{"id":4120,"date":"2023-01-22T20:37:00","date_gmt":"2023-01-22T20:37:00","guid":{"rendered":"https:\/\/campus.hesge.ch\/blog-master-is\/?p=4120"},"modified":"2023-10-01T13:56:46","modified_gmt":"2023-10-01T13:56:46","slug":"dis-ia-dessine-moi-un-mouton","status":"publish","type":"post","link":"https:\/\/campus.hesge.ch\/blog-master-is\/dis-ia-dessine-moi-un-mouton\/","title":{"rendered":"\u00abDis IA, dessine-moi un mouton!\u00bb"},"content":{"rendered":"\n<p>En ao\u00fbt 2022 une r\u00e9v\u00e9lation bouleverse les mondes de l\u2019art et du\u00a0<a>num\u00e9rique :<\/a>\u00a0l\u2019IA\u00a0<a href=\"https:\/\/www.moka.mag.com\/articles\/midjourney\">MidJourney<\/a>\u00a0gagne le prix des<a>\u00a0\u00ab artsnum\u00e9riques \u00bb<\/a>\u00a0avec une \u0153uvre\u00a0<a>in\u00e9dite !<\/a>\u00a0Cela prouve que l\u2019IA est capable de cr\u00e9er des dessins ultrar\u00e9alistes. Comment l\u2019IA peut-elle copier des dessins et imaginer des cr\u00e9ations artistiques\u00a0<a>originales ?<\/a>\u00a0Quelles technologies utilise l\u2019IA pour\u00a0<a>dessiner ?<\/a>\u00a0C\u2019est le sujet de ce billet.<\/p>\n\n\n\n<p><strong>IA qui es-tu?<br><\/strong>L\u2019<strong>intelligence artificielle (IA)<\/strong>&nbsp;utilise la technologie du&nbsp;<strong>Machine Learning<\/strong>&nbsp;qui permet aux ordinateurs d\u2019apprendre par eux-m\u00eames \u00e0 partir de donn\u00e9es fournies en input. L\u2019IA utilise le&nbsp;<strong>Deep Learning<\/strong>&nbsp;pour cr\u00e9er des r\u00e9seaux de neurones artificiels&nbsp;<a>(\u00ab&nbsp;<strong>Neural<\/strong><\/a><strong>&nbsp;Nets&nbsp;<\/strong>\u00bb) capables d\u2019apprentissage. Bas\u00e9s sur la structure des neurones humains les Neural Nets sont utilis\u00e9s pour traiter des donn\u00e9es et r\u00e9soudre des probl\u00e8mes (<a href=\"https:\/\/www.researchgate.net\/publication\/363372484_Image_inpainting_based_on_deep_learning_A_review\">Zhang &amp; al. 2022<\/a>).<\/p>\n\n\n\n<figure class=\"wp-block-image size-large\"><a href=\"http:\/\/www.ibm.com\/\"><img fetchpriority=\"high\" decoding=\"async\" width=\"428\" height=\"196\" src=\"https:\/\/campus.hesge.ch\/blog-master-is\/wp-content\/uploads\/2023\/09\/image.png\" alt=\"\" class=\"wp-image-4716\" srcset=\"https:\/\/campus.hesge.ch\/blog-master-is\/wp-content\/uploads\/2023\/09\/image.png 428w, https:\/\/campus.hesge.ch\/blog-master-is\/wp-content\/uploads\/2023\/09\/image-300x137.png 300w\" sizes=\"(max-width: 428px) 100vw, 428px\" \/><\/a><figcaption>Figure&nbsp;1&nbsp;<a href=\"http:\/\/www.ibm.com\/\">KAVLAKOGLU, 2020<\/a><\/figcaption><\/figure>\n\n\n\n<p><strong>Comment un Neural Net reconna\u00eet et reproduit des images?<\/strong><\/p>\n\n\n\n<p><strong>La composition d\u2019un Neural Net<\/strong><\/p>\n\n\n\n<p>Un Neural Net est compos\u00e9 de 3 couches (\u00ablayers\u00bb):<\/p>\n\n\n\n<ul class=\"wp-block-list\"><li>L\u2019<strong>input&nbsp;<\/strong>r\u00e9ceptionne les donn\u00e9es<\/li><li>Les&nbsp;<strong>hidden layers&nbsp;<\/strong>traitent les donn\u00e9es<\/li><li>L\u2019<strong>output&nbsp;<\/strong>les transcrit<\/li><\/ul>\n\n\n\n<figure class=\"wp-block-image size-large\"><a href=\"http:\/\/www.ibm.com\/\"><img decoding=\"async\" width=\"1024\" height=\"728\" src=\"https:\/\/campus.hesge.ch\/blog-master-is\/wp-content\/uploads\/2022\/11\/image-3-1024x728.png\" alt=\"\" class=\"wp-image-4126\" srcset=\"https:\/\/campus.hesge.ch\/blog-master-is\/wp-content\/uploads\/2022\/11\/image-3-1024x728.png 