user and the fabric are mined, and a complete personalized dynamic recommendation system is constructed;
preferably, the active feedback opinions of the user can directly reflect the search object and motivation of the user;
preferably, feedback of the user on the recommendation result is collected through the interactive system, a satisfaction degree model is constructed to quantize active feedback content, and a user portrait model or a sequencing mechanism is actively optimized so as to optimize the whole recommendation system;
the invention has the beneficial effects that:
the textile fabric personalized recommendation method based on image understanding provided by the invention comprises the steps of establishing a database of fabric images as a data set for fabric personalized recommendation technology research, establishing a knowledge system for understanding the fabric images from the viewpoints of fabric style, pattern, texture, color composition and the like, and constructing an algorithm model according to the established knowledge system to represent and abstract the fabric images; behavior data of a user in the system is stored in a network log mode, the data is analyzed to obtain complete data, the data is divided into characteristic fields which are analyzed by using the data, useless data are eliminated, a label is marked for the user, a user portrait of a fabric recommendation system is constructed, multi-level user portraits are adopted to complete group subdivision for the user, the search requirement of the user on fabrics is more comprehensive, the occupation ratio of the labels of each level is calculated, and the implicit preference of the user is deeply mined; by adopting a dynamic session recommendation method of the RNN and a fabric recommendation method of adaptive learning, rich transformation characteristics in projects can be better mined and accurate potential user vector representation can be generated by matching the RNN, and for the adaptive recommendation method, deep reinforcement learning and transfer learning are organically combined, and long-term or short-term preference of users is learned by adopting a generative antagonistic neural network; and finally, migrating knowledge in the auxiliary domain to the target domain by using a hybrid heterogeneous adaptive feedback mechanism, so as to realize better optimization of the recommendation system and enhance the controllability of the recommendation result.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in detail with reference to specific embodiments below.
The embodiment of the invention provides a textile fabric personalized recommendation method based on image understanding, which comprises the following steps of:
describing a fabric text, and mainly considering three aspects of texture, color distribution and pattern shape;
representing the fabric image in the existing data set, and representing the fabric by using the marking information of the fabric style, the pattern, the organizational structure, the pattern forming mode and other visual angles;
compressing text features, image manual features and deep network neural features extracted from multi-source heterogeneous data, and reducing the processing speed;
the method comprises the steps of completing collection of user data by adopting a relational heterogeneous database, realizing and playing a user portrait technology, and completing construction of a user tag system of a material recommendation system;
using a multi-level user portrait model to complete the group subdivision of the user;
mining rich conversion-type features in the database and generating accurate potential user vector representations;
the exploration cyclic neural network is used for excavating the sequence relation of the session context and dynamically blending the session context into a propulsion system;
and capturing the display or implicit evaluation of the user on the recommendation system through an interactive means, and finishing the continuous optimization of the recommendation system.
The method of the embodiment comprises the following steps:
step 101: the fabric text is described, and three aspects of texture, color distribution and pattern shape are mainly considered.
In the step, information of the fabric image including the product category, the attribute, the color, the texture period, the texture complexity and the like is extracted from the multi-element heterogeneous data such as the production process sheet, the product label, the product image and the like, and the texture and the color of the fabric image are globally and comprehensively characterized.
Step 102: and characterizing fabric images in the existing data set, and characterizing the fabric by using the marking information of the fabric style, the pattern, the organizational structure, the pattern forming mode and other visual angles.
In the step, the color moment, the gray level co-occurrence matrix, the ORB characteristic descriptor, the wavelet descriptor, the LBP characteristic descriptor and the like are adopted to carry out manual characteristic description on the fabric image; based on the fabric description of the low-layer characteristics, yarns with different colors are generally arranged periodically to form different patterns and patterns, the dominant color of the fabric image is grasped, and the color moment and the dominant color are combined to represent the color characteristics of the fabric image.
