PageSourceSearch

https://haruk1y.github.io/wiki/assets/js/200da040.adcf74e2.js

js haruk1y.github.io collected 2026-10-03 09:18:13 UTC 42,868 bytes, 1 lines download raw bytes

1"use strict";(self.webpackChunkwiki=self.webpackChunkwiki||[]).push([["263"],{82532(s,e,a){a.r(e),a.d(e,{metadata:()=>n,default:()=>d,frontMatter:()=>t,contentTitle:()=>m,toc:()=>r,assets:()=>c});var n=JSON.parse('{"id":"computer-vision/foundation-models/segment-anything/sam-2","title":"SAM 2","description":"SAM 2 \u306E image + video unified architecture\u3001memory attention\u3001video object segmentation \u3092\u6574\u7406\u3057\u307E\u3059\u3002","source":"@site/docs/computer-vision/foundation-models/segment-anything/sam-2.mdx","sourceDirName":"computer-vision/foundation-models/segment-anything","slug":"/computer-vision/foundation-models/segment-anything/sam-2","permalink":"/wiki/computer-vision/foundation-models/segment-anything/sam-2","draft":false,"unlisted":false,"tags":[],"version":"current","sidebarPosition":4,"frontMatter":{"sidebar_position":4,"title":"SAM 2","description":"SAM 2 \u306E image + video unified architecture\u3001memory attention\u3001video object segmentation \u3092\u6574\u7406\u3057\u307E\u3059\u3002"},"sidebar":"tutorialSidebar","previous":{"title":"SA-1B Dataset","permalink":"/wiki/computer-vision/foundation-models/segment-anything/sa-1b-dataset"},"next":{"title":"SAM 3","permalink":"/wiki/computer-vision/foundation-models/segment-anything/sam-3"}}'),i=a(74848),l=a(28453);let t={sidebar_position:4,title:"SAM 2",description:"SAM 2 \u306E image + video unified architecture\u3001memory attention\u3001video object segmentation \u3092\u6574\u7406\u3057\u307E\u3059\u3002"},m="SAM 2",c={},r=[{value:"\u4F55\u304C\u5909\u308F\u3063\u305F\u306E\u304B",id:"\u4F55\u304C\u5909\u308F\u3063\u305F\u306E\u304B",level:2},{value:"\u30A2\u30FC\u30AD\u30C6\u30AF\u30C1\u30E3\u306E\u76F4\u611F",id:"\u30A2\u30FC\u30AD\u30C6\u30AF\u30C1\u30E3\u306E\u76F4\u611F",level:2},{value:"Video Object Segmentation",id:"video-object-segmentation",level:2},{value:"3D Reconstruction \u3068\u306E\u95A2\u4FC2",id:"3d-reconstruction-\u3068\u306E\u95A2\u4FC2",level:2},{value:"SAM \u3068\u306E\u6BD4\u8F03",id:"sam-\u3068\u306E\u6BD4\u8F03",level:2},{value:"\u6570\u5F0F\u3067\u898B\u308B video segmentation \u306E\u6642\u9593\u65B9\u5411 propagation",id:"\u6570\u5F0F\u3067\u898B\u308B-video-segmentation-\u306E\u6642\u9593\u65B9\u5411-propagation",level:2},{value:"\u95A2\u9023\u30DA\u30FC\u30B8",id:"\u95A2\u9023\u30DA\u30FC\u30B8",level:2},{value:"\u4E3B\u306A\u30BD\u30FC\u30B9",id:"\u4E3B\u306A\u30BD\u30FC\u30B9",level:2}];function h(s){let e={a:"a",annotation:"annotation",h1:"h1",h2:"h2",header:"header",li:"li",math:"math",mermaid:"mermaid",mi:"mi",mn:"mn",mo:"mo",mover:"mover",mrow:"mrow",msub:"msub",msubsup:"msubsup",p:"p",semantics:"semantics",span:"span",strong:"strong",table:"table",tbody:"tbody",td:"td",th:"th",thead:"thead",tr:"tr",ul:"ul",...