1024w, https:\/\/campus.hesge.ch\/blog-master-is\/wp-content\/uploads\/2022\/11\/image-3-300x213.png 300w, https:\/\/campus.hesge.ch\/blog-master-is\/wp-content\/uploads\/2022\/11\/image-3-768x546.png 768w, https:\/\/campus.hesge.ch\/blog-master-is\/wp-content\/uploads\/2022\/11\/image-3.png 1309w\" sizes=\"(max-width: 1024px) 100vw, 1024px\" \/><\/a><figcaption><em>Figure 2<\/em> : <a href=\"http:\/\/www.ibm.com\/\">KAVLAKOGLU, 2020<\/a><\/figcaption><\/figure>\n\n\n\n<p>Un Neural Net est&nbsp;<em>simple&nbsp;<\/em>s\u2019il contient un seul hidden layer; il est&nbsp;<em>complexe&nbsp;<\/em>s\u2019il comporte entre 3 et 9 hidden layers (<a href=\"https:\/\/ibm.com\/cloud\/blog\/ai-vs-machine-learning-vs-deep-learning-vs-neural-networks\">Kavlakoglu 2020<\/a>;&nbsp;<a href=\"https:\/\/downloads.hindawi.com\/journals\/sp\/2022\/9691331.pdf\">Peng 2022<\/a>). Dans un Neural Net complexe l\u2019output d\u2019un hidden layer devient l\u2019input du hidden layer suivant, et ainsi de suite jusqu\u2019\u00e0 ce que l\u2019information traverse tous les hidden layers et arrive \u00e0 l\u2019output o\u00f9 elle est retranscrite.<\/p>\n\n\n\n<p><strong>Le Neural Net \u00e0 l\u2019\u00e9cole de l\u2019IA<\/strong><\/p>\n\n\n\n<p>Comme l\u2019enfant apprend le dessin, le Neural Net apprend \u00e0 identifier les caract\u00e9ristiques (\u00abpatterns\u00bb ou \u00abfeatures\u00bb: <a href=\"https:\/\/downloads.hindawi.com\/journals\/sp\/2022\/9691331.pdf\">Peng 2022<\/a>;&nbsp;<a href=\"https:\/\/www.researchgate.net\/publication\/361437472_Drawing_Conversations_Mediated_by_AI\">Yurman 2022<\/a>) d\u2019un dataset d\u2019images. Le Neural Net d\u00e9compose l\u2019image en patterns (points, traits, courbes) cod\u00e9s en pixel dans les hidden layers. C\u2019est en combinant les patterns comme des pi\u00e8ces de puzzle, que le Neural Net est capable de composer un dessin.<\/p>\n\n\n\n<figure class=\"wp-block-image size-large\"><a href=\"https:\/\/www.researchgate.net\/publication\/360440830_Cleanup_Sketched_Drawings_Deep_Learning-Based_Model\"><img decoding=\"async\" width=\"1018\" height=\"330\" src=\"https:\/\/campus.hesge.ch\/blog-master-is\/wp-content\/uploads\/2022\/11\/Capture-d\u2019\u00e9cran-2022-11-19-\u00e0-20.44.39.png\" alt=\"\" class=\"wp-image-4127\" srcset=\"https:\/\/campus.hesge.ch\/blog-master-is\/wp-content\/uploads\/2022\/11\/Capture-d\u2019\u00e9cran-2022-11-19-\u00e0-20.44.39.png 1018w, https:\/\/campus.hesge.ch\/blog-master-is\/wp-content\/uploads\/2022\/11\/Capture-d\u2019\u00e9cran-2022-11-19-\u00e0-20.44.39-300x97.png 300w, https:\/\/campus.hesge.ch\/blog-master-is\/wp-content\/uploads\/2022\/11\/Capture-d\u2019\u00e9cran-2022-11-19-\u00e0-20.44.39-768x249.png 768w\" sizes=\"(max-width: 1018px) 100vw, 1018px\" \/><\/a><figcaption>Une image d\u00e9compos\u00e9e en pixels (Mohammed &amp; al. 2022)<\/figcaption><\/figure>\n\n\n\n<p>Exemple: entrainons un Neural Net \u00e0 dessiner un chat.<\/p>\n\n\n\n<p><strong>Etape 1&nbsp;<\/strong>\u00ab<em>supervised learning<\/em>\u00bb : on fournit au Neural Net un dataset d\u2019entrainement avec des dessins de chats. Au d\u00e9but de l\u2019apprentissage le Neural Net aveugle ne voit pas les patterns. Pour comprendre les dessins, le Neural Net&nbsp;<strong>d\u00e9compose en pixels&nbsp;<\/strong>les patterns des dessins et les m\u00e9morise dans ses hidden layers (<a href=\"https:\/\/downloads.hindawi.com\/journals\/sp\/2022\/9691331.pdf\">Peng 2022<\/a>;&nbsp;<a href=\"https:\/\/www.researchgate.net\/publication\/361437472_Drawing_Conversations_Mediated_by_AI\">Yurman&nbsp;2022<\/a>).