The fabric image representation is learned by using a multi-view guide model, the fabric representation is carried out by using a labeling information guide model of view angles such as fabric style, pattern, organizational structure and pattern forming mode, loss is caused among different view angles of the fabric, and weighting summation is carried out on the fabric by using a formula:
Figure BDA0003060001540000041
Figure BDA0003060001540000042
step 103: compressing text features, image manual features and deep network neural features extracted from multi-source heterogeneous data, and reducing the processing speed;
the production process generally stores the fabric in a text form, analyzes text parameters and integrates the extracted visual characteristics into the fabric to jointly represent the fabric.
Text features, image manual features and deep neural network features extracted from multi-source heterogeneous data have certain differences in the aspects of feature dimensions, expression forms and the like, and the method of Hash coding, product quantization and the like and the image features are adopted for compression, so that the processing speed is reduced.
And (4) performing feature fusion and abstraction by using the DBF of the stack RBM, and extracting abstract high-level semantic features.
Step 104: and the relational heterogeneous database is adopted to complete the acquisition of user data, realize and exert the user portrait technology, and complete the construction of a user label system of the material recommendation system.
In the step, behavior data of a user in the system is stored in a network log mode, the data is sorted to obtain complete data, the data is divided into characteristic fields which are analyzed by using the data, the data characteristic fields are cleaned, and useless data are eliminated.
Dynamically quantifying the behavior characteristics of the user using attrition active learning, such as the formula:
Figure BDA0003060001540000051
constructing a user label system of the fabric recommendation system by using the probability C-mean clustering PCM, and adopting a fuzzy clustering objective function:
Figure BDA0003060001540000052
step 105: using a multi-level user portrait model to complete the group subdivision of the user;
based on the step 104, a user portrait of the fabric recommendation system is constructed, a multi-level user portrait model is adopted, group subdivision of users is completed, the searching requirements of the users on the fabric are displayed more comprehensively, the proportion of labels in each level is calculated based on an algorithm of graph attention, and implicit preference of the users is deeply mined.
Step 106: mining rich conversion-type features in the database and generating accurate potential user vector representations;
in this step, the session context of the user and the system is analyzed from multiple perspectives and dimensions and merged into a recommendation system mechanism, rich conversion-type features in the database are better mined in cooperation with the RNN, and accurate potential user vector representations are generated.
The RNN can be used for automatically extracting the characteristics of the conversation graph from the rich relationships among the nodes to perform hidden vector representation of each item, each conversation utilizes an attention mechanism to combine the overall security with the current preference to represent, and recommendation score values of all candidate items are calculated.
Step 107: the recurrent neural network excavates the sequence relation of the session context and is dynamically merged into a propulsion system;
adopting an attention mechanism, organically combining deep reinforcement learning and transfer learning, when carrying out personalized recommendation on a new user, if the new user belongs to the existing category, calculating according to a model corresponding to the existing category, if the new user belongs to the new category, starting the transfer learning mechanism, and adopting the following formula:
Figure BDA0003060001540000061
step 108: and capturing the display or implicit evaluation of the user on the recommendation system through an interactive means, and finishing the continuous optimization of the recommendation system.
Based on the steps, a hybrid heterogeneous adaptive feedback optimization mechanism is used, implicit feedback data of a user are used as an auxiliary domain, feedback data are displayed as a target domain, knowledge in the auxiliary domain is migrated to the target domain, better optimization of the recommendation system is achieved, and controllability of a recommendation result is enhanced. The objective function using the migration method is:
Figure BDA0003060001540000062
the constructed satisfaction model fusing the active decision and the implicit feedback of the user can be quantized.

