{"id":231621,"date":"2024-03-15T06:30:00","date_gmt":"2024-03-15T06:30:00","guid":{"rendered":"https:\/\/datascientest.com\/?p=231621"},"modified":"2024-03-14T14:59:03","modified_gmt":"2024-03-14T14:59:03","slug":"r-cnn-ou-region-based-convolutional-network","status":"publish","type":"post","link":"https:\/\/datascientest.com\/r-cnn-ou-region-based-convolutional-network","title":{"rendered":"R-CNN (Region Based Convolutional Network) :  tout sur ce mod\u00e8le de Machine Learning"},"content":{"rendered":"\t\t<div data-elementor-type=\"wp-post\" data-elementor-id=\"231621\" class=\"elementor elementor-231621\" data-elementor-post-type=\"post\">\n\t\t\t\t\t\t\t<div class=\"elementor-element elementor-element-a16a134 e-flex e-con-boxed e-con\" data-id=\"a16a134\" data-element_type=\"container\" data-settings=\"{&quot;content_width&quot;:&quot;boxed&quot;}\">\n\t\t\t\t\t<div class=\"e-con-inner\">\n\t\t\t\t<div class=\"elementor-element elementor-element-49026de elementor-widget elementor-widget-heading\" data-id=\"49026de\" data-element_type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<style>\/*! elementor - v3.15.0 - 02-08-2023 *\/\n.elementor-heading-title{padding:0;margin:0;line-height:1}.elementor-widget-heading .elementor-heading-title[class*=elementor-size-]>a{color:inherit;font-size:inherit;line-height:inherit}.elementor-widget-heading .elementor-heading-title.elementor-size-small{font-size:15px}.elementor-widget-heading .elementor-heading-title.elementor-size-medium{font-size:19px}.elementor-widget-heading .elementor-heading-title.elementor-size-large{font-size:29px}.elementor-widget-heading .elementor-heading-title.elementor-size-xl{font-size:39px}.elementor-widget-heading .elementor-heading-title.elementor-size-xxl{font-size:59px}<\/style><h2 class=\"elementor-heading-title elementor-size-default\">Comment enseigner \u00e0 un ordinateur la reconnaissance d\u2019\u00e9l\u00e9ments dans une image ? Ce probl\u00e8me a fait l\u2019objet de vastes recherches. L\u2019apparition de R-CNN a constitu\u00e9 une \u00e9tape majeure en la mati\u00e8re.<\/h2>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-35b83d7 elementor-widget elementor-widget-text-editor\" data-id=\"35b83d7\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<style>\/*! elementor - v3.15.0 - 02-08-2023 *\/\n.elementor-widget-text-editor.elementor-drop-cap-view-stacked .elementor-drop-cap{background-color:#69727d;color:#fff}.elementor-widget-text-editor.elementor-drop-cap-view-framed .elementor-drop-cap{color:#69727d;border:3px solid;background-color:transparent}.elementor-widget-text-editor:not(.elementor-drop-cap-view-default) .elementor-drop-cap{margin-top:8px}.elementor-widget-text-editor:not(.elementor-drop-cap-view-default) .elementor-drop-cap-letter{width:1em;height:1em}.elementor-widget-text-editor .elementor-drop-cap{float:left;text-align:center;line-height:1;font-size:50px}.elementor-widget-text-editor .elementor-drop-cap-letter{display:inline-block}<\/style>\t\t\t\t<p><span style=\"font-weight: 400;\">Doter un ordinateur de vision est une qu\u00eate qui s\u2019est \u00e9tendue sur plusieurs d\u00e9cennies. Comment amener une machine \u00e0 distinguer ce qui est quoi dans une sc\u00e8ne capt\u00e9e par une cam\u00e9ra ? Les applications de reconnaissance d\u2019image se sont acharn\u00e9es \u00e0 d\u00e9terminer :<\/span><\/p><ul><li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">les \u00e9l\u00e9ments pr\u00e9sents dans une image, \u00e0 partir de caract\u00e9ristiques telles que leurs contours ;<\/span><\/li><li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">la classification des objets distinctifs : personnages, animaux, meubles, v\u00e9hicules\u2026<\/span><\/li><\/ul>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-e47a08c elementor-widget elementor-widget-text-editor\" data-id=\"e47a08c\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p><span style=\"font-weight: 400;\">Diverses technologies ont \u00e9t\u00e9 imagin\u00e9es en vue de r\u00e9soudre ces probl\u00e8mes. CNN, puis R-CNN figurent parmi les r\u00e9ponses les plus efficaces apport\u00e9es.<\/span><\/p><p><span style=\"font-weight: 400;\">Pour mieux comprendre ce qu\u2019elles ont apport\u00e9es, il peut \u00eatre bon de revenir sur l\u2019approche mise en \u0153uvre auparavant.