<\/p>\n\n\n\n<p><strong>Etape 2<\/strong>: on fournit au Neural Net un 2e dataset avec des dessins diff\u00e9rents du dataset d\u2019entrainement. Le mod\u00e8le fait une&nbsp;<strong>pr\u00e9diction<\/strong>, et associe \u00e0 chaque pattern un \u00ab<strong>poids<\/strong>\u00bb qui s\u2019alourdit ou s\u2019all\u00e8ge selon que le pattern appris par le mod\u00e8le est correctement identifi\u00e9 (<a href=\"https:\/\/ibm.com\/cloud\/blog\/ai-vs-machine-learning-vs-deep-learning-vs-neural-networks\">Kavlakoglu 2020<\/a>).<\/p>\n\n\n\n<p><strong>Etape 3<\/strong>: le Neural Net&nbsp;<strong>compare&nbsp;<\/strong>sa pr\u00e9diction avec les dessins d\u2019input. Si la pr\u00e9diction est juste, le mod\u00e8le a identifi\u00e9 les bons patterns et renvoie un dessin coh\u00e9rent. Si la pr\u00e9diction est fausse, le mod\u00e8le corrige son erreur en modifiant le \u00abpoids\u00bb de chaque pattern. Les poids sont comparables aux muscles du corps: un Neural Net bien entrain\u00e9 a des muscles forts et des poids bien calibr\u00e9s qui identifient correctement chaque pattern d\u2019un dessin. Donc une bonne pr\u00e9diction augmente les poids, et une mauvaise pr\u00e9diction les diminue.<\/p>\n\n\n\n<p>Plus le Neural Net s\u2019entraine, pr\u00e9dit et se corrige, plus il apprend et s\u2019am\u00e9liore. Il devient capable de traiter de nouvelles donn\u00e9es et cr\u00e9er des oeuvres in\u00e9dites comme le fait l\u2019IA <a href=\"https:\/\/www.midjourney.com\/\">MidJourney<\/a>.<\/p>\n\n\n\n<p><strong>Quels composants des Neural Nets permettent \u00e0 l\u2019IA de dessiner?<br><\/strong>Nous avons vu le principe d\u2019un r\u00e9seau de neurones simple, comme l\u2019IA <a href=\"https:\/\/quickdraw.withgoogle.com\/\">QuickDraw<\/a> qui se nourrit des dessins propos\u00e9s par l\u2019humain pour pr\u00e9dire l\u2019objet dessin\u00e9. Mais l\u2019IA peut aussi utiliser un GAN<strong>\u00abGenerative Adversarial Net\u00bb&nbsp;<\/strong>qui assemble plusieurs r\u00e9seaux de neurones mis en concurrence.<\/p>\n\n\n\n<figure class=\"wp-block-image size-large\"><a href=\"https:\/\/www.researchgate.net\/publication\/363372484_Image_inpainting_based_on_deep_learning_A_review\"><img loading=\"lazy\" decoding=\"async\" width=\"745\" height=\"1024\" src=\"https:\/\/campus.hesge.ch\/blog-master-is\/wp-content\/uploads\/2022\/11\/Capture-d\u2019\u00e9cran-2022-11-19-\u00e0-20.46.05-745x1024.png\" alt=\"\" class=\"wp-image-4128\" srcset=\"https:\/\/campus.hesge.ch\/blog-master-is\/wp-content\/uploads\/2022\/11\/Capture-d\u2019\u00e9cran-2022-11-19-\u00e0-20.46.05-745x1024.png 745w, https:\/\/campus.hesge.ch\/blog-master-is\/wp-content\/uploads\/2022\/11\/Capture-d\u2019\u00e9cran-2022-11-19-\u00e0-20.46.05-218x300.png 218w, https:\/\/campus.hesge.ch\/blog-master-is\/wp-content\/uploads\/2022\/11\/Capture-d\u2019\u00e9cran-2022-11-19-\u00e0-20.46.05-768x1056.png 768w, https:\/\/campus.hesge.ch\/blog-master-is\/wp-content\/uploads\/2022\/11\/Capture-d\u2019\u00e9cran-2022-11-19-\u00e0-20.46.05.png 816w\" sizes=\"(max-width: 745px) 100vw, 745px\" \/><\/a><figcaption>Illustration d&#8217;un GAN compos\u00e9 d&#8217;un CNN discriminateur et un DNN g\u00e9n\u00e9rateur       (Zhang &amp; al. 2023)<\/figcaption><\/figure>\n\n\n\n<p>Un GAN comporte un \u00ab<strong>Convoluted Neural Net<\/strong>\u00bb (CNN) confront\u00e9 \u00e0 un \u00ab<strong>Deconvolutional Neural Net<\/strong>\u00bb (DNN)<\/p>\n\n\n\n<p>\u2022 Le CNN classe des donn\u00e9es pixelis\u00e9es et reconnait des dessins. Exemple: dessins de chiens et chats.