{"id":4087,"date":"2024-07-13T13:53:21","date_gmt":"2024-07-13T05:53:21","guid":{"rendered":"https:\/\/nullthought.net\/?p=4087"},"modified":"2025-02-23T11:08:03","modified_gmt":"2025-02-23T03:08:03","slug":"mambavision%ef%bc%9a%e4%b8%80%e7%a7%8d%e6%96%b0%e5%9e%8b%e6%b7%b7%e5%90%88mamba-transformer%e8%a7%86%e8%a7%89backbone","status":"publish","type":"post","link":"https:\/\/nullthought.net\/?p=4087","title":{"rendered":"MambaVision\uff1a\u4e00\u79cd\u65b0\u578b\u6df7\u5408Mamba-Transformer\u89c6\u89c9Backbone"},"content":{"rendered":"\n<p>Nvidia\u8fd9\u4e24\u5929\u53d1\u5e03\u4e86<a href=\"https:\/\/github.com\/NVlabs\/MambaVision\" target=\"_blank\" rel=\"noreferrer noopener\">MambaVision<\/a>\uff0c\u5373\u4e00\u79cd\u65b0\u578b\u6df7\u5408Mamba-Transformer\u89c6\u89c9Backbone\u3002\u9488\u5bf9\u7684MambaVision\u7684\u7814\u7a76\u5728\u8bba\u6587\u300a<a href=\"https:\/\/arxiv.org\/abs\/2407.08083\" target=\"_blank\" rel=\"noreferrer noopener\">MambaVision: A Hybrid Mamba-Transformer Vision Backbone<\/a>\u300b\u6709\u5177\u4f53\u9610\u8ff0\u3002<\/p>\n\n\n\n<p>\u4f5c\u4e3a\u4e00\u79cd\u65b0\u578b\u6df7\u5408Mamba-Transformer\u89c6\u89c9Backbone\uff0cMambaVision\u65e8\u5728\u63d0\u9ad8\u89c6\u89c9\u4efb\u52a1\u4e2d\u7684\u5efa\u6a21\u80fd\u529b\u3002\u8bba\u6587\u7684\u6838\u5fc3\u8d21\u732e\u5305\u62ec\u91cd\u65b0\u8bbe\u8ba1Mamba\u6a21\u578b\uff0c\u4ee5\u589e\u5f3a\u5176\u5bf9\u89c6\u89c9\u7279\u5f81\u7684\u9ad8\u6548\u5efa\u6a21\u80fd\u529b\uff0c\u5e76\u7efc\u5408\u7814\u7a76\u4e86Vision Transformer\uff08ViT\uff09\u4e0eMamba\u7684\u96c6\u6210\u53ef\u884c\u6027\u3002<strong>\u7814\u7a76\u7ed3\u679c\u8868\u660e\uff0c\u5728Mamba\u67b6\u6784\u7684\u6700\u7ec8\u5c42\u4e2d\u52a0\u5165\u82e5\u5e72\u81ea\u6ce8\u610f\u529b\u5757\u80fd\u591f\u663e\u8457\u63d0\u5347\u6355\u83b7\u957f\u8ddd\u79bb\u7a7a\u95f4\u4f9d\u8d56\u6027\u7684\u80fd\u529b<\/strong>\u3002\u57fa\u4e8e\u8fd9\u4e9b\u53d1\u73b0\uff0c\u63d0\u51fa\u4e86MambaVision\u6a21\u578b\u7cfb\u5217\uff0c\u5177\u6709\u5206\u5c42\u67b6\u6784\uff0c\u4ee5\u6ee1\u8db3\u4e0d\u540c\u7684\u8bbe\u8ba1\u6807\u51c6\u3002\u5728ImageNet-1K\u6570\u636e\u96c6\u4e0a\u7684\u56fe\u50cf\u5206\u7c7b\u4e2d\uff0cMambaVision\u6a21\u578b\u53d8\u4f53\u5728Top-1\u51c6\u786e\u7387\u548c\u56fe\u50cf\u541e\u5410\u91cf\u65b9\u9762\u8fbe\u5230\u4e86\u65b0\u7684SOTA\uff08State-of-the-Art\uff09\u6027\u80fd\u3002\u5728MS COCO\u548cADE20K\u6570\u636e\u96c6\u4e0a\u7684\u76ee\u6807\u68c0\u6d4b\u3001\u5b9e\u4f8b\u5206\u5272\u548c\u8bed\u4e49\u5206\u5272\u7b49\u4e0b\u6e38\u4efb\u52a1\u4e2d\uff0cMambaVision\u7684\u8868\u73b0\u4f18\u4e8e\u7c7b\u4f3c\u89c4\u6a21\u7684\u4e3b\u5e72\u7f51\u7edc\u3002<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-full\"><img