{"id":102454,"date":"2021-12-03T15:36:38","date_gmt":"2021-12-03T13:36:42","guid":{"rendered":"https:\/\/prohoster.info\/blog\/novosti-interneta\/hyperstyle-adaptacziya-sistemy-mashinnogo-obucheniya-stylegan-dlya-redaktirovaniya-izobrazhenij"},"modified":"2021-12-03T15:36:38","modified_gmt":"2021-12-03T13:36:42","slug":"hyperstyle-adaptacziya-sistemy-mashinnogo-obucheniya-stylegan-dlya-redaktirovaniya-izobrazhenij","status":"publish","type":"post","link":"https:\/\/prohoster.info\/fr\/blog\/news\/hyperstyle-adaptacziya-sistemy-mashinnogo-obucheniya-stylegan-dlya-redaktirovaniya-izobrazhenij","title":{"rendered":"HyperStyle \u2014 adaptation du syst\u00e8me d'apprentissage automatique StyleGAN pour l'\u00e9dition d'images","gt_translate_keys":[{"key":"rendered","format":"text"}]},"content":{"rendered":"

Un groupe de chercheurs de l'universit\u00e9 de Tel-Aviv a pr\u00e9sent\u00e9 HyperStyle, une version invers\u00e9e du syst\u00e8me d'apprentissage automatique StyleGAN2 d\u00e9velopp\u00e9 par NVIDIA, qui a \u00e9t\u00e9 retravaill\u00e9e pour recr\u00e9er les parties manquantes lors de l'\u00e9dition d'images r\u00e9elles. Le code est \u00e9crit en Python en utilisant le cadre PyTorch et est distribu\u00e9 sous la licence MIT. <\/p>\n

Alors que StyleGAN permet de synth\u00e9tiser de nouveaux visages r\u00e9alistes en d\u00e9finissant des param\u00e8tres tels que l'\u00e2ge, le sexe, la longueur des cheveux, le caract\u00e8re du sourire, la forme du nez, la couleur de la peau, les lunettes et l'angle de la photo, HyperStyle permet de modifier des param\u00e8tres similaires sur des photographies existantes, sans alt\u00e9rer leurs caract\u00e9ristiques distinctives et en pr\u00e9servant la reconnaissabilit\u00e9 du visage d'origine. Par exemple, \u00e0 l'aide d'HyperStyle, on peut simuler le vieillissement d'une personne sur une photo, changer la coiffure, ajouter des lunettes, une barbe ou une moustache, donner \u00e0 l'image l'apparence d'un personnage de dessin anim\u00e9 ou d'une peinture, ou modifier l'expression du visage pour la rendre triste ou joyeuse. De plus, le syst\u00e8me peut \u00eatre entra\u00een\u00e9 non seulement pour changer les visages des gens, mais aussi pour tout objet, comme pour l'\u00e9dition d'images de voitures.

\"HyperStyle<\/center> <\/p>\n

La m\u00e9thode propos\u00e9e vise \u00e0 r\u00e9soudre le probl\u00e8me de la reconstruction des parties manquantes d'une image lors de l'\u00e9dition. Dans les m\u00e9thodes pr\u00e9c\u00e9demment propos\u00e9es, le compromis entre reconstruction et \u00e9ditabilit\u00e9 \u00e9tait r\u00e9gl\u00e9 par un ajustement fin du g\u00e9n\u00e9rateur d'images pour substituer les parties de l'image cible lors de la recr\u00e9ation des zones \u00e9ditables initialement manquantes. Le principal inconv\u00e9nient de ces approches est la n\u00e9cessit\u00e9 d'une longue formation cibl\u00e9e du r\u00e9seau neuronal pour chaque image. <\/p>\n

La m\u00e9thode bas\u00e9e sur l'algorithme StyleGAN permet d'utiliser un mod\u00e8le standard, pr\u00e9alablement entra\u00een\u00e9 sur des collections d'images communes, pour g\u00e9n\u00e9rer des \u00e9l\u00e9ments caract\u00e9ristiques de l'image d'origine avec un niveau de fid\u00e9lit\u00e9 comparable \u00e0 celui des algorithmes n\u00e9cessitant un entra\u00eenement individuel du mod\u00e8le pour chaque image. Parmi les avantages de cette nouvelle m\u00e9thode, on note \u00e9galement la capacit\u00e9 \u00e0 modifier des images avec une performance proche du temps r\u00e9el.