<\/p>\n\n\n\n<p>\u2022 Le DNN a l\u2019objectif inverse: il g\u00e9n\u00e8re des dessins depuis le mot \u00e9crit \u00abchat\u00bb ou \u00abchien\u00bb qu\u2019il a identifi\u00e9.<\/p>\n\n\n\n<p>Confront\u00e9 au CNN, le DNN s\u2019entraine a cr\u00e9er des dessins r\u00e9alistes, avant de soumettre son dessin au CNN qui d\u00e9cide si c\u2019est coh\u00e9rent ou non.<br>Ainsi, il semble que le secret de l\u2019IA soit inscrit dans le GAN dont la cr\u00e9ativit\u00e9 simule celle de l\u2019humain (<a href=\"https:\/\/www.researchgate.net\/publication\/361437472_Drawing_Conversations_Mediated_by_AI\">Yurman 2022<\/a>). Plus le GAN compare et m\u00e9morise des patterns de dessins, plus sa compr\u00e9hension s\u2019affine, lui permettant de mieux interpr\u00e9ter les dessins.<\/p>\n\n\n\n<p>Il s\u2019en inspire et s\u2019en abstrait pour cr\u00e9er des \u0153uvres d\u2019art in\u00e9dites. Ainsi l\u2019IA contribue aux processus cr\u00e9atifs \u00e0 travers des applications comme <a href=\"https:\/\/www.researchgate.net\/publication\/335694599_AI-Sketcher_A_Deep_Generative_Model_for_Producing_High-Quality_Sketches\">AI-Sketcher<\/a>&nbsp;(<a href=\"https:\/\/www.researchgate.net\/publication\/335694599_AI-Sketcher_A_Deep_Generative_Model_for_Producing_High-Quality_Sketches\">Cao &amp; al. 2020<\/a>),&nbsp;<a href=\"https:\/\/openai.com\/dall-e-2\/\">Dall-e<\/a>,&nbsp;<a href=\"https:\/\/www.midjourney.com\/\">MidJourney<\/a>,&nbsp;<a href=\"https:\/\/quickdraw.withgoogle.com\/\">QuickDraw<\/a>. Ces mod\u00e8les d\u2019AI apprennent des s\u00e9quences de patterns pr\u00e9cis qu\u2019ils ont int\u00e9gr\u00e9 dans leurs hidden layers. Ils peuvent ainsi g\u00e9n\u00e9rer un dessin artificiel de chat tr\u00e8s r\u00e9aliste.<\/p>\n\n\n\n<p><strong>CONCLUSION<\/strong><\/p>\n\n\n\n<p>Traitant une grande quantit\u00e9 de donn\u00e9es, les Neural Nets apprennent \u00e0 reconnaitre des \u00abpatterns\u00bb pour cr\u00e9er des dessins, ce qui permet \u00e0 l\u2019IA de d\u00e9velopper de surprenantes capacit\u00e9s, allant de l\u2019esquisse \u00e0 l\u2019\u0153uvre d\u2019art. L\u2019IA nous permet de dessiner, et nous soutient dans nos t\u00e2ches quotidiennes, notamment gr\u00e2ce aux assistants virtuels, bas\u00e9s sur l\u2019IA. Ainsi, pourrait-t-on imaginer qu\u2019un jour nos assistants SIRI, Alexa, OK Google soient capables d\u2019utiliser les Neural Network pour dessiner directement sur nos \u00e9crans, ce que l\u2019utilisateur lui demande? Et si un jour SIRI devenait capable de \u00abdessiner un mouton\u00bb?<\/p>\n\n\n\n<p><strong>BIBLIOGRAPHIE<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\"><li>AI vs. Machine Learning vs. Deep Learning vs. Neural Networks: What\u2019s the Difference?, KAVLAKOGLU, Eda [sans date]. [en ligne]. [Consult\u00e9 le 19 novembre 2022]. Disponible \u00e0 l\u2019adresse: https:\/\/www.ibm.com\/cloud\/blog\/ai-vs-machine-learning-vs-deep-learning-vs- neural-networks<\/li><li>CAO, Nan, YAN, Xin, SHI, Yang et CHEN, Chaoran, 2019. AI-Sketcher : A Deep Generative Model for Producing High-Quality Sketches.&nbsp;<em>Proceedings of the AAAI Conference on Artificial Intelligence<\/em>. [en ligne]. 17 juillet 2019. Vol. 33, no. 01, pp. 2564-2571. [Consult\u00e9 le 19 novembre 2022]. DOI 10.1609\/aaai.v33i01.33012564.<\/li><li>DALL\u00b7E 2, [sans date].