<\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t<div class=\"elementor-element elementor-element-cea408d e-flex e-con-boxed e-con\" data-id=\"cea408d\" data-element_type=\"container\" data-settings=\"{&quot;content_width&quot;:&quot;boxed&quot;}\">\n\t\t\t\t\t<div class=\"e-con-inner\">\n\t\t\t\t<div class=\"elementor-element elementor-element-0007cdd elementor-widget elementor-widget-heading\" data-id=\"0007cdd\" data-element_type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<h3 class=\"elementor-heading-title elementor-size-default\">De SVM aux CNN<\/h3>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-a9b353a elementor-widget elementor-widget-text-editor\" data-id=\"a9b353a\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p><span style=\"font-weight: 400;\">L\u2019approche <a href=\"https:\/\/datascientest.com\/svm\">SVM<\/a> (Support Vector Machine) reposait sur les valeurs des pixels individuels au sein d\u2019une image. Elle p\u00eachait \u00e0 de nombreux niveaux, notamment\u00a0<\/span><span style=\"font-weight: 400;\">celui-ci : le d\u00e9veloppeur devait lui-m\u00eame s\u00e9lectionner et identifier les caract\u00e9ristiques des donn\u00e9es utilis\u00e9es pour l&rsquo;entra\u00eenement de l&rsquo;algorithme.\u00a0<\/span><\/p><p><span style=\"font-weight: 400;\">Il \u00e9tait n\u00e9cessaire de pouvoir t\u00f4t ou tard automatiser cette t\u00e2che, sans la n\u00e9cessit\u00e9 d\u2019une intervention humaine. \u00c0 cette fin, le chercheur fran\u00e7ais Yann Le Cun a pr\u00e9sent\u00e9 le concept des CNN en 1998.<\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-c60951f elementor-align-center button-principale elementor-widget elementor-widget-button\" data-id=\"c60951f\" data-element_type=\"widget\" data-widget_type=\"button.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<div class=\"elementor-button-wrapper\">\n\t\t\t<a class=\"elementor-button elementor-button-link elementor-size-sm elementor-animation-grow\" href=\"https:\/\/datascientest.com\/formation-data-scientist\">\n\t\t\t\t\t\t<span class=\"elementor-button-content-wrapper\">\n\t\t\t\t\t\t<span class=\"elementor-button-text\">Tout apprendre du R-CNN<\/span>\n\t\t<\/span>\n\t\t\t\t\t<\/a>\n\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t<div class=\"elementor-element elementor-element-e761d7a e-flex e-con-boxed e-con\" data-id=\"e761d7a\" data-element_type=\"container\" data-settings=\"{&quot;content_width&quot;:&quot;boxed&quot;}\">\n\t\t\t\t\t<div class=\"e-con-inner\">\n\t\t\t\t<div class=\"elementor-element elementor-element-53377c5 elementor-widget elementor-widget-heading\" data-id=\"53377c5\" data-element_type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<h3 class=\"elementor-heading-title elementor-size-default\">R\u00e9seaux neuronaux convolutif (CNN)<\/h3>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-1f87a6d elementor-widget elementor-widget-text-editor\" data-id=\"1f87a6d\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p><span style=\"font-weight: 400;\">Les <a href=\"https:\/\/datascientest.com\/convolutional-neural-network\">CNN<\/a> peuvent \u00eatre consid\u00e9r\u00e9s comme des extracteurs automatiques de caract\u00e9ristiques au sein d\u2019une image. Si NN signifie Neural Network (<a href=\"https:\/\/datascientest.com\/fonctionnement-des-reseaux-neurones\">r\u00e9seau de neurones<\/a>), le C signifie convolutif.<\/span><\/p><p><span style=\"font-weight: 400;\">La convolution est une technique qui utilise un filtre pour transformer une image en une autre, \u00e0 l\u2019aide d\u2019op\u00e9rations math\u00e9matiques.<\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t<div class=\"elementor-element elementor-element-c8bfd5f e-flex e-con-boxed e-con\" data-id=\"c8bfd5f\" data-element_type=\"container\" data-settings=\"{&quot;content_width&quot;:&quot;boxed&quot;}\">\n\t\t\t\t\t<div class=\"e-con-inner\">\n\t\t\t\t<div class=\"elementor-element elementor-element-f21af5f elementor-widget elementor-widget-heading\" data-id=\"f21af5f\" data-element_type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<h3 class=\"elementor-heading-title elementor-size-default\">AlexNet<\/h3>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-3f683e0 elementor-widget elementor-widget-text-editor\" data-id=\"3f683e0\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p><span style=\"font-weight: 400;\">Plus tard, en 2012, trois ing\u00e9nieurs, Alex Krizhevsky, Ilya Sutskever et Geoffrey Hinton ont propos\u00e9 un CNN \u00e9labor\u00e9, AlexNet, reposant sur plusieurs couches de convolution. AlexNet s\u2019est distingu\u00e9 en remportant une importante comp\u00e9tition de reconnaissance d\u2019image, <strong>ILSVRC (ImageNet Large Scale Visual Recognition Challenge)<\/strong>. D\u00e8s lors, il est apparu que les CNN \u00e9taient en mesure de surpasser les techniques traditionnelles de reconnaissance d\u2019image.