(0,l.R)(),...s.components};return(0,i.jsxs)(i.Fragment,{children:[(0,i.jsx)(e.header,{children:(0,i.jsx)(e.h1,{id:"sam-2",children:"SAM 2"})}),"\n",(0,i.jsxs)(e.p,{children:["SAM 2 \u306F\u3001SAM \u306E\u67A0\u7D44\u307F\u3092 ",(0,i.jsx)(e.strong,{children:"image \u3068 video \u306E\u4E21\u65B9"}),"\u306B\u62E1\u5F35\u3057\u305F promptable segmentation model \u3067\u3059\u3002Frame \u3054\u3068\u306B mask \u3092\u51FA\u3059\u3060\u3051\u3067\u306A\u304F\u3001prompt \u3055\u308C\u305F object \u3092",(0,i.jsx)(e.strong,{children:"\u6642\u9593\u65B9\u5411\u306B\u8FFD\u8DE1"}),"\u3067\u304D\u307E\u3059\u3002"]}),"\n",(0,i.jsx)(e.h2,{id:"\u4F55\u304C\u5909\u308F\u3063\u305F\u306E\u304B",children:"\u4F55\u304C\u5909\u308F\u3063\u305F\u306E\u304B"}),"\n",(0,i.jsx)(e.p,{children:"SAM \u306F image \u3092 1 \u679A\u305A\u3064\u72EC\u7ACB\u306B\u6271\u3044\u307E\u3057\u305F\u3002SAM 2 \u3067\u306F\u3001"}),"\n",(0,i.jsxs)(e.ul,{children:["\n",(0,i.jsx)(e.li,{children:"Video \u3092\u4E00\u9023\u306E frame \u3068\u3057\u3066\u6271\u3046"}),"\n",(0,i.jsx)(e.li,{children:"\u3069\u3053\u304B\u306E frame \u3067 prompt \u3092\u4E0E\u3048\u308B\u3068\u3001\u4ED6\u306E frame \u3067\u3082\u540C\u3058 object \u306E mask \u3092\u51FA\u3059"}),"\n",(0,i.jsxs)(e.li,{children:["\u6642\u9593\u65B9\u5411\u306E\u4F9D\u5B58\u3092 ",(0,i.jsx)(e.strong,{children:"memory attention"})," \u3067\u6271\u3046"]}),"\n"]}),"\n",(0,i.jsx)(e.p,{children:"\u3068\u3044\u3046\u62E1\u5F35\u304C\u52A0\u308F\u3063\u3066\u3044\u307E\u3059\u3002Image \u3082 video \u3082\u540C\u3058 model \u3067\u6271\u3048\u308B\u3001unified \u306A\u8A2D\u8A08\u3067\u3059\u3002"}),"\n",(0,i.jsx)(e.h2,{id:"\u30A2\u30FC\u30AD\u30C6\u30AF\u30C1\u30E3\u306E\u76F4\u611F",children:"\u30A2\u30FC\u30AD\u30C6\u30AF\u30C1\u30E3\u306E\u76F4\u611F"}),"\n",(0,i.jsx)(e.mermaid,{value:"flowchart LR\n  F[Video frames] --\x3e E[Image encoder]\n  P[Prompt on a frame] --\x3e PE[Prompt encoder]\n  E --\x3e MA[Memory attention]\n  PE --\x3e MA\n  MEM[Memory bank] --\x3e MA\n  MA --\x3e D[Mask decoder]\n  D --\x3e M[Per-frame mask]\n  M --\x3e MEM"}),"\n",(0,i.jsxs)(e.ul,{children:["\n",(0,i.jsx)(e.li,{children:"\u5404 frame \u306F image encoder \u3067 embedding \u306B\u306A\u308A\u307E\u3059\u3002"}),"\n",(0,i.jsx)(e.li,{children:"Memory bank \u306F\u3001\u904E\u53BB frame \u306E mask \u3084 embedding \u3092\u4FDD\u5B58\u3057\u307E\u3059\u3002"}),"\n",(0,i.jsx)(e.li,{children:"Memory attention \u306B\u3088\u3063\u3066\u3001\u73FE\u5728 frame \u306E\u4E88\u6E2C\u304C\u904E\u53BB frame \u306E\u60C5\u5831\u3092\u53C2\u7167\u3057\u307E\u3059\u3002"}),"\n",(0,i.jsx)(e.li,{children:"\u51FA\u529B mask \u306F memory bank \u306B\u30D5\u30A3\u30FC\u30C9\u30D0\u30C3\u30AF\u3055\u308C\u3001\u6B21 frame \u3067\u518D\u5229\u7528\u3055\u308C\u307E\u3059\u3002"}),"\n"]}),"\n",(0,i.jsx)(e.p,{children:"\u3053\u308C\u306B\u3088\u3063\u3066\u3001object \u306E\u898B\u305F\u76EE\u304C\u6642\u9593\u3068\u3068\u3082\u306B\u5909\u308F\u3063\u3066\u3082\u3001tracking \u304C\u7D99\u7D9A\u3057\u3084\u3059\u304F\u306A\u308A\u307E\u3059\u3002"}),"\n",(0,i.jsx)(e.h2,{id:"video-object-segmentation",children:"Video Object Segmentation"}),"\n",(0,i.jsx)(e.p,{children:"SAM 2 \u304C\u89E3\u3044\u3066\u3044\u308B task \u306F\u3001video object segmentation \u306E\u6C4E\u7528\u7248\u3068\u307F\u306A\u305B\u307E\u3059\u3002"}),"\n",(0,i.jsxs)(e.table,{children:[(0,i.jsx)(e.thead,{children:(0,i.jsxs)(e.tr,{children:[(0,i.jsx)(e.th,{children:"\u8A2D\u5B9A"}),(0,i.jsx)(e.th,{children:"\u8AAC\u660E"})]})}),(0,i.jsxs)(e.tbody,{children:[(0,i.jsxs)(e.tr,{children:[(0,i.jsx)(e.td,{children:"Click prompt"}),(0,i.jsx)(e.td,{children:"\u3042\u308B frame \u3067\u5BFE\u8C61\u3092 click\u3002Model \u306F\u5168 frame \u3067\u305D\u306E object \u306E mask \u3092\u8FD4\u3059\u3002"})]}),(0,i.jsxs)(e.tr,{children:[(0,i.jsx)(e.td,{children:"Box prompt"}),(0,i.jsx)(e.td,{children:"Bounding box \u3067\u6307\u5B9A\u3002"})]}),(0,i.jsxs)(e.tr,{children:[(0,i.jsx)(e.td,{children:"Mask prompt"}),(0,i.jsx)(e.td,{children:"\u65E2\u5B58 mask \u3092 prompt \u3068\u3059\u308B (refine \u7528)\u3002"})]}),(0,i.jsxs)(e.tr,{children:[(0,i.jsx)(e.td,{children:"Multi-object"}),(0,i.jsx)(e.td,{children:"\u8907\u6570 object \u3092\u5225\u3005\u306B segment / track\u3002"})]})]})]}),"\n",(0,i.jsx)(e.h2,{id:"3d-reconstruction-\u3068\u306E\u95A2\u4FC2",children:"3D Reconstruction \u3068\u306E\u95A2\u4FC2"}),"\n",(0,i.jsx)(e.p,{children:"Video \u304C\u5165\u529B\u306B\u306A\u308B\u3053\u3068\u3067\u3001SAM 2 \u306F 3D Reconstruction \u3068\u7279\u306B\u76F8\u6027\u304C\u826F\u304F\u306A\u308A\u307E\u3059\u3002"}),"\n",(0,i.jsxs)(e.ul,{children:["\n",(0,i.jsx)(e.li,{children:"Multi-view image \u304C\u6642\u9593 sequence \u3067\u6765\u308B\u5834\u5408\u3001SAM 2 \u3067 view \u6A2A\u65AD\u306E object identity \u3092\u4FDD\u3066\u308B"}),"\n",(0,i.jsx)(e.li,{children:"Dynamic scene reconstruction \u3067\u3001moving object \u3092 tracking \u3057\u306A\u304C\u3089\u9664\u5916\u3067\u304D\u308B"}),"\n",(0,i.jsx)(e.li,{children:"Video \u304B\u3089\u306E NeRF / 3DGS training \u3067\u3001object mask \u306E time consistency \u3092\u4FDD\u3066\u308B"}),"\n"]}),"\n",(0,i.jsx)(e.mermaid,{value:"flowchart LR\n  