&nbsp;<em>OpenAI<\/em>. [en ligne]. [Consult\u00e9 le 19 novembre 2022]. Disponible \u00e0 l\u2019adresse: https:\/\/openai.com\/dall-e-2\/<\/li><li>Midjourney, [sans date].&nbsp;<em>Midjourney<\/em>. [en ligne]. [Consult\u00e9 le 19 novembre 2022]. Disponible \u00e0 l\u2019adresse: https:\/\/www.midjourney.com\/<\/li><li>MidJourney | Comment l\u2019IA cr\u00e9e des images, [sans date].&nbsp;<em>MokaMag<\/em>. [en ligne]. [Consult\u00e9 le 19 novembre 2022]. Disponible \u00e0 l\u2019adresse: https:\/\/www.moka- mag.com\/articles\/midjourney<\/li><li>MOHAMMED, Amal Ahmed Hasan et CHEN, Jiazhou, 2022. Cleanup Sketched Drawings: Deep Learning-Based Model. ALGALIL, Fahd Abd (\u00e9d.),&nbsp;<em>Applied Bionics and Biomechanics<\/em>. [en ligne]. 6 mai 2022. Vol. 2022, pp. 1-17. [Consult\u00e9 le 17 novembre 2022]. DOI 10.1155\/2022\/2238077.<\/li><li>PENG, Yuanbo, 2022. Digital Recognition Methods Based on Deep Learning. LI, Lianhui (\u00e9d.),&nbsp;<em>Scientific Programming<\/em>. [en ligne]. 25 juillet 2022. Vol. 2022, pp. 1-12. [Consult\u00e9 le 17 novembre 2022]. DOI 10.1155\/2022\/9691331.<\/li><li>Quick, Draw!, [sans date]. [en ligne]. [Consult\u00e9 le 19 novembre 2022]. Disponible \u00e0 l\u2019adrese:https:\/\/quickdraw.withgoogle.com\/<\/li><li>YURMAN, Paulina et REDDY, Anuradha Venugopal, 2022. Drawing Conversations Mediated by AI. In:&nbsp;<em>Creativity and Cognition<\/em>. [en ligne]. Venice Italy: ACM. 20 juin 2022. pp. 56-70. [Consult\u00e9 le 19 novembre 2022]. ISBN 9781450393270. DOI 10.1145\/3527927.3531448.<\/li><li>ZHANG, Xiaobo, ZHAI, Donghai, LI, Tianrui, ZHOU, Yuxin et LIN, Yang, 2023. Image inpainting based on deep learning: A review.&nbsp;<em>Information Fusion<\/em>. [en ligne]. f\u00e9vrier 2023. Vol. 90, pp. 74-94. [Consult\u00e9 le 17 novembre 2022]. DOI 10.1016\/j.inffus.2022.08.033.<\/li><\/ul>\n","protected":false},"excerpt":{"rendered":"<p>En ao\u00fbt 2022 une r\u00e9v\u00e9lation bouleverse les mondes de l\u2019art et du\u00a0num\u00e9rique :\u00a0l\u2019IA\u00a0MidJourney\u00a0gagne le prix des\u00a0\u00ab artsnum\u00e9riques \u00bb\u00a0avec une \u0153uvre\u00a0in\u00e9dite !\u00a0Cela prouve que l\u2019IA est capable de cr\u00e9er des dessins ultrar\u00e9alistes. Comment l\u2019IA peut-elle copier des dessins et imaginer des &hellip; <a href=\"https:\/\/campus.hesge.ch\/blog-master-is\/dis-ia-dessine-moi-un-mouton\/\">Lire la suite\u00ad\u00ad<\/a><\/p>\n","protected":false},"author":119,"featured_media":4130,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[18,19,12,1],"tags":[58],"class_list":["post-4120","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-advanced-neural-net","category-machine-learning","category-reflexion-is","category-uncategorized","tag-advanced-neural-network"],"yoast_head":"<!-- This site is optimized with the Yoast SEO plugin v23.5 - https:\/\/yoast.com\/wordpress\/plugins\/seo\/ -->\n<title>\u00abDis IA, dessine-moi un mouton!\u00bb - Recherche d&#039;Id\u00e9eS<\/title>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/campus.hesge.ch\/blog-master-is\/dis-ia-dessine-moi-un-mouton\/\" \/>\n<meta property=\"og:locale\" content=\"fr_FR\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"\u00abDis IA, dessine-moi un mouton!\u00bb - Recherche d&#039;Id\u00e9eS\" \/>\n<meta property=\"og:description\" content=\"En ao\u00fbt 2022 une r\u00e9v\u00e9lation bouleverse les mondes de l\u2019art et du\u00a0num\u00e9rique :\u00a0l\u2019IA\u00a0MidJourney\u00a0gagne le prix des\u00a0\u00ab artsnum\u00e9riques \u00bb\u00a0avec une \u0153uvre\u00a0in\u00e9dite !