<\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-9c2c7f3 elementor-widget elementor-widget-image\" data-id=\"9c2c7f3\" data-element_type=\"widget\" data-widget_type=\"image.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<style>\/*! elementor - v3.15.0 - 02-08-2023 *\/\n.elementor-widget-image{text-align:center}.elementor-widget-image a{display:inline-block}.elementor-widget-image a img[src$=\".svg\"]{width:48px}.elementor-widget-image img{vertical-align:middle;display:inline-block}<\/style>\t\t\t\t\t\t\t\t\t\t\t\t<img decoding=\"async\" width=\"767\" height=\"600\" src=\"https:\/\/datascientest.com\/wp-content\/uploads\/2024\/02\/R-CNN-datascientest1.jpg\" class=\"attachment-full size-full wp-image-231642\" alt=\"\" srcset=\"https:\/\/datascientest.com\/wp-content\/uploads\/2024\/02\/R-CNN-datascientest1.jpg 767w, https:\/\/datascientest.com\/wp-content\/uploads\/2024\/02\/R-CNN-datascientest1-300x235.jpg 300w\" sizes=\"(max-width: 767px) 100vw, 767px\" \/>\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t<div class=\"elementor-element elementor-element-5f88e49 e-flex e-con-boxed e-con\" data-id=\"5f88e49\" data-element_type=\"container\" data-settings=\"{&quot;content_width&quot;:&quot;boxed&quot;}\">\n\t\t\t\t\t<div class=\"e-con-inner\">\n\t\t\t\t<div class=\"elementor-element elementor-element-b002dc4 elementor-widget elementor-widget-heading\" data-id=\"b002dc4\" data-element_type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<h3 class=\"elementor-heading-title elementor-size-default\">R-CNN ou les CNN bas\u00e9 sur des R\u00e9gions<\/h3>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-ea56941 elementor-widget elementor-widget-text-editor\" data-id=\"ea56941\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p><span style=\"font-weight: 400;\">Les CNN \u00e9taient encore limit\u00e9s car leur classification ne pouvait \u00eatre accomplie que sur un seul objet d\u2019une image. De ce fait, en 2014, Ross Girshick a propos\u00e9 un <strong>algorithme de d\u00e9tection d\u2019objets<\/strong> plus pouss\u00e9 : <strong>R-CNN ou CNN bas\u00e9 sur les R\u00e9gions<\/strong>. Ce mod\u00e8le pionnier a fortement aid\u00e9 \u00e0 faire avancer le domaine de la d\u00e9tection d&rsquo;objets. L&rsquo;architecture R-CNN, \u00e0 partir d&rsquo;une image d&rsquo;entr\u00e9e, extrait des propositions de r\u00e9gions, calcule ensuite leurs caract\u00e9ristiques CNN et s\u2019\u00e9vertue \u00e0 les classifier.<\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t<div class=\"elementor-element elementor-element-05b7c91 e-flex e-con-boxed e-con\" data-id=\"05b7c91\" data-element_type=\"container\" data-settings=\"{&quot;content_width&quot;:&quot;boxed&quot;}\">\n\t\t\t\t\t<div class=\"e-con-inner\">\n\t\t\t\t<div class=\"elementor-element elementor-element-7e4cd4b elementor-widget elementor-widget-heading\" data-id=\"7e4cd4b\" data-element_type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<h3 class=\"elementor-heading-title elementor-size-default\">Comment fonctionne R-CNN ?<\/h3>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-45cc2ff elementor-widget elementor-widget-text-editor\" data-id=\"45cc2ff\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p><span style=\"font-weight: 400;\">Le fonctionnement de R-CNN se d\u00e9roule ainsi.