loading=\"lazy\" decoding=\"async\" width=\"993\" height=\"967\" src=\"https:\/\/nullthought.net\/wp-content\/uploads\/2024\/07\/image-9.png\" alt=\"\" class=\"wp-image-4088\" srcset=\"https:\/\/nullthought.net\/wp-content\/uploads\/2024\/07\/image-9.png 993w, https:\/\/nullthought.net\/wp-content\/uploads\/2024\/07\/image-9-300x292.png 300w, https:\/\/nullthought.net\/wp-content\/uploads\/2024\/07\/image-9-768x748.png 768w\" sizes=\"auto, (max-width: 993px) 100vw, 993px\" \/><figcaption class=\"wp-element-caption\"><a href=\"https:\/\/arxiv.org\/abs\/2407.08083\"><\/a><a href=\"https:\/\/arxiv.org\/abs\/2407.08083\" target=\"_blank\" rel=\"noreferrer noopener\">MambaVision: A Hybrid Mamba-Transformer Vision Backbone<\/a>, Nvidia AI<\/figcaption><\/figure>\n\n\n\n<p>\u8bba\u6587\u7684\u5185\u5bb9\u6982\u8981\u5982\u4e0b\uff1a<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">\u4e00\u3001\u76f8\u5173\u5de5\u4f5c<\/h4>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>ViT<\/strong>\uff1aVision Transformer\u5229\u7528\u81ea\u6ce8\u610f\u529b\u5c42\u63d0\u4f9b\u4e86\u66f4\u5927\u7684\u611f\u53d7\u91ce\uff0c\u4f46\u521d\u671f\u7f3a\u4e4fCNN\u7684\u67d0\u4e9b\u4f18\u52bf\uff0c\u5982\u5f52\u7eb3\u504f\u7f6e\u548c\u5e73\u79fb\u4e0d\u53d8\u6027\uff0c\u9700\u8981\u5927\u89c4\u6a21\u8bad\u7ec3\u6570\u636e\u96c6\u6765\u8fbe\u5230\u7ade\u4e89\u6027\u80fd\u3002<\/li>\n\n\n\n<li><strong>Mamba<\/strong>\uff1a\u5f15\u5165\u4e86\u4e00\u79cd\u65b0\u7684\u72b6\u6001\u7a7a\u95f4\u6a21\u578b\uff08SSM\uff09\uff0c\u5b9e\u73b0\u4e86\u7ebf\u6027\u65f6\u95f4\u590d\u6742\u5ea6\uff0c\u5728\u4e0d\u540c\u8bed\u8a00\u5efa\u6a21\u4efb\u52a1\u4e2d\u8868\u73b0\u4f18\u5f02\uff0c\u4f46\u5728\u8ba1\u7b97\u673a\u89c6\u89c9\u4efb\u52a1\u4e2d\u6548\u7387\u8f83\u4f4e\u3002<\/li>\n<\/ul>\n\n\n\n<h4 class=\"wp-block-heading\">\u4e8c\u3001\u65b9\u6cd5<\/h4>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>\u5b8f\u89c2\u67b6\u6784<\/strong>\uff1aMambaVision\u7684\u5b8f\u89c2\u67b6\u6784\u5305\u62ec\u56db\u4e2a\u4e0d\u540c\u7684\u9636\u6bb5\u3002\u524d\u4e24\u4e2a\u9636\u6bb5\u4f7f\u7528\u57fa\u4e8eCNN\u7684\u5c42\u8fdb\u884c\u9ad8\u5206\u8fa8\u7387\u7279\u5f81\u7684\u5feb\u901f\u63d0\u53d6\uff0c\u800c\u7b2c\u4e09\u548c\u7b2c\u56db\u9636\u6bb5\u5305\u62ecMambaVision\u548cTransformer\u5757\u3002\u5177\u4f53\u6765\u8bf4\uff0c\u8f93\u5165\u56fe\u50cf\u9996\u5148\u8f6c\u6362\u4e3a\u91cd\u53e0\u7684patch\uff0c\u5e76\u901a\u8fc7stem\u6295\u5f71\u5230\u5d4c\u5165\u7a7a\u95f4\u4e2d\u3002\u6bcf\u4e2a\u9636\u6bb5\u4e4b\u95f4\u7684\u964d\u91c7\u6837\u5668\u5305\u62ec\u4e00\u4e2a3\u00d73\u7684CNN\u5c42\u3002<\/li>\n\n\n\n<li><strong>\u5fae\u89c2\u67b6\u6784<\/strong>\uff1a\u91cd\u65b0\u8bbe\u8ba1\u4e86