\u00a0Cela prouve que l\u2019IA est capable de cr\u00e9er des dessins ultrar\u00e9alistes. Comment l\u2019IA peut-elle copier des dessins et imaginer des &hellip; Lire la suite\u00ad\u00ad\" \/>\n<meta property=\"og:url\" content=\"https:\/\/campus.hesge.ch\/blog-master-is\/dis-ia-dessine-moi-un-mouton\/\" \/>\n<meta property=\"og:site_name\" content=\"Recherche d&#039;Id\u00e9eS\" \/>\n<meta property=\"article:published_time\" content=\"2023-01-22T20:37:00+00:00\" \/>\n<meta property=\"article:modified_time\" content=\"2023-10-01T13:56:46+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/campus.hesge.ch\/blog-master-is\/wp-content\/uploads\/2022\/11\/Capture-d\u2019\u00e9cran-2022-11-19-\u00e0-21.12.44.png\" \/>\n\t<meta property=\"og:image:width\" content=\"950\" \/>\n\t<meta property=\"og:image:height\" content=\"930\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/png\" \/>\n<meta name=\"author\" content=\"Pamela von Scarpatetti\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"\u00c9crit par\" \/>\n\t<meta name=\"twitter:data1\" content=\"Pamela von Scarpatetti\" \/>\n\t<meta name=\"twitter:label2\" content=\"Dur\u00e9e de lecture estim\u00e9e\" \/>\n\t<meta name=\"twitter:data2\" content=\"6 minutes\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\/\/schema.org\",\"@graph\":[{\"@type\":\"WebPage\",\"@id\":\"https:\/\/campus.hesge.ch\/blog-master-is\/dis-ia-dessine-moi-un-mouton\/\",\"url\":\"https:\/\/campus.hesge.ch\/blog-master-is\/dis-ia-dessine-moi-un-mouton\/\",\"name\":\"\u00abDis IA, dessine-moi un mouton!\u00bb - Recherche d&#039;Id\u00e9eS\",\"isPartOf\":{\"@id\":\"https:\/\/campus.hesge.ch\/blog-master-is\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\/\/campus.hesge.ch\/blog-master-is\/dis-ia-dessine-moi-un-mouton\/#primaryimage\"},\"image\":{\"@id\":\"https:\/\/campus.hesge.ch\/blog-master-is\/dis-ia-dessine-moi-un-mouton\/#primaryimage\"},\"thumbnailUrl\":\"https:\/\/campus.hesge.ch\/blog-master-is\/wp-content\/uploads\/2022\/11\/Capture-d\u2019\u00e9cran-2022-11-19-\u00e0-21.12.44.png\",\"datePublished\":\"2023-01-22T20:37:00+00:00\",\"dateModified\":\"2023-10-01T13:56:46+00:00\",\"author\":{\"@id\":\"https:\/\/campus.hesge.ch\/blog-master-is\/#\/schema\/person\/c595b21e2f3bd4d24114157f1b2be2df\"},\"breadcrumb\":{\"@id\":\"https:\/\/campus.hesge.ch\/blog-master-is\/dis-ia-dessine-moi-un-mouton\/#breadcrumb\"},\"inLanguage\":\"fr-FR\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/campus.hesge.ch\/blog-master-is\/dis-ia-dessine-moi-un-mouton\/\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"fr-FR\",\"@id\":\"https:\/\/campus.hesge.ch\/blog-master-is\/dis-ia-dessine-moi-un-mouton\/#primaryimage\",\"url\":\"https:\/\/campus.hesge.ch\/blog-master-is\/wp-content\/uploads\/2022\/11\/Capture-d\u2019\u00e9cran-2022-11-19-\u00e0-21.12.44.png\",\"contentUrl\":\"https:\/\/campus.hesge.ch\/blog-master-is\/wp-content\/uploads\/2022\/11\/Capture-d\u2019\u00e9cran-2022-11-19-\u00e0-21.12.44.png\",\"width\":950,\"height\":930,\"caption\":\"Chat roux tenant un smartphone dans sa patte.\"},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\/\/campus.hesge.ch\/blog-master-is\/dis-ia-dessine-moi-un-mouton\/#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Accueil\",\"item\":\"https:\/\/campus.hesge.ch\/blog-master-is\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"\u00abDis IA, dessine-moi un mouton!