<\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-d38463c elementor-widget elementor-widget-heading\" data-id=\"d38463c\" data-element_type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<h4 class=\"elementor-heading-title elementor-size-medium\">Proposition de r\u00e9gions<\/h4>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-164e0ee elementor-widget elementor-widget-text-editor\" data-id=\"164e0ee\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p><span style=\"font-weight: 400;\">R-CNN commence par diviser l&rsquo;image d&rsquo;entr\u00e9e en plusieurs r\u00e9gions potentielles. <\/span><span style=\"font-weight: 400;\">Les caract\u00e9ristiques o\u00f9 un objet peut \u00eatre localis\u00e9 sont d\u00e9termin\u00e9es au moyen de m\u00e9thodes telles que le <strong>selective search (recherche s\u00e9lective)<\/strong>. Ce selective search fonctionne en fusionnant ou en divisant des segments de l&rsquo;image selon divers indices : couleur, forme, texture&#8230;<\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-0e96910 elementor-widget elementor-widget-heading\" data-id=\"0e96910\" data-element_type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<h4 class=\"elementor-heading-title elementor-size-medium\">Extraction de caract\u00e9ristiques<\/h4>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-fb0f496 elementor-widget elementor-widget-text-editor\" data-id=\"fb0f496\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<span style=\"font-weight: 400;\">Une fois ces propositions de r\u00e9gions g\u00e9n\u00e9r\u00e9es, quelques 2 000 r\u00e9gions sont extraites et d\u00e9form\u00e9es \u00e0 une taille coh\u00e9rente. Chacune de ces r\u00e9gions est ensuite donn\u00e9e en entr\u00e9e \u00e0 un mod\u00e8le CNN, qui va t\u00e2cher d\u2019en extraire des caract\u00e9ristiques pr\u00e9cises.<\/span>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-defbf28 elementor-align-center button-principale elementor-widget elementor-widget-button\" data-id=\"defbf28\" data-element_type=\"widget\" data-widget_type=\"button.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<div class=\"elementor-button-wrapper\">\n\t\t\t<a class=\"elementor-button elementor-button-link elementor-size-sm elementor-animation-grow\" href=\"https:\/\/datascientest.com\/formation-data-scientist\">\n\t\t\t\t\t\t<span class=\"elementor-button-content-wrapper\">\n\t\t\t\t\t\t<span class=\"elementor-button-text\">Ma\u00eetriser le fonctionnement du R-CNN<\/span>\n\t\t<\/span>\n\t\t\t\t\t<\/a>\n\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-b5bea9f elementor-widget elementor-widget-heading\" data-id=\"b5bea9f\" data-element_type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<h4 class=\"elementor-heading-title elementor-size-medium\">Classification des objets<\/h4>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-068f516 elementor-widget elementor-widget-text-editor\" data-id=\"068f516\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p><span style=\"font-weight: 400;\">Les vecteurs de caract\u00e9ristiques issus des propositions de r\u00e9gions sont aliment\u00e9s dans un <strong>classificateur d&rsquo;<\/strong><a href=\"https:\/\/datascientest.com\/apprentissage-automatique\">apprentissage automatique<\/a> visant \u00e0 d\u00e9terminer si la proposition de r\u00e9gion contient ou non une instance d\u2019une classe pr\u00e9d\u00e9finie.<\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-48dd6dc elementor-widget elementor-widget-heading\" data-id=\"48dd6dc\" data-element_type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<h4 class=\"elementor-heading-title elementor-size-medium\">Bounding Box Regression - R\u00e9gression de la bo\u00eete englobante<\/h4>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-86b407a elementor-widget elementor-widget-text-editor\" data-id=\"86b407a\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p><span style=\"font-weight: 400;\">R-CNN effectue une<strong> r\u00e9gression de la bo\u00eete englobante<\/strong>, afin d\u2019affiner l&#8217;emplacement et la taille de la bo\u00eete englobante autour de l&rsquo;objet d\u00e9tect\u00e9.<\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-c6b6d2a elementor-widget elementor-widget-heading\" data-id=\"c6b6d2a\" data-element_type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<h4 class=\"elementor-heading-title elementor-size-medium\">Non-Maximum Suppression (NMS)<\/h4>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-fe3dd04 elementor-widget elementor-widget-text-editor\" data-id=\"fe3dd04\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p><span style=\"font-weight: 400;\">Une fois qu\u2019il a class\u00e9 les bo\u00eetes englobantes et op\u00e9r\u00e9 une r\u00e9gression pour chaque r\u00e9gion propos\u00e9e, R-CNN applique un algorithme de \u00ab\u00a0suppression non maximale\u00a0\u00bb qui vise \u00e0 \u00e9liminer les bo\u00eetes englobantes en double ou fortement superpos\u00e9es.