V[Video] --\x3e S2[SAM 2]\n  S2 --\x3e M[Time-consistent masks]\n  M --\x3e R[Dynamic-aware 3D reconstruction]"}),"\n",(0,i.jsx)(e.h2,{id:"sam-\u3068\u306E\u6BD4\u8F03",children:"SAM \u3068\u306E\u6BD4\u8F03"}),"\n",(0,i.jsxs)(e.table,{children:[(0,i.jsx)(e.thead,{children:(0,i.jsxs)(e.tr,{children:[(0,i.jsx)(e.th,{children:"\u89B3\u70B9"}),(0,i.jsx)(e.th,{children:"SAM"}),(0,i.jsx)(e.th,{children:"SAM 2"})]})}),(0,i.jsxs)(e.tbody,{children:[(0,i.jsxs)(e.tr,{children:[(0,i.jsx)(e.td,{children:"\u5BFE\u8C61"}),(0,i.jsx)(e.td,{children:"Image"}),(0,i.jsx)(e.td,{children:"Image + Video"})]}),(0,i.jsxs)(e.tr,{children:[(0,i.jsx)(e.td,{children:"Tracking"}),(0,i.jsx)(e.td,{children:"\u306A\u3057"}),(0,i.jsx)(e.td,{children:"\u3042\u308A"})]}),(0,i.jsxs)(e.tr,{children:[(0,i.jsx)(e.td,{children:"Memory"}),(0,i.jsx)(e.td,{children:"\u306A\u3057"}),(0,i.jsx)(e.td,{children:"Memory attention"})]}),(0,i.jsxs)(e.tr,{children:[(0,i.jsx)(e.td,{children:"Prompt"}),(0,i.jsx)(e.td,{children:"Visual"}),(0,i.jsx)(e.td,{children:"Visual"})]}),(0,i.jsxs)(e.tr,{children:[(0,i.jsx)(e.td,{children:"\u7528\u9014"}),(0,i.jsx)(e.td,{children:"Interactive segmentation"}),(0,i.jsx)(e.td,{children:"Interactive segmentation + tracking"})]})]})]}),"\n",(0,i.jsx)(e.h2,{id:"\u6570\u5F0F\u3067\u898B\u308B-video-segmentation-\u306E\u6642\u9593\u65B9\u54
111-propagation",children:"\u6570\u5F0F\u3067\u898B\u308B video segmentation \u306E\u6642\u9593\u65B9\u5411 propagation"}),"\n",(0,i.jsxs)(e.p,{children:["SAM 2 \u306F\u3001video ",(0,i.jsxs)(e.span,{className:"katex",children:[(0,i.jsx)(e.span,{className:"katex-mathml",children:(0,i.jsx)(e.math,{xmlns:"http://www.w3.org/1998/Math/MathML",children:(0,i.jsxs)(e.semantics,{children:[(0,i.jsxs)(e.mrow,{children:[(0,i.jsx)(e.mi,{children:"V"}),(0,i.jsx)(e.mo,{children:"="}),(0,i.jsx)(e.mo,{stretchy:"false",children:"("}),(0,i.jsxs)(e.msub,{children:[(0,i.jsx)(e.mi,{children:"I"}),(0,i.jsx)(e.mn,{children:"1"})]}),(0,i.jsx)(e.mo,{separator:"true",children:","}),(0,i.jsx)(e.mo,{children:"\u2026"}),(0,i.jsx)(e.mo,{separator:"true",children:","}),(0,i.jsxs)(e.msub,{children:[(0,i.jsx)(e.mi,{children:"I"}),(0,i.jsx)(e.mi,{children:"T"})]}),(0,i.jsx)(e.mo,{stretchy:"false",children:")"})]}),(0,i.jsx)(e.annotation,{encoding:"application/x-tex",children:"V=(I_1,\\ldots,I_T)"})]})})}),(0,i.jsxs)(e.span,{className:"katex-html","aria-hidden":"true",children:[(0,i.jsxs)(e.span,{className:"base",children:[(0,i.jsx)(e.span,{className:"strut",style:{height:"0.6833em"}}),(0,i.jsx)(e.span,{className:"mord