\u00bb\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\/\/campus.hesge.ch\/blog-master-is\/#website\",\"url\":\"https:\/\/campus.hesge.ch\/blog-master-is\/\",\"name\":\"Recherche d&#039;Id\u00e9eS\",\"description\":\"Carnet de recherche des \u00e9tudiants du Master en sciences de l&#039;information et des donn\u00e9es (Information Science - IS) de la Haute \u00e9cole de gestion de Gen\u00e8ve\",\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\/\/campus.hesge.ch\/blog-master-is\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"fr-FR\"},{\"@type\":\"Person\",\"@id\":\"https:\/\/campus.hesge.ch\/blog-master-is\/#\/schema\/person\/c595b21e2f3bd4d24114157f1b2be2df\",\"name\":\"Pamela von Scarpatetti\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"fr-FR\",\"@id\":\"https:\/\/campus.hesge.ch\/blog-master-is\/#\/schema\/person\/image\/\",\"url\":\"https:\/\/secure.gravatar.com\/avatar\/e865cc48263ef6b4fda3369c4ff52d0f?s=96&d=mm&r=g\",\"contentUrl\":\"https:\/\/secure.gravatar.com\/avatar\/e865cc48263ef6b4fda3369c4ff52d0f?s=96&d=mm&r=g\",\"caption\":\"Pamela von Scarpatetti\"},\"url\":\"https:\/\/campus.hesge.ch\/blog-master-is\/author\/pamela\/\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"\u00abDis IA, dessine-moi un mouton!\u00bb - Recherche d&#039;Id\u00e9eS","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/campus.hesge.ch\/blog-master-is\/dis-ia-dessine-moi-un-mouton\/","og_locale":"fr_FR","og_type":"article","og_title":"\u00abDis IA, dessine-moi un mouton!\u00bb - Recherche d&#039;Id\u00e9eS","og_description":"En ao\u00fbt 2022 une r\u00e9v\u00e9lation bouleverse les mondes de l\u2019art et du\u00a0num\u00e9rique :\u00a0l\u2019IA\u00a0MidJourney\u00a0gagne le prix des\u00a0\u00ab artsnum\u00e9riques \u00bb\u00a0avec une \u0153uvre\u00a0in\u00e9dite !\u00a0Cela prouve que l\u2019IA est capable de cr\u00e9er des dessins ultrar\u00e9alistes. Comment l\u2019IA peut-elle copier des dessins et imaginer des &hellip; Lire la suite\u00ad\u00ad","og_url":"https:\/\/campus.hesge.ch\/blog-master-is\/dis-ia-dessine-moi-un-mouton\/","og_site_name":"Recherche d&#039;Id\u00e9eS","article_published_time":"2023-01-22T20:37:00+00:00","article_modified_time":"2023-10-01T13:56:46+00:00","og_image":[{"width":950,"height":930,"url":"https:\/\/campus.hesge.ch\/blog-master-is\/wp-content\/uploads\/2022\/11\/Capture-d\u2019\u00e9cran-2022-11-19-\u00e0-21.12.44.png","type":"image\/png"}],"author":"Pamela von Scarpatetti","twitter_card":"summary_large_image","twitter_misc":{"\u00c9crit par":"Pamela von Scarpatetti","Dur\u00e9e de lecture estim\u00e9e":"6 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"WebPage","@id":"https:\/\/campus.hesge.ch\/blog-master-is\/dis-ia-dessine-moi-un-mouton\/","url":"https:\/\/campus.hesge.ch\/blog-master-is\/dis-ia-dessine-moi-un-mouton\/","name":"\u00abDis IA, dessine-moi un mouton!\u00bb - Recherche d&#039;Id\u00e9eS","isPartOf":{"@id":"https:\/\/campus.hesge.ch\/blog-master-is\/#website"},"primaryImageOfPage":{"@id":"https:\/\/campus.hesge.ch\/blog-master-is\/dis-ia-dessine-moi-un-mouton\/#primaryimage"},"image":{"@id":"https:\/\/campus.hesge.ch\/blog-master-is\/dis-ia-dessine-moi-un-mouton\/#primaryimage"},"thumbnailUrl":"https:\/\/campus.hesge.ch\/blog-master-is\/wp-content\/uploads\/2022\/11\/Capture-d\u2019\u00e9cran-2022-11-19-\u00e0-21.12.44.png","datePublished":"2023-01-22T20:37:00+00:00","dateModified":"2023-10-01T13:56:46+00:00","author":{"@id":"https:\/\/campus.hesge.ch\/blog-master-is\/#\/schema\/person