<\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-5ffefc6 elementor-widget elementor-widget-image\" data-id=\"5ffefc6\" data-element_type=\"widget\" data-widget_type=\"image.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<img decoding=\"async\" width=\"684\" height=\"600\" src=\"https:\/\/datascientest.com\/wp-content\/uploads\/2024\/02\/R-CNN-datascientest2.jpg\" class=\"attachment-full size-full wp-image-231643\" alt=\"\" srcset=\"https:\/\/datascientest.com\/wp-content\/uploads\/2024\/02\/R-CNN-datascientest2.jpg 684w, https:\/\/datascientest.com\/wp-content\/uploads\/2024\/02\/R-CNN-datascientest2-300x263.jpg 300w\" sizes=\"(max-width: 684px) 100vw, 684px\" \/>\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t<div class=\"elementor-element elementor-element-aed98ce e-flex e-con-boxed e-con\" data-id=\"aed98ce\" data-element_type=\"container\" data-settings=\"{&quot;content_width&quot;:&quot;boxed&quot;}\">\n\t\t\t\t\t<div class=\"e-con-inner\">\n\t\t\t\t<div class=\"elementor-element elementor-element-fbe5867 elementor-widget elementor-widget-heading\" data-id=\"fbe5867\" data-element_type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<h3 class=\"elementor-heading-title elementor-size-default\">R-CNN : flexible et robuste\n<\/h3>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-05f06ef elementor-widget elementor-widget-text-editor\" data-id=\"05f06ef\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p><span style=\"font-weight: 400;\">La publication de l&rsquo;article de Ross Girshick sur R-CNN a \u00e9t\u00e9 une \u00e9tape majeure dans le domaine de la vision par ordinateur. Il a d\u00e9montr\u00e9 qu\u2019il \u00e9tait possible d\u2019<strong>atteindre des performances \u00e9lev\u00e9es dans la d\u00e9tection d&rsquo;objets dans une image<\/strong> et a permis de les localiser de fa\u00e7on pr\u00e9cise gr\u00e2ce aux bo\u00eetes englobantes.<\/span><\/p><p><span style=\"font-weight: 400;\">L\u2019exp\u00e9rience a d\u00e9montr\u00e9 que l\u2019approche R-CNN \u00e9tait \u00e0 la fois flexible &#8211; adaptable \u00e0 des cas de figure tr\u00e8s vari\u00e9s &#8211; mais aussi robuste. Elle a \u00e9t\u00e9 mise \u00e0 profit sur sur de nombreux sc\u00e9narios du monde r\u00e9el comme l\u2019<strong>identification d\u2019objets abandonn\u00e9s dans des zones publiques ou encore la d\u00e9tection d\u2019obstacles par des v\u00e9hicules autonomes<\/strong>.<\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t<div class=\"elementor-element elementor-element-19002e7 e-flex e-con-boxed e-con\" data-id=\"19002e7\" data-element_type=\"container\" data-settings=\"{&quot;content_width&quot;:&quot;boxed&quot;}\">\n\t\t\t\t\t<div class=\"e-con-inner\">\n\t\t\t\t<div class=\"elementor-element elementor-element-aaf4414 elementor-widget elementor-widget-heading\" data-id=\"aaf4414\" data-element_type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<h3 class=\"elementor-heading-title elementor-size-default\">Le Fast R-CNN<\/h3>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-0df369c elementor-widget elementor-widget-text-editor\" data-id=\"0df369c\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t\t\t<p><span style=\"font-weight: 400;\">Pourtant, R-CNN n\u2019est pas le mod\u00e8le le plus rapide qui soit. Dans la mesure o\u00f9 chaque image doit classifier 2000 propositions de r\u00e9gion, il faut un certain temps pour entra\u00eener le r\u00e9seau et aussi un espace disque important pour stocker la carte de caract\u00e9ristiques de la proposition de r\u00e9gion.