{"id":22099,"date":"2024-07-29T21:22:38","date_gmt":"2024-07-30T04:22:38","guid":{"rendered":"http:\/\/jnack.com\/blog\/?p=22099"},"modified":"2024-07-29T21:22:38","modified_gmt":"2024-07-30T04:22:38","slug":"meta-releases-sam-2-for-fast-segmentation","status":"publish","type":"post","link":"http:\/\/jnack.com\/blog\/2024\/07\/29\/meta-releases-sam-2-for-fast-segmentation\/","title":{"rendered":"Meta releases SAM 2 for fast segmentation"},"content":{"rendered":"\n<p>Man, I&#8217;m old enough to remember rotoscoping video by hand\u2014a process that quickly made me want to jump right out a window. Years later, when we were working on realtime video segmentation at Google, I was so proud to show the tech to a bunch of high school design students\u2014only to have them shrug and treat it as completely normal.<\/p>\n\n\n\n<p>Ah, but so it goes: &#8220;One of history\u2019s few iron laws is that luxuries tend to become necessities and to spawn new obligations. Once people get used to a certain luxury, they take it for granted.&#8221; \u2014 Yuval Noah Harari<\/p>\n\n\n\n<p>In any case, Meta has just released what looks like a great update to their excellent\u2014and open-source\u2014Segment Anything Model. Check it out:<\/p>\n\n\n<blockquote class=\"twitter-tweet\">\n<p dir=\"ltr\" lang=\"en\">Introducing Meta Segment Anything Model 2 (SAM 2) \u2014 the first unified model for real-time, promptable object segmentation in images &amp; videos.<\/p>\n<p>SAM 2 is available today under Apache 2.0 so that anyone can use it to build their own experiences<\/p>\n<p>Details <a href=\"https:\/\/t.co\/eTTDpxI60h\">https:\/\/t.co\/eTTDpxI60h<\/a> <a href=\"https:\/\/t.co\/mOFiF1kZfE\">pic.twitter.com\/mOFiF1kZfE<\/a><\/p>\n<p>\u2014 AI at Meta (@AIatMeta) <a href=\"https:\/\/twitter.com\/AIatMeta\/status\/1818055906179105010?ref_src=twsrc%5Etfw\">July 29, 2024<\/a><\/p>\n<\/blockquote>\n<p><script async=\"\" src=\"https:\/\/platform.twitter.com\/widgets.js\" charset=\"utf-8\"><\/script><\/p>\n\n\n<p>You can play with the <a href=\"https:\/\/sam2.metademolab.com\/demo\">demo<\/a> and learn more on the <a href=\"https:\/\/ai.meta.com\/SAM2\/\">site<\/a>:<\/p>\n\n\n\n<ul>\n<li><em>Following up on the success of the\u00a0<a rel=\"noreferrer noopener\" href=\"https:\/\/ai.meta.com\/blog\/segment-anything-foundation-model-image-segmentation\/\" target=\"_blank\"><u>Meta Segment Anything Model<\/u><\/a>\u00a0(SAM) for images, we\u2019re releasing\u00a0<a rel=\"noreferrer noopener\" href=\"https:\/\/github.com\/facebookresearch\/segment-anything-2\" target=\"_blank\"><u>SAM 2<\/u><\/a>, a unified model for real-time promptable object segmentation in images and videos that achieves state-of-the-art performance.<\/em><\/li>\n\n\n\n<li><em>In keeping with\u00a0<a rel=\"noreferrer noopener\" href=\"https:\/\/about.fb.com\/news\/2024\/07\/open-source-ai-is-the-path-forward\/\" target=\"_blank\"><u>our approach to open science<\/u><\/a>, we\u2019re sharing the code and model weights with a permissive Apache 2.0 license.<\/em><\/li>\n\n\n\n<li><em>We\u2019re also sharing the\u00a0<a rel=\"noreferrer noopener\" href=\"https:\/\/ai.meta.com\/datasets\/segment-anything-video\" target=\"_blank\"><u>SA-V dataset<\/u><\/a>, which includes approximately 51,000 real-world videos and more than 600,000 masklets (spatio-temporal masks).<\/em><\/li>\n\n\n\n<li><em>SAM 2 can segment any object in any video or image\u2014even for objects and visual domains it has not seen previously, enabling a diverse range of use cases without custom adaptation.<\/em><\/li>\n<\/ul>\n","protected":false},"excerpt":{"rendered":"<p>Man, I&#8217;m old enough to remember rotoscoping video by hand\u2014a process that quickly made me want to jump right out a window. Years later, when we were working on realtime video segmentation at Google, I was so proud to show the tech to a bunch of high school design students\u2014only to have them shrug and [&hellip;]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"","sticky":false,"template":"","format":"standard","meta":[],"categories":[66],"tags":[],"_links":{"self":[{"href":"http:\/\/jnack.com\/blog\/wp-json\/wp\/v2\/posts\/22099"}],"collection":[{"href":"http:\/\/jnack.com\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"http:\/\/jnack.com\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"http:\/\/jnack.com\/blog\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"http:\/\/jnack.com\/blog\/wp-json\/wp\/v2\/comments?post=22099"}],"version-history":[{"count":1,"href":"http:\/\/jnack.com\/blog\/wp-json\/wp\/v2\/posts\/22099\/revisions"}],"predecessor-version":[{"id":22100,"href":"http:\/\/jnack.com\/blog\/wp-json\/wp\/v2\/posts\/22099\/revisions\/22100"}],"wp:attachment":[{"href":"http:\/\/jnack.com\/blog\/wp-json\/wp\/v2\/media?parent=22099"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"http:\/\/jnack.com\/blog\/wp-json\/wp\/v2\/categories?post=22099"},{"taxonomy":"post_tag","embeddable":true,"href":"http:\/\/jnack.com\/blog\/wp-json\/wp\/v2\/tags?post=22099"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}