\"HyperStyle<\/center> <\/p>\n

Des mod\u00e8les pr\u00e9entrain\u00e9s sont disponibles pour les visages humains, les machines et les animaux, bas\u00e9s sur des collections telles que Flickr-Faces-HQ (FFHQ, 70 000 images PNG haute qualit\u00e9 de visages humains), Stanford Cars (16 000 images de voitures) et AFHQ (photos d'animaux). Un outil d'entra\u00eenement de mod\u00e8les est \u00e9galement fourni, ainsi que des mod\u00e8les pr\u00e9entrain\u00e9s pr\u00eats \u00e0 l'emploi de codeurs et de g\u00e9n\u00e9rateurs standard. Par exemple, des g\u00e9n\u00e9rateurs sont disponibles pour cr\u00e9er des images dans le style Toonify, des personnages Pixar, pour dessiner des esquisses et m\u00eame pour styliser \u00e0 la mani\u00e8re des princesses des dessins anim\u00e9s Disney.

\"HyperStyle<\/center>
\"HyperStyle<\/center>
\"HyperStyle<\/center>
\"HyperStyle<\/center>
\n
Source : opennet.ru<\/a> <\/p>","protected":false,"gt_translate_keys":[{"key":"rendered","format":"html"}]},"excerpt":{"rendered":"