<\/span><\/p><p><span style=\"font-weight: 400;\">Par la force des choses, R-CNN a donc ouvert la voie \u00e0 des am\u00e9liorations notables en terme de vitesse dans la d\u00e9tection des objets comme <strong>Fast R-CNN<\/strong>. Celui-ci traite l\u2019image de fa\u00e7on int\u00e9grale plut\u00f4t que chaque proposition de r\u00e9gion individuelle. Il identifie des <strong>r\u00e9gions d&rsquo;int\u00e9r\u00eat<\/strong> (<strong>RoI<\/strong>) pouvant contenir des objets et extrait pour chacune les caract\u00e9ristiques correspondantes. Une fois redimensionn\u00e9es, ces caract\u00e9ristiques sont soumises \u00e0 des couches suppl\u00e9mentaires du r\u00e9seau qui classent l&rsquo;objet dans la r\u00e9gion d&rsquo;int\u00e9r\u00eat et affinent les coordonn\u00e9es de la bo\u00eete englobante. Il r\u00e9sulte de cette approche un traitement plus v\u00e9loce. D\u2019autres tentatives similaires sont apparues comme Faster R-CNN.<\/span><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-7a2d8cd elementor-align-center button-principale elementor-widget elementor-widget-button\" data-id=\"7a2d8cd\" data-element_type=\"widget\" data-widget_type=\"button.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t\t\t<div class=\"elementor-button-wrapper\">\n\t\t\t<a class=\"elementor-button elementor-button-link elementor-size-sm elementor-animation-grow\" href=\"https:\/\/datascientest.com\/formation-intelligence-artificielle\">\n\t\t\t\t\t\t<span class=\"elementor-button-content-wrapper\">\n\t\t\t\t\t\t<span class=\"elementor-button-text\">Se former au R-CNN<\/span>\n\t\t<\/span>\n\t\t\t\t\t<\/a>\n\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<div class=\"elementor-element elementor-element-c4b809b elementor-widget elementor-widget-html\" data-id=\"c4b809b\" data-element_type=\"widget\" data-widget_type=\"html.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<script type=\"application\/ld+json\">\r\n{\r\n  \"@context\": \"https:\/\/schema.org\",\r\n  \"@type\": \"FAQPage\",\r\n  \"mainEntity\": [{\r\n    \"@type\": \"Question\",\r\n    \"name\": \"De SVM aux CNN\",\r\n    \"acceptedAnswer\": {\r\n      \"@type\": \"Answer\",\r\n      \"text\": \"L\u2019approche SVM (Support Vector Machine) reposait sur les valeurs des pixels individuels au sein d\u2019une image. Elle p\u00eachait \u00e0 de nombreux niveaux, notamment celui-ci : le d\u00e9veloppeur devait lui-m\u00eame s\u00e9lectionner et identifier les caract\u00e9ristiques des donn\u00e9es utilis\u00e9es pour l'entra\u00eenement de l'algorithme.\"\r\n    }\r\n  },{\r\n    \"@type\": \"Question\",\r\n    \"name\": \"R\u00e9seaux neuronaux convolutif (CNN)\",\r\n    \"acceptedAnswer\": {\r\n      \"@type\": \"Answer\",\r\n      \"text\": \"Les CNN peuvent \u00eatre consid\u00e9r\u00e9s comme des extracteurs automatiques de caract\u00e9ristiques au sein d\u2019une image. Si NN signifie Neural Network (r\u00e9seau de neurone), le C signifie convolutif. La convolution est une technique qui utilise un filtre pour transformer une image en une autre, \u00e0 l\u2019aide d\u2019op\u00e9rations math\u00e9matiques.\"\r\n    }\r\n  },{\r\n    \"@type\": \"Question\",\r\n    \"name\": \"R-CNN ou les CNN bas\u00e9 sur des R\u00e9gions\",\r\n    \"acceptedAnswer\": {\r\n      \"@type\": \"Answer\",\r\n      \"text\": \"Les CNN \u00e9taient encore limit\u00e9s car leur classification ne pouvait \u00eatre accomplie que sur un seul objet d\u2019une image. De ce fait, en 2014, Ross Girshick a propos\u00e9 un algorithme de d\u00e9tection d\u2019objets plus pouss\u00e9 : R-CNN ou CNN bas\u00e9 sur les R\u00e9gions. Ce mod\u00e8le pionnier a fortement aid\u00e9 \u00e0 faire avancer le domaine de la d\u00e9tection d'objets. L'architecture R-CNN, \u00e0 partir d'une image d'entr\u00e9e, extrait des propositions de r\u00e9gions, calcule ensuite leurs caract\u00e9ristiques CNN et s\u2019\u00e9vertue \u00e0 les classifier.\"\r\n    }\r\n  }]\r\n}\r\n<\/script>\r\n\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t\t\t<\/div>\n\t\t","protected":false},"excerpt":{"rendered":"<p>Comment enseigner \u00e0 un ordinateur la reconnaissance d\u2019\u00e9l\u00e9ments dans une image ? Ce probl\u00e8me a fait l\u2019objet de vastes recherches. L\u2019apparition de R-CNN a constitu\u00e9 une \u00e9tape majeure en la mati\u00e8re. Doter un ordinateur de vision est une qu\u00eate qui s\u2019est \u00e9tendue sur plusieurs d\u00e9cennies. Comment amener une machine \u00e0 distinguer ce qui est quoi [&hellip;]<\/p>\n","protected":false},"author":77,"featured_media":232799,"comment_status":"open","ping_status":"open","sticky":false,"template":"elementor_theme","format":"standard","meta":{"content-type":"","inline_featured_image":false},"categories":[2497],"tags":[],"yoast_head":"<!