Mamba\u6df7\u5408\u5668\uff0c\u66ff\u6362\u4e86\u56e0\u679c\u5377\u79ef\u5c42\uff0c\u5e76\u589e\u52a0\u4e86\u5bf9\u79f0\u8def\u5f84\u4ee5\u589e\u5f3a\u5168\u5c40\u4e0a\u4e0b\u6587\u7684\u5efa\u6a21\u80fd\u529b\u3002\u901a\u8fc7\u4f7f\u7528\u9009\u62e9\u673a\u5236\u548c\u8f93\u5165\u4f9d\u8d56\u7684\u5e8f\u5217\u5904\u7406\u6765\u6269\u5c55S4\u516c\u5f0f\uff0cMamba\u6a21\u578b\u80fd\u591f\u6839\u636e\u8f93\u5165\u52a8\u6001\u8c03\u6574\u53c2\u6570B\u3001C\u548c\u2206\u3002<\/li>\n<\/ul>\n\n\n\n<h4 class=\"wp-block-heading\">\u4e09\u3001\u5b9e\u9a8c<\/h4>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>\u56fe\u50cf\u5206\u7c7b<\/strong>\uff1a\u5728ImageNet-1K\u6570\u636e\u96c6\u4e0a\u7684\u5206\u7c7b\u7ed3\u679c\u8868\u660e\uff0cMambaVision-B\u5728Top-1\u51c6\u786e\u7387\u548c\u56fe\u50cf\u541e\u5410\u91cf\u65b9\u9762\u90fd\u4f18\u4e8eConvNeXt-B\u548cSwin-B\u7b49\u6d41\u884c\u6a21\u578b\u3002<\/li>\n\n\n\n<li><strong>\u76ee\u6807\u68c0\u6d4b\u548c\u5206\u5272<\/strong>\uff1a\u5728MS COCO\u6570\u636e\u96c6\u4e0a\u7684\u76ee\u6807\u68c0\u6d4b\u548c\u5b9e\u4f8b\u5206\u5272\u7ed3\u679c\u663e\u793a\uff0cMambaVision-T\u5728box AP\u548cmask AP\u65b9\u9762\u4f18\u4e8eConvNeXt-T\u548cSwin-T\u3002Cascade Mask-RCNN\u7f51\u7edc\u7684\u7ed3\u679c\u4e5f\u8868\u660e\uff0cMambaVision\u5728\u5404\u4e2a\u5c3a\u5bf8\u7684\u6a21\u578b\u4e2d\u90fd\u8868\u73b0\u4f18\u5f02\u3002<\/li>\n\n\n\n<li><strong>\u8bed\u4e49\u5206\u5272<\/strong>\uff1a\u5728ADE20K\u6570\u636e\u96c6\u4e0a\u7684\u8bed\u4e49\u5206\u5272\u57fa\u51c6\u6d4b\u8bd5\u4e2d\uff0cMambaVision\u6a21\u578b\u5728mIoU\u65b9\u9762\u4e5f\u4f18\u4e8e\u7c7b\u4f3c\u89c4\u6a21\u7684\u7ade\u4e89\u6a21\u578b\u3002<\/li>\n<\/ul>\n\n\n\n<h4 class=\"wp-block-heading\">\u56db\u3001\u7ed3\u8bba<\/h4>\n\n\n\n<p>\u672c\u6587\u4ecb\u7ecd\u4e86MambaVision\uff0c\u8fd9\u662f\u7b2c\u4e00\u4e2a\u4e13\u4e3a\u89c6\u89c9\u5e94\u7528\u8bbe\u8ba1\u7684Mamba-Transformer\u6df7\u5408\u4e3b\u5e72\u7f51\u7edc\u3002\u901a\u8fc7\u91cd\u65b0\u8bbe\u8ba1Mamba\u516c\u5f0f\uff0c\u589e\u5f3a\u4e86\u5168\u5c40\u4e0a\u4e0b\u6587\u8868\u793a\u5b66\u4e60\u80fd\u529b\uff0c\u5e76\u7cfb\u7edf\u5730\u7814\u7a76\u4e86\u6df7\u5408\u8bbe\u8ba1\u7684\u96c6\u6210\u6a21\u5f0f\u3002MambaVision\u5728Top-1\u51c6\u786e\u7387\u548c\u56fe\u50cf\u541e\u5410\u91cf\u65b9\u9762\u8fbe\u5230\u4e86\u65b0\u7684SOTA\uff0c\u663e\u8457\u8d85\u8d8a\u4e86\u57fa\u4e8eTransformer\u548cMamba\u7684\u6a21\u578b\u3002<\/p>\n\n\n\n<hr class=\"wp-block-separator has-alpha-channel-opacity\"\/>\n\n\n\n<p>MambaVision on GitHub: <a href=\"https:\/\/github.com\/NVlabs\/MambaVision\" target=\"_blank\" rel=\"noreferrer