{"id":3748,"date":"2024-10-14T14:10:54","date_gmt":"2024-10-14T05:10:54","guid":{"rendered":"https:\/\/vds.sogang.ac.kr\/?p=3748"},"modified":"2024-10-14T14:11:41","modified_gmt":"2024-10-14T05:11:41","slug":"2024%eb%85%84-the-18th-european-conference-on-computer-vision-eccv-%ec%9c%a0%ed%98%84%ec%9a%b0-%ec%a1%b0%ec%9c%a0%eb%b9%88-%eb%ac%b8%ec%8a%b9%ed%9b%88","status":"publish","type":"post","link":"https:\/\/vds.sogang.ac.kr\/?p=3748","title":{"rendered":"2024\ub144 The 18th European Conference on Computer Vision (ECCV) &#8211; \uc720\ud604\uc6b0, \uc870\uc720\ube48, \ubb38\uc2b9\ud6c8"},"content":{"rendered":"<p>&nbsp;<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-3749\" src=\"https:\/\/vds.sogang.ac.kr\/wp-content\/uploads\/2024\/10\/eccv\ud6c4\uae30-scaled-e1728882686232.jpg\" alt=\"\" width=\"960\" height=\"1280\" \/><\/p>\n<p>[\ud559\ud68c \ud6c4\uae30]<br \/>\n\ucef4\ud4e8\ud130\ube44\uc804 \ubd84\uc57c top-tier \ud559\ud68c\uc778 ECCV 2024\uc5d0 \ubc1c\ud45c\uc790\ub85c \ucc38\uc11d\ud558\uc600\ub2e4. \uc774\ubc88 ECCV\ub294 9\uc6d4 29\uc77c\ubd80\ud130 10\uc6d4 4\uc77c\uae4c\uc9c0 \ucd1d 6\uc77c\uac04 \uc5f4\ub838\uace0, 9\uc6d4 29-30\uc77c\uc740 \uc6cc\ud06c\uc0f5\uc774 \uc9c4\ud589\ub418\uace0 10\uc6d4 1-4\uc77c\uc740 main conference\uac00 \uc9c4\ud589\ub418\uc5c8\ub2e4. \uc6b0\uc120 Workshop \uae30\uac04\uc5d0\ub294 \uae30\uc874\uc5d0 \uad00\uc2ec\uc788\ub358 efficiency\uc640 \uad00\ub828\ub41c \uc8fc\uc81c\ub97c \uc120\uc815\ud558\uc5ec \uc138\ubbf8\ub098\uc5d0 \ucc38\uc11d\ud558\uc600\ub2e4. \ucc38\uc11d\ud55c \uc8fc\uc81c\ub294 \u201cEfficient Deep Learning for Foundation Models\u201d, \u201cComputational Aspects of Deep Learning\u201d \ub4f1\uc774 \uc788\uc5c8\ub2e4. \ub180\ub77c\uc6e0\ub358 \uc810\uc740 \uacbd\ub7c9\ud654 \ud558\ub294 \ub300\uc0c1\uc774 \ub300\uaddc\ubaa8 foundation model\uc744 \ud0c0\uac9f\uc73c\ub85c \ud558\ub294 \uc5f0\uad6c\ub4e4\uc774\uc5c8\uae30 \ub54c\ubb38\uc5d0, \uacbd\ub7c9\ud654 \ub41c \ubaa8\ub378 \uc0ac\uc774\uc988\ub3c4 1B \uc2a4\ucf00\uc77c\uc758 \ud30c\ub77c\ubbf8\ud130\ub97c \uac16\uace0 \uc788\ub294 \ub525\ub7ec\ub2dd \ubaa8\ub378\uc774\uc5c8\ub2e4. \uc989, 70B \uc815\ub3c4\uc758 foundation \ubaa8\ub378\uc744 1B \uc2a4\ucf00\uc77c\ub85c \uc904\uc774\ub294 \uc5f0\uad6c \ub4f1\uc774 \uc788\uc5c8\ub2e4. \uc774\ub97c \ud1b5\ud574 \ucd5c\uadfc\uc5d0\ub294 \ub300\uaddc\ubaa8 \ubaa8\ub378\uc5d0 \uad00\uc2ec\uc774 \ub9ce\uc73c\uba70 \uacbd\ub7c9\ud654\ub97c \ud55c\ub2e4\uace0 \ud558\ub354\ub77c\ub3c4 \ub9ce\uc740 GPU \uc790\uc6d0\uc744 \uc694\uad6c\ud558\ub294 \uc5f0\uad6c\uac00 \ub300\uc138\ub97c \uc774\ub8e8\uace0 \uc788\ub2e4\ub294 \uac83\uc744 \uc54c \uc218 \uc788\uc5c8\ub2e4. \ub610\ud55c main conference\uc5d0\ub3c4 \ucc38\uc11d\ud558\uc5ec \uc5ec\ub7ec \ud3ec\uc2a4\ud130 \ubc1c\ud45c\uc790\ub4e4\uacfc \uc5f0\uad6c \uc774\uc57c\uae30\ub97c \ub098\ub20c \uc218 \uc788\uc5c8\ub2e4. \uc778\uc0c1\uae4a\uc5c8\ub358 \ub17c\ubb38\uc740 \u201cDenseNets Reloaded: Paradigm Shift Beyond ResNets and ViTs\u201d\uacfc \u201cSeiT++: Masked Token Modeling Improves Storage-efficient Training\u201d \uc774\uc5c8\ub2e4. DensNets\uc740 \ub9e4\uc6b0 \uc61b\ub0a0\uc5d0 \uc81c\uc548\ub41c \uad6c\uc870\ub85c residual summation \uad6c\uc870 \ub300\uc2e0 concatenate \ubc29\ubc95\uc73c\ub85c \uc774\uc804 layer\uc758 feature information\uc744 \ub2e4\uc74c\uc73c\ub85c \ub118\uaca8\uc900\ub2e4\ub294 \uac83\uc774\uc5c8\ub2e4. \ud558\uc9c0\ub9cc \uc774 \ubc29\ubc95\uc5d0\ub294 computational cost\uac00 \ub9ce\uc774 \ub4e0\ub2e4\ub294 \ub2e8\uc810\uc774 \uc788\uc5c8\ub294\ub370 channel\uc744 \uc870\uc815\ud558\uc5ec \uc774 \ub2e8\uc810\uc744 \uac1c\uc120\ud558\uc5ec \uc131\ub2a5 \uac1c\uc120\uc744 \uc774\ub8ec \uc5f0\uad6c\uc600\ub2e4. \uc774\uc640 \uac19\uc774 \ub2e8\uc21c\ud558\uba74\uc11c \ud6a8\uacfc\uc801\uc778 \uad6c\uc870\ub97c \uc81c\uc548\ud558\ub294 \uc5f0\uad6c\ub4e4\uc774 \ud56d\uc0c1 \ub300\ub2e8\ud558\ub2e4\uace0 \ub290\ub080\ub2e4. \uadf8\ub9ac\uace0 \ub450\ubc88\uc9f8\ub85c SeiT\ub294 vision data\ub97c tokenize\ub97c \ud1b5\ud574 size\ub97c \uc904\uc5ec\uc11c \ud559\uc2b5\uc744 \uc9c4\ud589\ud558\ub294 \ubd84\uc57c\uc5d0 \ub300\ud55c \uc5f0\uad6c\uc600\ub2e4. \uc774\ub97c \ud1b5\ud574 ImageNet-1k\ub97c 1%\uae4c\uc9c0 \uc904\uc5ec\uc11c \ud559\uc2b5\uc2dc\ud0ac \uc218 \uc788\ub294 \uacb0\uacfc\ub97c \ubcf4\uc5ec\uc8fc\uace0 \uc788\ub2e4. ImageNet\uc740 \ub370\uc774\ud130 \ud06c\uae30\uac00 \ub9e4\uc6b0 \ucee4\uc11c \ud559\uc2b5\uc774 \uc624\ub798\uac78\ub9ac\ub294\ub370 \ucd94\ud6c4\uc758 \uc5f0\uad6c\uc5d0 \ud574\ub2f9 framework\ub97c \ub3c4\uc785\ud560 \uc218 \uc788\uaca0\ub2e4\ub294 \uc0dd\uac01\uc774 \ub4e4\uc5b4\uc11c \uc778\uc0c1\uc801\uc774\uc5c8\ub2e4.