Claims (7)

1. The method for recommending the textile fabric individuation based on image understanding is characterized by comprising the following steps of:
step 1: describing a fabric text, and considering three aspects of texture, color distribution and pattern shape;
extracting information including product types, attributes, colors, texture periods and texture complexity fabric images from the multi-element heterogeneous data, and performing global and comprehensive characterization on the textures and colors of the fabric images;
step 2: representing the fabric image in the existing data set, and representing the fabric by using the marking information of the fabric style, the pattern, the organizational structure, the pattern forming mode and other visual angles;
and step 3: compressing text features, image manual features and deep network neural features extracted from multi-source heterogeneous data, and reducing the processing speed;
text features, image manual features and deep neural network features extracted from multi-source heterogeneous data are compressed by adopting methods such as Hash coding, product quantization and the like and image features, so that the processing speed is reduced; performing feature fusion and abstraction by using a DBF of a stack RBM, and extracting abstract high-level semantic features;
and 4, step 4: the method comprises the steps of completing collection of user data by adopting a relational heterogeneous database, realizing and playing a user portrait technology, and completing construction of a user tag system of a material recommendation system;
storing behavior data of a user in a system in a network log mode, sorting the data to obtain complete data, dividing the data into characteristic fields for data analysis, and cleaning the data characteristic fields to remove useless data;
dynamically quantifying the behavior characteristics of the user using attrition active learning, such as the formula:
Figure FDA0003060001530000011
constructing a user label system of the fabric recommendation system by using the probability C-mean clustering PCM, and adopting a fuzzy clustering objective function:
Figure FDA0003060001530000012
and 5: using a multi-level user portrait model to complete the group subdivision of the user;
step 6: mining rich conversion-type features in the database and generating accurate potential user vector representations;
and 7: the exploration cyclic neural network is used for excavating the sequence relation of the session context and dynamically blending the session context into a propulsion system;
and 8: and capturing the display or implicit evaluation of the user on the recommendation system through an interactive means, and finishing the continuous optimization of the recommendation system.
2. The method for personalized recommendation of textile fabric based on image understanding according to claim 1, characterized in that the step 8 is as follows:
by using a hybrid heterogeneous adaptive feedback optimization mechanism, implicit feedback data of a user is an auxiliary domain, feedback data is displayed as a target domain, and knowledge in the auxiliary domain is migrated into the target domain, so that better optimization of a recommendation system is realized, and controllability of a recommendation result is enhanced; the objective function using the migration method is:
Figure FDA0003060001530000021
the established satisfaction model fusing the active decision and the implicit feedback of the user can realize quantification.
3. The method for recommending textile fabric individuality based on image understanding according to claim 1 or 2, characterized in that in step 2, a fabric image is manually characterized by using a color moment, a gray level co-occurrence matrix, an ORB feature descriptor, a wavelet descriptor, and an LBP feature descriptor; based on the fabric description of the low-layer characteristics, yarns with different colors are generally arranged periodically to form different patterns and patterns, the dominant color of the fabric image is grasped, and the color moment and the dominant color are combined to represent the color characteristics of the fabric image;
the fabric image representation is learned by using a multi-view guide model, the fabric representation is carried out by using a labeling information guide model of view angles such as fabric style, pattern, organizational structure and pattern forming mode, loss is caused among different view angles of the fabric, and weighting summation is carried out on the fabric by using a formula:
Figure FDA0003060001530000022
Figure FDA0003060001530000023
4. the method for personalized recommendation of textile fabric based on image understanding according to claim 1 or 2, wherein in the step 6, the session context of the user and the system is analyzed from multiple perspectives and dimensions and is merged into a recommendation system mechanism, rich conversion-type features in the database are better mined by matching with RNN, and accurate potential user vector representation is generated.
5. The method for personalized recommendation of textile fabrics based on image understanding according to claim 1 or 2, characterized in that in step 7, an attention mechanism is adopted, and depth reinforcement learning and transfer learning are organically combined, when a new user is personalized recommended, if the new user belongs to an existing category, calculation is performed according to a model corresponding to the existing category, if the new user belongs to a new category, a transfer learning mechanism is started, and the following formula is adopted:
Figure FDA0003060001530000031
6. the method for personalized recommendation of textile fabric based on image understanding according to claim 3, wherein in the step 7, an attention mechanism is adopted, deep reinforcement learning and transfer learning are organically combined, when a new user is personalized recommended, if the new user belongs to an existing category, calculation is performed according to a model corresponding to the existing category, if the new user belongs to a new category, a transfer learning mechanism is started, and the following formula is adopted:
Figure FDA0003060001530000032
7. the method for personalized recommendation of textile fabric based on image understanding according to claim 4, wherein in the step 7, an attention mechanism is adopted, deep reinforcement learning and transfer learning are organically combined, when a new user is personalized recommended, if the new user belongs to an existing category, calculation is performed according to a model corresponding to the existing category, if the new user belongs to a new category, a transfer learning mechanism is started, and the following formula is adopted:
Figure FDA0003060001530000033
CN202110510068.4A 2021-05-11 2021-05-11 Textile fabric personalized recommendation method based on image understanding Pending CN113158063A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110510068.4A CN113158063A (en) 2021-05-11 2021-05-11 Textile fabric personalized recommendation method based on image understanding