noopener\">https:\/\/github.com\/NVlabs\/MambaVision<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p>Nvidia\u8fd9\u4e24\u5929\u53d1\u5e03\u4e86MambaVision\uff0c\u5373\u4e00\u79cd\u65b0\u578b\u6df7\u5408Mamba-Transformer\u89c6\u89c9Backb [&hellip;]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"site-sidebar-layout":"default","site-content-layout":"","ast-site-content-layout":"default","site-content-style":"default","site-sidebar-style":"default","ast-global-header-display":"","ast-banner-title-visibility":"","ast-main-header-display":"","ast-hfb-above-header-display":"","ast-hfb-below-header-display":"","ast-hfb-mobile-header-display":"","site-post-title":"","ast-breadcrumbs-content":"","ast-featured-img":"","footer-sml-layout":"","ast-disable-related-posts":"","theme-transparent-header-meta":"","adv-header-id-meta":"","stick-header-meta":"","header-above-stick-meta":"","header-main-stick-meta":"","header-below-stick-meta":"","astra-migrate-meta-layouts":"set","ast-page-background-enabled":"default","ast-page-background-meta":{"desktop":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"ast-content-background-meta":{"desktop":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"footnotes":""},"categories":[8,36],"tags":[39,84,95,80],"class_list":["post-4087","post","type-post","status-publish","format-standard","hentry","category-tech","category-36","tag-ai","tag-nvidia","tag-transformer","tag-cv"],"rttpg_featured_image_url":null,"rttpg_author":{"display_name":"NullThought","author_link":"https:\/\/nullthought.net\/?author=1"},"rttpg_comment":1,"rttpg_category":"<a href=\"https:\/\/nullthought.net\/?cat=8\" rel=\"category\">Tech<\/a> <a href=\"https:\/\/nullthought.net\/?cat=36\" rel=\"category\">\u79d1\u5b66<\/a>","rttpg_excerpt":"Nvidia\u8fd9\u4e24\u5929\u53d1\u5e03\u4e86MambaVision\uff0c\u5373\u4e00\u79cd\u65b0\u578b\u6df7\u5408Mamba-Transformer\u89c6\u89c9Backb&hellip;","_links":{"self":[{"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/posts\/4087","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/nullthought.net\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=4087"}],"version-history":[{"count":3,"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/posts\/4087\/revisions"}],"predecessor-version":[{"id":4091,"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/posts\/4087\/revisions\/4091"}],"wp:attachment":[{"href":"https:\/\/nullthought.net\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=4087"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/nullthought.net\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=4087"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/nullthought.net\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=4087"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}