\u0413\u0440\u0443\u043f\u043f\u0430 \u0438\u0441\u0441\u043b\u0435\u0434\u043e\u0432\u0430\u0442\u0435\u043b\u0435\u0439 \u0438\u0437 \u0422\u0435\u043b\u044c-\u0410\u0432\u0438\u0432\u0441\u043a\u043e\u0433\u043e \u0443\u043d\u0438\u0432\u0435\u0440\u0441\u0438\u0442\u0435\u0442\u0430 \u043f\u0440\u0435\u0434\u0441\u0442\u0430\u0432\u0438\u043b\u0430 HyperStyle, \u0438\u043d\u0432\u0435\u0440\u0442\u0438\u0440\u043e\u0432\u0430\u043d\u043d\u044b\u0439 \u0432\u0430\u0440\u0438\u0430\u043d\u0442 \u0440\u0430\u0437\u0432\u0438\u0432\u0430\u0435\u043c\u043e\u0439 \u043a\u043e\u043c\u043f\u0430\u043d\u0438\u0435\u0439 NVIDIA \u0441\u0438\u0441\u0442\u0435\u043c\u044b \u043c\u0430\u0448\u0438\u043d\u043d\u043e\u0433\u043e \u043e\u0431\u0443\u0447\u0435\u043d\u0438\u044f StyleGAN2, \u043a\u043e\u0442\u043e\u0440\u044b\u0439 \u043f\u0435\u0440\u0435\u0440\u0430\u0431\u043e\u0442\u0430\u043d \u0434\u043b\u044f \u0432\u043e\u0441\u0441\u043e\u0437\u0434\u0430\u043d\u0438\u044f \u043d\u0435\u0434\u043e\u0441\u0442\u0430\u044e\u0449\u0438\u0445 \u0447\u0430\u0441\u0442\u0435\u0439 \u043f\u0440\u0438 \u0440\u0435\u0434\u0430\u043a\u0442\u0438\u0440\u043e\u0432\u0430\u043d\u0438\u0438 \u0440\u0435\u0430\u043b\u044c\u043d\u044b\u0445 \u0438\u0437\u043e\u0431\u0440\u0430\u0436\u0435\u043d\u0438\u0439. \u041a\u043e\u0434 \u043d\u0430\u043f\u0438\u0441\u0430\u043d \u043d\u0430 \u044f\u0437\u044b\u043a\u0435 Python c \u0438\u0441\u043f\u043e\u043b\u044c\u0437\u043e\u0432\u0430\u043d\u0438\u0435\u043c \u0444\u0440\u0435\u0439\u043c\u0432\u043e\u0440\u043a\u0430 PyTorch \u0438 \u0440\u0430\u0441\u043f\u0440\u043e\u0441\u0442\u0440\u0430\u043d\u044f\u0435\u0442\u0441\u044f \u043f\u043e\u0434 \u043b\u0438\u0446\u0435\u043d\u0437\u0438\u0435\u0439 MIT. \u0415\u0441\u043b\u0438 StyleGAN \u043f\u043e\u0437\u0432\u043e\u043b\u044f\u0435\u0442 \u0441\u0438\u043d\u0442\u0435\u0437\u0438\u0440\u043e\u0432\u0430\u0442\u044c \u0440\u0435\u0430\u043b\u0438\u0441\u0442\u0438\u0447\u043d\u043e \u0432\u044b\u0433\u043b\u044f\u0434\u044f\u0449\u0438\u0435 \u043d\u043e\u0432\u044b\u0435 \u043b\u0438\u0446\u0430 \u043b\u044e\u0434\u0435\u0439, \u0437\u0430\u0434\u0430\u0432\u0430\u044f \u0442\u0430\u043a\u0438\u0435 \u043f\u0430\u0440\u0430\u043c\u0435\u0442\u0440\u044b, \u043a\u0430\u043a \u0432\u043e\u0437\u0440\u0430\u0441\u0442, \u043f\u043e\u043b, […]<\/p>\n","protected":false,"gt_translate_keys":[{"key":"rendered","format":"html"}]},"author":1,"featured_media":102455,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[702],"tags":[],"class_list":["post-102454","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-news"],"aioseo_notices":[],"aioseo_head":"\n\t\t\n\t\n\t\n\t\n\t\n\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\n","aioseo_head_json":{"title":"\ud83e\udd47HyperStyle \u2014 adaptation du syst\u00e8me d'apprentissage automatique StyleGAN pour l'\u00e9dition d'images | ProHoster","description":"Un groupe de chercheurs de l'universit\u00e9 de Tel Aviv a pr\u00e9sent\u00e9 HyperStyle, une variante invers\u00e9e du syst\u00e8me d'apprentissage automatique StyleGAN2 d\u00e9velopp\u00e9 par NVIDIA, qui a \u00e9t\u00e9 retravaill\u00e9 pour recr\u00e9er les \u00e9l\u00e9ments manquants.","canonical_url":"https:\/\/prohoster.info\/fr\/blog\/news\/hyperstyle-adaptacziya-sistemy-mashinnogo-obucheniya-stylegan-dlya-redaktirovaniya-izobrazhenij","robots":"max-image-preview:large","keywords":"","webmasterTools":{"miscellaneous":""},"schema":null,"og:locale":"fr_FR","og:site_name":"ProHoster | \u041a\u0443\u043f\u0438\u0442\u044c \u043d\u0430\u0434\u0435\u0436\u043d\u044b\u0439 \u0445\u043e\u0441\u0442\u0438\u043d\u0433 \u0434\u043b\u044f \u0441\u0430\u0439\u0442\u043e\u0432 \u0441 \u0437\u0430\u0449\u0438\u0442\u043e\u0439 \u043e\u0442 DDoS, VPS VDS \u0441\u0435\u0440\u0432\u0435\u0440\u044b","og:type":"article","og:title":"\ud83e\udd47HyperStyle \u2014 \u0430\u0434\u0430\u043f\u0442\u0430\u0446\u0438\u044f \u0441\u0438\u0441\u0442\u0435\u043c\u044b \u043c\u0430\u0448\u0438\u043d\u043d\u043e\u0433\u043e \u043e\u0431\u0443\u0447\u0435\u043d\u0438\u044f StyleGAN \u0434\u043b\u044f \u0440\u0435\u0434\u0430\u043a\u0442\u0438\u0440\u043e\u0432\u0430\u043d\u0438\u044f \u0438\u0437\u043e\u0431\u0440\u0430\u0436\u0435\u043d\u0438\u0439 | ProHoster","og:description":"\u0413\u0440\u0443\u043f\u043f\u0430 \u0438\u0441\u0441\u043b\u0435\u0434\u043e\u0432\u0430\u0442\u0435\u043b\u0435\u0439 \u0438\u0437 \u0422\u0435\u043b\u044c-\u0410\u0432\u0438\u0432\u0441\u043a\u043e\u0433\u043e \u0443\u043d\u0438\u0432\u0435\u0440\u0441\u0438\u0442\u0435\u0442\u0430 \u043f\u0440\u0435\u0434\u0441\u0442\u0430\u0432\u0438\u043b\u0430 HyperStyle, \u0438\u043d\u0432\u0435\u0440\u0442\u0438\u0440\u043e\u0432\u0430\u043d\u043d\u044b\u0439 \u0432\u0430\u0440\u0438\u0430\u043d\u0442 \u0440\u0430\u0437\u0432\u0438\u0432\u0430\u0435\u043c\u043e\u0439 \u043a\u043e\u043c\u043f\u0430\u043d\u0438\u0435\u0439 NVIDIA \u0441\u0438\u0441\u0442\u0435\u043c\u044b \u043c\u0430\u0448\u0438\u043d\u043d\u043e\u0433\u043e \u043e\u0431\u0443\u0447\u0435\u043d\u0438\u044f StyleGAN2, \u043a\u043e\u0442\u043e\u0440\u044b\u0439 \u043f\u0435\u0440\u0435\u0440\u0430\u0431\u043e\u0442\u0430\u043d \u0434\u043b\u044f \u0432\u043e\u0441\u0441\u043e\u0437\u0434\u0430\u043d\u0438\u044f \u043d\u0435\u0434\u043e\u0441\u0442\u0430\u044e\u0449\u0438\u0445.","og:url":"https:\/\/prohoster.info\/fr\/blog\/news\/hyperstyle-adaptacziya-sistemy-mashinnogo-obucheniya-stylegan-dlya-redaktirovaniya-izobrazhenij","og:image":"https:\/\/prohoster.info\/wp-content\/uploads\/2021\/11\/logo-350.jpg","og:image:secure_url":"https:\/\/prohoster.info\/wp-content\/uploads\/2021\/11\/logo-350.jpg","og:image:width":350,"og:image:height":350,"article:published_time":"2021-12-03T13:36:42+00:00","article:modified_time":"2021-12-03T13:36:42+00:00","article:publisher":"https:\/\/www.facebook.com\/prohoster","article:author":"https:\/\/www.facebook.com\/prohoster"},"aioseo_meta_data":{"post_id":"102454","title":null,"description":null,"keywords":null,"keyphrases":null,"primary_term":null,"canonical_url":null,"og_title":null,"og_description":null,"og_object_type":"default","og_image_type":"default","og_image_url":null,"og_image_width":null,"og_image_height":null,"og_image_custom_url":null,"og_image_custom_fields":null,"og_video":null,"og_custom_url":null,"og_article_section":null,"og_article_tags":null,"twitter_use_og":false,"twitter_card":"default","twitter_image_type":"default","twitter_image_url":null,"twitter_image_custom_url":null,"twitter_image_custom_fields":null,"twitter_title":null,"twitter_description":null,"schema":{"blockGraphs":[],"customGraphs":[],"default":{"data":{"Article":[],"Course":[],"Dataset":[],"FAQPage":[],"Movie":[],"Person":[],"Product":[],"ProductReview":[],"Car":[],"Recipe":[],"Service":[],"SoftwareApplication":[],"WebPage":[]},"graphName":"","isEnabled":true},"graphs":[]},"schema_type":"default","schema_type_options":null,"pillar_content":false,"robots_default":true,"robots_noindex":false,"robots_noarchive":false,"robots_nosnippet":false,"robots_nofollow":false,"robots_noimageindex":false,"robots_noodp":false,"robots_notranslate":false,"robots_max_snippet":null,"robots_max_videopreview":null,"robots_max_imagepreview":"large","priority":null,"frequency":null,"local_seo":null,"seo_analyzer_scan_date":null,"breadcrumb_settings":null,"limit_modified_date":false,"reviewed_by":null,"ai":null,"created":"2021-12-03 13:37:36","updated":"2022-10-04 22:45:20","focus_keyword":null,"additional_keywords":null,"truseo_locale":null},"gt_translate_keys":[{"key":"link","format":"url"}],"_links":{"self":[{"href":"https:\/\/prohoster.info\/fr\/wp-json\/wp\/v2\/posts\/102454","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/prohoster.info\/fr\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/prohoster.info\/fr\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/prohoster.info\/fr\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/prohoster.info\/fr\/wp-json\/wp\/v2\/comments?post=102454"}],"version-history":[{"count":0,"href":"https:\/\/prohoster.info\/fr\/wp-json\/wp\/v2\/posts\/102454\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/prohoster.info\/fr\/wp-json\/wp\/v2\/media\/102455"}],"wp:attachment":[{"href":"https:\/\/prohoster.info\/fr\/wp-json\/wp\/v2\/media?parent=102454"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/prohoster.info\/fr\/wp-json\/wp\/v2\/categories?post=102454"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/prohoster.info\/fr\/wp-json\/wp\/v2\/tags?post=102454"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}