FR3114423B1 - KNOWLEDGE DISSEMINATION WITH PRESERVATION OF SEMANTIC RELATIONSHIPS FOR IMAGE TO IMAGE CONVERSION - Google Patents

KNOWLEDGE DISSEMINATION WITH PRESERVATION OF SEMANTIC RELATIONSHIPS FOR IMAGE TO IMAGE CONVERSION Download PDF

Info

Publication number
FR3114423B1
FR3114423B1 FR2009478A FR2009478A FR3114423B1 FR 3114423 B1 FR3114423 B1 FR 3114423B1 FR 2009478 A FR2009478 A FR 2009478A FR 2009478 A FR2009478 A FR 2009478A FR 3114423 B1 FR3114423 B1 FR 3114423B1
Authority
FR
France
Prior art keywords
image
learned
knowledge
model
preservation
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.)
Active
Application number
FR2009478A
Other languages
French (fr)
Other versions
FR3114423A1 (en
Inventor
Zeqi Li
Ruowei Jiang
Parham Aarabi
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.)
LOreal SA
Original Assignee
LOreal SA
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
Priority to FR2009478A priority Critical patent/FR3114423B1/en
Application filed by LOreal SA filed Critical LOreal SA
Priority to JP2022580295A priority patent/JP2023531263A/en
Priority to EP21737083.2A priority patent/EP4150528A1/en
Priority to PCT/EP2021/067860 priority patent/WO2022002943A1/en
Priority to CN202180046657.XA priority patent/CN116075830A/en
Priority to KR1020237002607A priority patent/KR20230027274A/en
Priority to US17/361,779 priority patent/US20220004803A1/en
Publication of FR3114423A1 publication Critical patent/FR3114423A1/en
Application granted granted Critical
Publication of FR3114423B1 publication Critical patent/FR3114423B1/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/088—Non-supervised learning, e.g. competitive learning
    • 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
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/047—Probabilistic or stochastic networks

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Health & Medical Sciences (AREA)
  • Computing Systems (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Molecular Biology (AREA)
  • Artificial Intelligence (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Health & Medical Sciences (AREA)
  • Probability & Statistics with Applications (AREA)
  • Image Analysis (AREA)
  • Processing Or Creating Images (AREA)

Abstract

Diffusion de connaissances avec préservation des relations sémantiques pour une conversion d'image à image L’invention concerne des générateurs basés sur un GAN utiles pour exécuter des conversions d'image à image. Des systèmes et des procédés définissent, par l'intermédiaire d'un conditionnement, un modèle de GAN appris qui présente un générateur appris qui est réduit à partir d'un modèle de GAN d’apprentissage (et d'un générateur) en utilisant la diffusion de connaissance. Une perte de diffusion de connaissance des relations sémantiques est utilisée pour transférer des connaissances sémantiques d'une couche intermédiaire du modèle d’apprentissage à une couche intermédiaire au modèle appris. Des générateurs appris ainsi définis sont stockés et exécutés par des dispositifs mobiles tels que des téléphones intelligents et des ordinateurs portables pour effectuer des expériences de réalité augmentée. Des effets sont simulés sur des images, comprenant des effets de maquillage, capillaires, d'ongles ainsi que de simulation d'âge. Figure pour l'abrégé : figure 11Knowledge dissemination with preservation of semantic relationships for image-to-image conversion The invention relates to GAN-based generators useful for performing image-to-image conversions. Systems and methods define, through conditioning, a learned GAN model that has a learned generator that is reduced from a learned GAN model (and generator) using the dissemination of knowledge. A loss of knowledge diffusion of semantic relations is used to transfer semantic knowledge from an intermediate layer of the learning model to an intermediate layer of the learned model. Learned generators thus defined are stored and executed by mobile devices such as smart phones and laptops to perform augmented reality experiences. Effects are simulated on images, including makeup, hair, nail, and age simulation effects. Figure for the abstract: Figure 11

FR2009478A 2020-06-29 2020-09-18 KNOWLEDGE DISSEMINATION WITH PRESERVATION OF SEMANTIC RELATIONSHIPS FOR IMAGE TO IMAGE CONVERSION Active FR3114423B1 (en)

Priority Applications (7)