\/c595b21e2f3bd4d24114157f1b2be2df"},"breadcrumb":{"@id":"https:\/\/campus.hesge.ch\/blog-master-is\/dis-ia-dessine-moi-un-mouton\/#breadcrumb"},"inLanguage":"fr-FR","potentialAction":[{"@type":"ReadAction","target":["https:\/\/campus.hesge.ch\/blog-master-is\/dis-ia-dessine-moi-un-mouton\/"]}]},{"@type":"ImageObject","inLanguage":"fr-FR","@id":"https:\/\/campus.hesge.ch\/blog-master-is\/dis-ia-dessine-moi-un-mouton\/#primaryimage","url":"https:\/\/campus.hesge.ch\/blog-master-is\/wp-content\/uploads\/2022\/11\/Capture-d\u2019\u00e9cran-2022-11-19-\u00e0-21.12.44.png","contentUrl":"https:\/\/campus.hesge.ch\/blog-master-is\/wp-content\/uploads\/2022\/11\/Capture-d\u2019\u00e9cran-2022-11-19-\u00e0-21.12.44.png","width":950,"height":930,"caption":"Chat roux tenant un smartphone dans sa patte."},{"@type":"BreadcrumbList","@id":"https:\/\/campus.hesge.ch\/blog-master-is\/dis-ia-dessine-moi-un-mouton\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Accueil","item":"https:\/\/campus.hesge.ch\/blog-master-is\/"},{"@type":"ListItem","position":2,"name":"\u00abDis IA, dessine-moi un mouton!\u00bb"}]},{"@type":"WebSite","@id":"https:\/\/campus.hesge.ch\/blog-master-is\/#website","url":"https:\/\/campus.hesge.ch\/blog-master-is\/","name":"Recherche d&#039;Id\u00e9eS","description":"Carnet de recherche des \u00e9tudiants du Master en sciences de l&#039;information et des donn\u00e9es (Information Science - IS) de la Haute \u00e9cole de gestion de Gen\u00e8ve","potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/campus.hesge.ch\/blog-master-is\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"fr-FR"},{"@type":"Person","@id":"https:\/\/campus.hesge.ch\/blog-master-is\/#\/schema\/person\/c595b21e2f3bd4d24114157f1b2be2df","name":"Pamela von Scarpatetti","image":{"@type":"ImageObject","inLanguage":"fr-FR","@id":"https:\/\/campus.hesge.ch\/blog-master-is\/#\/schema\/person\/image\/","url":"https:\/\/secure.gravatar.com\/avatar\/e865cc48263ef6b4fda3369c4ff52d0f?s=96&d=mm&r=g","contentUrl":"https:\/\/secure.gravatar.com\/avatar\/e865cc48263ef6b4fda3369c4ff52d0f?s=96&d=mm&r=g","caption":"Pamela von Scarpatetti"},"url":"https:\/\/campus.hesge.ch\/blog-master-is\/author\/pamela\/"}]}},"_links":{"self":[{"href":"https:\/\/campus.hesge.ch\/blog-master-is\/wp-json\/wp\/v2\/posts\/4120","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/campus.hesge.ch\/blog-master-is\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/campus.hesge.ch\/blog-master-is\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/campus.hesge.ch\/blog-master-is\/wp-json\/wp\/v2\/users\/119"}],"replies":[{"embeddable":true,"href":"https:\/\/campus.hesge.ch\/blog-master-is\/wp-json\/wp\/v2\/comments?post=4120"}],"version-history":[{"count":7,"href":"https:\/\/campus.hesge.ch\/blog-master-is\/wp-json\/wp\/v2\/posts\/4120\/revisions"}],"predecessor-version":[{"id":4840,"href":"https:\/\/campus.hesge.ch\/blog-master-is\/wp-json\/wp\/v2\/posts\/4120\/revisions\/4840"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/campus.hesge.ch\/blog-master-is\/wp-json\/wp\/v2\/media\/4130"}],"wp:attachment":[{"href":"https:\/\/campus.hesge.ch\/blog-master-is\/wp-json\/wp\/v2\/media?parent=4120"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/campus.hesge.ch\/blog-master-is\/wp-json\/wp\/v2\/categories?post=4120"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/campus.hesge.ch\/blog-master-is\/wp-json\/wp\/v2\/tags?post=4120"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}