-- This site is optimized with the Yoast SEO Premium plugin v21.7 (Yoast SEO v21.7) - https:\/\/yoast.com\/wordpress\/plugins\/seo\/ -->\n<title>R-CNN (Region Based Convolutional Network) : tout sur ce mod\u00e8le de Machine Learning<\/title>\n<meta name=\"description\" content=\"Comment enseigner \u00e0 un ordinateur la reconnaissance d\u2019\u00e9l\u00e9ments dans une image ? Ce probl\u00e8me a fait l\u2019objet de vastes recherches. L\u2019apparition de R-CNN a constitu\u00e9 une \u00e9tape majeure en la mati\u00e8re.\" \/>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/datascientest.com\/r-cnn-ou-region-based-convolutional-network\" \/>\n<meta property=\"og:locale\" content=\"fr_FR\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"R-CNN (Region Based Convolutional Network) : tout sur ce mod\u00e8le de Machine Learning\" \/>\n<meta property=\"og:url\" content=\"https:\/\/datascientest.com\/r-cnn-ou-region-based-convolutional-network\" \/>\n<meta property=\"og:site_name\" content=\"Formation Data Science | DataScientest.com\" \/>\n<meta property=\"article:published_time\" content=\"2024-03-15T06:30:00+00:00\" \/>\n<meta property=\"article:modified_time\" content=\"2024-03-14T14:59:03+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/datascientest.com\/wp-content\/uploads\/2024\/02\/rcnn-datascientest.jpg\" \/>\n\t<meta property=\"og:image:width\" content=\"1200\" \/>\n\t<meta property=\"og:image:height\" content=\"686\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/jpeg\" \/>\n<meta name=\"author\" content=\"Raphael Kassel\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"\u00c9crit par\" \/>\n\t<meta name=\"twitter:data1\" content=\"Raphael Kassel\" \/>\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:\/\/datascientest.com\/r-cnn-ou-region-based-convolutional-network\",\"url\":\"https:\/\/datascientest.com\/r-cnn-ou-region-based-convolutional-network\",\"name\":\"R-CNN (Region Based Convolutional Network) : tout sur ce mod\u00e8le de Machine Learning\",\"isPartOf\":{\"@id\":\"https:\/\/datascientest.com\/#website\"},\"datePublished\":\"2024-03-15T06:30:00+00:00\",\"dateModified\":\"2024-03-14T14:59:03+00:00\",\"author\":{\"@id\":\"https:\/\/datascientest.com\/#\/schema\/person\/6eca16f63daaafcbc96e15549b9d4ee1\"},\"description\":\"Comment enseigner \u00e0 un ordinateur la reconnaissance d\u2019\u00e9l\u00e9ments dans une image ? Ce probl\u00e8me a fait l\u2019objet de vastes recherches. L\u2019apparition de R-CNN a constitu\u00e9 une \u00e9tape majeure en la mati\u00e8re.\",\"inLanguage\":\"fr-FR\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/datascientest.com\/r-cnn-ou-region-based-convolutional-network\"]}]},{\"@type\":\"WebSite\",\"@id\":\"https:\/\/datascientest.com\/#website\",\"url\":\"https:\/\/datascientest.com\/\",\"name\":\"Formation Data Science | DataScientest.com\",\"description\":\"Leader de la formation en Data Science\",\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\/\/datascientest.com\/?s={search_term_string}\"},\"query-input\":\"required name=search_term_string\"}],\"inLanguage\":\"fr-FR\"},{\"@type\":\"Person\",\"@id\":\"https:\/\/datascientest.com\/#\/schema\/person\/6eca16f63daaafcbc96e15549b9d4ee1\",\"name\":\"Raphael Kassel\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"fr-FR\",\"@id\":\"https:\/\/datascientest.com\/#\/schema\/person\/image\/\",\"url\":\"https:\/\/datascientest.com\/wp-content\/uploads\/2023\/11\/raph.jpeg\",\"contentUrl\":\"https:\/\/datascientest.com\/wp-content\/uploads\/2023\/11\/raph.jpeg\",\"caption\":\"Raphael Kassel\"},\"description\":\"Raphael Kassel est le Head of Data Programs chez DataScientest. Il supervise une \u00e9quipe Data, impl\u00e9mente de nouveaux parcours comme \\\"Chef de projet IA\\\" et \\\"Data Product Manager\\\", et contribue \u00e0 la modernisation des op\u00e9rations du groupe. En collaboration \u00e9troite avec l'\u00e9quipe de d\u00e9veloppement, il adapte l'accompagnement acad\u00e9mique aux \u00e9volutions de l'\u00e9quipe data et \u00e0 l'internationalisation.\",\"sameAs\":[\"https:\/\/www.linkedin.com\/in\/raphaelkassel\/\",\"Data Science\"],\"url\":\"https:\/\/datascientest.com\/author\/raphaelkassel\"}]}<\/script>\n<!