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110510068.4A CN113158063A (en) 2021-05-11 2021-05-11 Textile fabric personalized recommendation method based on image understanding

Publications (1)

Publication Number Publication Date
CN113158063A true CN113158063A (en) 2021-07-23

Family

ID=76874343

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110510068.4A Pending CN113158063A (en) 2021-05-11 2021-05-11 Textile fabric personalized recommendation method based on image understanding

Country Status (1)

Country Link
CN (1) CN113158063A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114842043A (en) * 2022-07-04 2022-08-02 南通中豪超纤制品有限公司 Fabric style identification method and system based on image processing
CN117291111A (en) * 2023-11-24 2023-12-26 宁波博洋服饰集团有限公司 Digital fabric simulation optimization method combined with garment fabric cloud computing platform

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114842043A (en) * 2022-07-04 2022-08-02 南通中豪超纤制品有限公司 Fabric style identification method and system based on image processing
CN117291111A (en) * 2023-11-24 2023-12-26 宁波博洋服饰集团有限公司 Digital fabric simulation optimization method combined with garment fabric cloud computing platform
CN117291111B (en) * 2023-11-24 2024-04-05 宁波博洋服饰集团有限公司 Digital fabric simulation optimization method combined with garment fabric cloud computing platform

Similar Documents

Publication Publication Date Title
Al-Halah et al. Fashion forward: Forecasting visual style in fashion
CN110110181A (en) A kind of garment coordination recommended method based on user styles and scene preference
CN109299396A (en) Merge the convolutional neural networks collaborative filtering recommending method and system of attention model
CN106779087A (en) A kind of general-purpose machinery learning data analysis platform
CN113158063A (en) Textile fabric personalized recommendation method based on image understanding
Zhou et al. Fashion recommendations through cross-media information retrieval
CN107818105A (en) The recommendation method and server of application program
CN106897370B (en) Picture examination expert recommendation method based on Pearson similarity and FP-Growth
CN101706780A (en) Image semantic retrieving method based on visual attention model
CN108109055B (en) Cross-scene clothing retrieval method based on image rendering
CN101458732A (en) Knowledge application method orienting ornament innovation design
CN103984721A (en) Personalized book searching method based on interactive evolutionary optimization
CN106897776A (en) A kind of continuous type latent structure method based on nominal attribute
CN111061959A (en) Developer characteristic-based crowd-sourcing software task recommendation method
CN108804577A (en) A kind of predictor method of information label interest-degree
CN111311385B (en) Commodity recommendation grammar generation method and system based on commodity selling points
CN101021926A (en) Personal image design platform and method for designing by the same
CN110458292A (en) A kind of clothes recommended method based on expertise
CN112256895A (en) Fabric image retrieval method based on multi-task learning
Shah et al. Random patterns clothing image retrieval using convolutional neural network
CN109697257A (en) It is a kind of based on the network information retrieval method presorted with feature learning anti-noise
Podlasov et al. Japanese street fashion for young people: A multimodal digital humanities approach for identifying sociocultural patterns and trends
Li et al. Image emotion recognition based on deep neural network
Hu Analysis of clothing structure and management in clothing design oriented to market demand via recommendation algorithm
Varshney Design and Development of an Intelligent Tool to Assist Textile Designers

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