Application Number Priority Date Filing Date Title
FR2009478A FR3114423B1 (en) 2020-09-18 2020-09-18 KNOWLEDGE DISSEMINATION WITH PRESERVATION OF SEMANTIC RELATIONSHIPS FOR IMAGE TO IMAGE CONVERSION
EP21737083.2A EP4150528A1 (en) 2020-06-29 2021-06-29 Semantic relation preserving knowledge distillation for image-to-image translation
PCT/EP2021/067860 WO2022002943A1 (en) 2020-06-29 2021-06-29 Semantic Relation Preserving Knowledge Distillation For Image-To-Image Translation
CN202180046657.XA CN116075830A (en) 2020-06-29 2021-06-29 Knowledge refinement of semantic relationship preservation in image-to-image conversion
JP2022580295A JP2023531263A (en) 2020-06-29 2021-06-29 Semantic Relations Maintaining Knowledge Distillation for Image-to-Image Transformation
KR1020237002607A KR20230027274A (en) 2020-06-29 2021-06-29 Distillation of Semantic Relationship Preservation Knowledge for Image-to-Image Transformation
US17/361,779 US20220004803A1 (en) 2020-06-29 2021-06-29 Semantic relation preserving knowledge distillation for image-to-image translation

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
FR2009478A FR3114423B1 (en) 2020-09-18 2020-09-18 KNOWLEDGE DISSEMINATION WITH PRESERVATION OF SEMANTIC RELATIONSHIPS FOR IMAGE TO IMAGE CONVERSION
FR2009478 2020-09-18

Publications (2)

Publication Number Publication Date
FR3114423A1 FR3114423A1 (en) 2022-03-25
FR3114423B1 true FR3114423B1 (en) 2023-07-14

Family

ID=74045635

Family Applications (1)

Application Number Title Priority Date Filing Date
FR2009478A Active FR3114423B1 (en) 2020-06-29 2020-09-18 KNOWLEDGE DISSEMINATION WITH PRESERVATION OF SEMANTIC RELATIONSHIPS FOR IMAGE TO IMAGE CONVERSION

Country Status (1)

Country Link
FR (1) FR3114423B1 (en)

Also Published As

Publication number Publication date
FR3114423A1 (en) 2022-03-25

Similar Documents

Publication Publication Date Title
KR102073351B1 (en) Server and operating method thereof
US20210035047A1 (en) Utilizing a machine learning model to determine anonymized avatars for employment interviews
WO2020029708A1 (en) Image processing method and apparatus, electronic device, storage medium, and program product
WO2018000270A1 (en) Personalized response generating method and system based on profile
US11769018B2 (en) System and method for temporal attention behavioral analysis of multi-modal conversations in a question and answer system
US11727717B2 (en) Data-driven, photorealistic social face-trait encoding, prediction, and manipulation using deep neural networks
CN108280451B (en) Semantic segmentation and network training method and device, equipment and medium
WO2020020041A1 (en) Sentence processing method and system and electronic device
US10909457B2 (en) Compound model scaling for neural networks
CN111666416B (en) Method and device for generating semantic matching model
WO2018000271A1 (en) Intention scene recognition method and system based on user portrait
CN110088835A (en) Use the blind source separating of similarity measure
US11769315B2 (en) Diagnostic tool for deep learning similarity models
CN110232920B (en) Voice processing method and device
FR3114423B1 (en) KNOWLEDGE DISSEMINATION WITH PRESERVATION OF SEMANTIC RELATIONSHIPS FOR IMAGE TO IMAGE CONVERSION
US20230153630A1 (en) Method and apparatus for training language model for multi-modal dialog
KR20210065629A (en) Chatbot integration agent platform system and service method thereof
CN110222166A (en) Question and answer service providing method, device, equipment, medium and immersion coaching system
CN110619602A (en) Image generation method and device, electronic equipment and storage medium
KR20230065339A (en) Model data processing method, device, electronic device and computer readable medium
JP2019185134A (en) Information processing device, learning method, and program
CN113553386A (en) Embedded representation model training method, knowledge graph-based question-answering method and device
US9299339B1 (en) Parsing rule augmentation based on query sequence and action co-occurrence
WO2021077333A1 (en) Simultaneous interpretation method and device, and storage medium
Richardson et al. Retrofitting realities: affordances and limitations in porting an interactive geospatial visualization from augmented to virtual reality

Legal Events

Date Code Title Description
PLFP Fee payment

Year of fee payment: 2

PLSC Publication of the preliminary search report

Effective date: 20220325

PLFP Fee payment

Year of fee payment: 3

PLFP Fee payment

Year of fee payment: 4