<\/p>\n<p>[\ubc1c\ud45c \ud6c4\uae30]<br \/>\n\uc6b0\ub9ac \ud300\uc740 Transformer \uad6c\uc870\ub97c \uae30\ubc18\uc73c\ub85c \ud55c \uacbd\ub7c9 \uad6c\uc870 \ubc0f \ubc29\ubc95\uc5d0 \ub300\ud55c \ub0b4\uc6a9\uc73c\ub85c \ud3ec\uc2a4\ud130 \ubc1c\ud45c\ub97c \uc9c4\ud589\ud558\uc600\ub2e4. \uac10\uc0ac\ud558\uac8c\ub3c4 \uc0dd\uac01\ubcf4\ub2e4 \ub9ce\uc740 \uc0ac\ub78c\ub4e4\uc774 \uad00\uc2ec\uc744 \uac00\uc838\uc8fc\uc5c8\uace0, \ub2e4\uc591\ud55c \uae30\uc5c5\uc758 \uc0ac\ub78c\ub4e4\uacfc \uc5f0\uad6c\uc5d0 \ub300\ud55c \uc774\uc57c\uae30\ub97c \ud560 \uc218 \uc788\ub2e4\ub294 \uc810\uc774 \uc990\uac70\uc6e0\uace0 \uc55e\uc73c\ub85c\uc758 \uc5f0\uad6c \ubc29\ud5a5\uc131\uc5d0\ub3c4 \ud070 \ub3c4\uc6c0\uc774 \ub418\uc5c8\ub2e4. \ubc1c\ud45c\uc5d0\uc11c \ub098\uc628 \uc9c8\ubb38\uc740 \ub2e4\uc74c\uacfc \uac19\ub2e4.<br \/>\n\uc9c8\ubb38 : \uc81c\uc548\ud558\ub294 \uacbd\ub7c9 \uad6c\uc870 \ubc0f \ubc29\ubc95\uc774 vision task \ubfd0\ub9cc \uc544\ub2c8\ub77c diffusion \uc774\ub098 multi-modal\uacfc \uac19\uc740 task\uc5d0\ub3c4 \uc801\uc6a9\ub420 \uc218 \uc788\ub294\uc9c0?<br \/>\n\ub300\ub2f5 : \uc81c\uc548\ud558\ub294 \ubc29\ubc95\uc740 attention \uae30\ubc18\uc758 transformer \uad6c\uc870\ub97c \uae30\ubc18\uc73c\ub85c \ud55c\ub2e4. \ub530\ub77c\uc11c transformer \uad6c\uc870\ub97c \uae30\ubc18\uc73c\ub85c \ud558\ub294 \ub2e4\uc591\ud55c task\uc5d0 \ubc94\uc6a9\uc801\uc73c\ub85c \uc801\uc6a9\ud560 \uc218 \uc788\ub2e4.<br \/>\n\uc704\uc758 \uc9c8\ubb38\uc744 \ud1b5\ud574 \ucd5c\uadfc\uc5d0\ub294 vision \ubd84\uc57c\uc5d0\uc11c diffusion \ubc0f multi-modal task\uac00 \ud65c\ubc1c\ud558\uac8c \uc5f0\uad6c\ub418\uace0 \uc788\ub2e4\ub294 \uac83\uc744 \uc54c \uc218 \uc788\uc5c8\ub2e4. \ub530\ub77c\uc11c \ucd94\ud6c4 \uc5f0\uad6c\ub97c \uc9c4\ud589\ud560 \ub54c \uadf8\ucabd\uc5d0\uc11c\uc758 \ucd94\uac00\uc801\uc778 \uac80\uc99d \uc2e4\ud5d8\uc774 \ud544\uc694\ud558\ub2e4\ub294 \uac83\uc744 \ubc30\uc6b8 \uc218 \uc788\uc5c8\ub2e4.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>&nbsp; [\ud559\ud68c \ud6c4\uae30] \ucef4\ud4e8\ud130\ube44\uc804 \ubd84\uc57c top-tier \ud559\ud68c\uc778 ECCV 2024\uc5d0 \ubc1c\ud45c\uc790\ub85c \ucc38\uc11d\ud558\uc600\ub2e4. \uc774\ubc88 ECCV\ub294 9\uc6d4 29\uc77c\ubd80\ud130 10\uc6d4 4\uc77c\uae4c\uc9c0 \ucd1d 6\uc77c\uac04 \uc5f4\ub838\uace0, 9\uc6d4 29-30\uc77c\uc740 \uc6cc\ud06c\uc0f5\uc774 \uc9c4\ud589\ub418\uace0 10\uc6d4 1-4\uc77c\uc740 main conference\uac00 \uc9c4\ud589\ub418\uc5c8\ub2e4. \uc6b0\uc120 Workshop \uae30\uac04\uc5d0\ub294 \uae30\uc874\uc5d0 \uad00\uc2ec\uc788\ub358 efficiency\uc640 \uad00\ub828\ub41c \uc8fc\uc81c\ub97c \uc120\uc815\ud558\uc5ec \uc138\ubbf8\ub098\uc5d0 \ucc38\uc11d\ud558\uc600\ub2e4. \ucc38\uc11d\ud55c \uc8fc\uc81c\ub294 \u201cEfficient Deep Learning for Foundation Models\u201d, \u201cComputational Aspects of Deep Learning\u201d \ub4f1\uc774 \uc788\uc5c8\ub2e4.&hellip;&nbsp;<a href=\"https:\/\/vds.sogang.ac.kr\/?p=3748\" class=\"\" rel=\"bookmark\">\ub354 \ubcf4\uae30 &raquo;<span class=\"screen-reader-text\">2024\ub144 The 18th European Conference on Computer Vision (ECCV) &#8211; \uc720\ud604\uc6b0, \uc870\uc720\ube48, \ubb38\uc2b9\ud6c8<\/span><\/a><\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"neve_meta_sidebar":"","neve_meta_container":"","neve_meta_enable_content_width":"off","neve_meta_content_width":70,"neve_meta_title_alignment":"","neve_meta_author_avatar":"","neve_post_elements_order":"","neve_meta_disable_header":"","neve_meta_disable_footer":"","neve_meta_disable_title":"","footnotes":""},"categories":[27],"tags":[],"class_list":["post-3748","post","type-post","status-publish","format-standard","hentry","category-conference"],"_links":{"self":[{"href":"https:\/\/vds.sogang.ac.kr\/index.php?rest_route=\/wp\/v2\/posts\/3748","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/vds.sogang.ac.kr\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/vds.sogang.ac.kr\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/vds.sogang.ac.kr\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/vds.sogang.ac.kr\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=3748"}],"version-history":[{"count":2,"href":"https:\/\/vds.sogang.ac.kr\/index.php?rest_route=\/wp\/v2\/posts\/3748\/revisions"}],"predecessor-version":[{"id":3751,"href":"https:\/\/vds.sogang.ac.kr\/index.php?rest_route=\/wp\/v2\/posts\/3748\/revisions\/3751"}],"wp:attachment":[{"href":"https:\/\/vds.sogang.ac.kr\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=3748"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/vds.sogang.ac.kr\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=3748"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/vds.sogang.ac.kr\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=3748"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}