-- \/ Yoast SEO Premium plugin. -->","yoast_head_json":{"title":"R-CNN (Region Based Convolutional Network) : tout sur ce mod\u00e8le de Machine Learning","description":"Comment enseigner \u00e0 un ordinateur la reconnaissance d\u2019\u00e9l\u00e9ments dans une image ? Ce probl\u00e8me a fait l\u2019objet de vastes recherches. L\u2019apparition de R-CNN a constitu\u00e9 une \u00e9tape majeure en la mati\u00e8re.","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:\/\/datascientest.com\/r-cnn-ou-region-based-convolutional-network","og_locale":"fr_FR","og_type":"article","og_title":"R-CNN (Region Based Convolutional Network) : tout sur ce mod\u00e8le de Machine Learning","og_url":"https:\/\/datascientest.com\/r-cnn-ou-region-based-convolutional-network","og_site_name":"Formation Data Science | DataScientest.com","article_published_time":"2024-03-15T06:30:00+00:00","article_modified_time":"2024-03-14T14:59:03+00:00","og_image":[{"width":1200,"height":686,"url":"https:\/\/datascientest.com\/wp-content\/uploads\/2024\/02\/rcnn-datascientest.jpg","type":"image\/jpeg"}],"author":"Raphael Kassel","twitter_card":"summary_large_image","twitter_misc":{"\u00c9crit par":"Raphael Kassel","Dur\u00e9e de lecture estim\u00e9e":"6 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"WebPage","@id":"https:\/\/datascientest.com\/r-cnn-ou-region-based-convolutional-network","url":"https:\/\/datascientest.com\/r-cnn-ou-region-based-convolutional-network","name":"R-CNN (Region Based Convolutional Network) : tout sur ce mod\u00e8le de Machine Learning","isPartOf":{"@id":"https:\/\/datascientest.com\/#website"},"datePublished":"2024-03-15T06:30:00+00:00","dateModified":"2024-03-14T14:59:03+00:00","author":{"@id":"https:\/\/datascientest.com\/#\/schema\/person\/6eca16f63daaafcbc96e15549b9d4ee1"},"description":"Comment enseigner \u00e0 un ordinateur la reconnaissance d\u2019\u00e9l\u00e9ments dans une image ? Ce probl\u00e8me a fait l\u2019objet de vastes recherches. L\u2019apparition de R-CNN a constitu\u00e9 une \u00e9tape majeure en la mati\u00e8re.","inLanguage":"fr-FR","potentialAction":[{"@type":"ReadAction","target":["https:\/\/datascientest.com\/r-cnn-ou-region-based-convolutional-network"]}]},{"@type":"WebSite","@id":"https:\/\/datascientest.com\/#website","url":"https:\/\/datascientest.com\/","name":"Formation Data Science | DataScientest.com","description":"Leader de la formation en Data Science","potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/datascientest.com\/?s={search_term_string}"},"query-input":"required name=search_term_string"}],"inLanguage":"fr-FR"},{"@type":"Person","@id":"https:\/\/datascientest.com\/#\/schema\/person\/6eca16f63daaafcbc96e15549b9d4ee1","name":"Raphael Kassel","image":{"@type":"ImageObject","inLanguage":"fr-FR","@id":"https:\/\/datascientest.com\/#\/schema\/person\/image\/","url":"https:\/\/datascientest.com\/wp-content\/uploads\/2023\/11\/raph.jpeg","contentUrl":"https:\/\/datascientest.com\/wp-content\/uploads\/2023\/11\/raph.jpeg","caption":"Raphael Kassel"},"description":"Raphael Kassel est le Head of Data Programs chez DataScientest. Il supervise une \u00e9quipe Data, impl\u00e9mente de nouveaux parcours comme \"Chef de projet IA\" et \"Data Product Manager\", et contribue \u00e0 la modernisation des op\u00e9rations du groupe. En collaboration \u00e9troite avec l'\u00e9quipe de d\u00e9veloppement, il adapte l'accompagnement acad\u00e9mique aux \u00e9volutions de l'\u00e9quipe data et \u00e0 l'internationalisation.","sameAs":["https:\/\/www.linkedin.com\/in\/raphaelkassel\/","Data Science"],"url":"https:\/\/datascientest.com\/author\/raphaelkassel"}]}},"_links":{"self":[{"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/posts\/231621"}],"collection":[{"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/users\/77"}],"replies":[{"embeddable":true,"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/comments?post=231621"}],"version-history":[{"count":0,"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/posts\/231621\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/media\/232799"}],"wp:attachment":[{"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/media?parent=231621"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/categories?post=231621"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/datascientest.com\/wp-json\/wp\/v2\/tags?post=231621"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}