mathnormal",style:{marginRight:"0.2222em"},children:"V"}),(0,i.jsx)(e.span,{className:"mspace",style:{marginRight:"0.2778em"}}),(0,i.jsx)(e.span,{className:"mrel",children:"="}),(0,i.jsx)(e.span,{className:"mspace",style:{marginRight:"0.2778em"}})]}),(0,i.jsxs)(e.span,{className:"base",children:[(0,i.jsx)(e.span,{className:"strut",style:{height:"1em",verticalAlign:"-0.25em"}}),(0,i.jsx)(e.span,{className:"mopen",children:"("}),(0,i.jsxs)(e.span,{className:"mord",children:[(0,i.jsx)(e.span,{className:"mord mathnormal",style:{marginRight:"0.0785em"},children:"I"}),(0,i.jsx)(e.span,{className:"msupsub",children:(0,i.jsxs)(e.span,{className:"vlist-t vlist-t2",children:[(0,i.jsxs)(e.span,{className:"vlist-r",children:[(0,i.jsx)(e.span,{className:"vlist",style:{height:"0.3011em"},children:(0,i.jsxs)(e.span,{style:{top:"-2.55em",marginLeft:"-0.0785em",marginRight:"0.05em"},children:[(0,i.jsx)(e.span,{className:"pstrut",style:{height:"2.7em"}}),(0,i.jsx)(e.span,{className:"sizing reset-size6 size3 mtight",children:(0,i.jsx)(e.span,{className:"mord mtight",children:"1"})})]})}),(0,i.jsx)(e.span,{className:"vlist-s",children:"\u200B"})]}),(0,i.jsx)(e.span,{className:"vlist-r",children:(0,i.jsx)(e.span,{className:"vlist",style:{height:"0.15em"},children:(0,i.jsx)(e.span,{})})})]})})]}),(0,i.jsx)(e.span,{className:"mpunct",children:","}),(0,i.jsx)(e.span,{className:"mspace",style:{marginRight:"0.1667em"}}),(0,i.jsx)(e.span,{className:"minner",children:"\u2026"}),(0,i.jsx)(e.span,{className:"mspace",style:{marginRight:"0.1667em"}}),(0,i.jsx)(e.span,{className:"mpunct",children:","}),(0,i.jsx)(e.span,{className:"mspace",style:{marginRight:"0.1667em"}}),(0,i.jsxs)(e.span,{className:"mord",children:[(0,i.jsx)(e.span,{className:"mord mathnormal",style:{marginRight:"0.0785em"},children:"I"}),(0,i.jsx)(e.span,{className:"msupsub",children:(0,i.jsxs)(e.span,{className:"vlist-t vlist-t2",children:[(0,i.jsxs)(e.span,{className:"vlist-r",children:[(0,i.jsx)(e.span,{className:"vlist",style:{height:"0.3283em"},children:(0,i.jsxs)(e.span,{style:{top:"-2.55em",marginLeft:"-0.0785em",marginRight:"0.05em"},children:[(0,i.jsx)(e.span,{className:"pstrut",style:{height:"2.7em"}}),(0,i.jsx)(e.span,{className:"sizing reset-size6 size3 mtight",children:(0,i.jsx)(e.span,{className:"mord mathnormal mtight",style:{marginRight:"0.1389em"},children:"T"})})]})}),(0,i.jsx)(e.span,{className:"vlist-s",children:"\u200B"})]}),(0,i.jsx)(e.span,{className:"vlist-r",children:(0,i.jsx)(e.span,{className:"vlist",style:{height:"0.15em"},children:(0,i.jsx)(e.span,{})})})]})})]}),(0,i.jsx)(e.span,{className:"mclose",children:")"})]})]})]})," \u3068\u4EFB\u610F\u30D5\u30EC\u30FC\u30E0\u3067\u306E prompt ",(0,i.jsxs)(e.span,{className:"katex",children:[(0,i.jsx)(e.span,{className:"katex-mathml",children:(0,i.jsx)(e.math,{xmlns:"http://www.w3.org/1998/Math/MathML",children:(0,i.jsxs)(e.semantics,{children:[(0,i.jsx)(e.mrow,{children:(0,i.jsxs)(e.msub,{children:[(0,i.jsx)(e.mi,{children:"p"}),(0,i.jsxs)(e.msub,{children:[(0,i.jsx)(e.mi,{children:"t"}),(0,i.jsx)(e.mn,{children:"0"})]})]})}),(0,i.jsx)(e.annotation,{encoding:"application/x-tex",children:"p_{t_0}"})]})})}),(0,i.jsx)(e.span,{className:"katex-html","aria-hidden":"true",children:(0,i.jsxs)(e.span,{className:"base",children:[(0,i.jsx)(e.span,{className:"strut",style:{height:"0.6807em",verticalAlign:"-0.2501em"}}),(0,i.jsxs)(e.span,{className:"mord",children:[(0,i.jsx)(e.span,{className:"mord mathnormal",children:"p"}),(0,i.jsx)(e.span,{className:"msupsub",children:(0,i.jsxs)(e.span,{className:"vlist-t vlist-t2",children:[(0,i.jsxs)(e.span,{className:"vlist-r",children:[(0,i.jsx)(e.span,{className:"vlist",style:{height:"0.2806em"}
1,children:(0,i.jsxs)(e.span,{style:{top:"-2.55em",marginLeft:"0em",marginRight:"0.05em"},children:[(0,i.jsx)(e.span,{className:"pstrut",style:{height:"2.7em"}}),(0,i.jsx)(e.span,{className:"sizing reset-size6 size3 mtight",children:(0,i.jsx)(e.span,{className:"mord mtight",children:(0,i.jsxs)(e.span,{className:"mord mtight",children:[(0,i.jsx)(e.span,{className:"mord mathnormal mtight",children:"t"}),(0,i.jsx)(e.span,{className:"msupsub",children:(0,i.jsxs)(e.span,{className:"vlist-t vlist-t2",children:[(0,i.jsxs)(e.span,{className:"vlist-r",children:[(0,i.jsx)(e.span,{className:"vlist",style:{height:"0.3173em"},children:(0,i.jsxs)(e.span,{style:{top:"-2.357em",marginLeft:"0em",marginRight:"0.0714em"},children:[(0,i.jsx)(e.span,{className:"pstrut",style:{height:"2.5em"}}),(0,i.jsx)(e.span,{className:"sizing reset-size3 size1 mtight",children:(0,i.jsx)(e.span,{className:"mord mtight",children:"0"})})]})}),(0,i.jsx)(e.span,{className:"vlist-s",children:"\u200B"})]}),(0,i.jsx)(e.span,{className:"vlist-r",children:(0,i.jsx)(e.span,{className:"vlist",style:{height:"0.143em"},children:(0,i.jsx)(e.span,{})})})]})})]})})})]})}),(0,i.jsx)(e.span,{className:"vlist-s",children:"\u200B"})]}),(0,i.jsx)(e.span,{className:"vlist-r",children:(0,i.jsx)(e.span,{className:"vlist",style:{height:"0.2501em"},children:(0,i.jsx)(e.span,{})})})]})})]})]})})]})," \u3092\u53D7\u3051\u53D6\u308A\u3001\u5404\u30D5\u30EC\u30FC\u30E0\u306E mask \u7CFB\u5217\u3092\u51FA\u529B\u3057\u307E\u3059\u3002"]}),"\n",(0,i.jsx)(e.span,{className:"katex-display",children:(0,i.jsxs)(e.span,{className:"katex",children:[(0,i.jsx)(e.span,{className:"katex-mathml",children:(0,i.jsx)(e.math,{xmlns:"http://www.w3.org/1998/Math/MathML",display:"block",children:(0,i.jsxs)(e.semantics,{children:[(0,i.jsxs)(e.mrow,{children:[(0,i.jsx)(e.mo,{stretchy:"false",children:"{"}),(0,i.jsxs)(e.msub,{children:[(0,i.jsxs)(e.mover,{accent:"true",children:[(0,i.jsx)(e.mi,{children:"M"}),(0,i.jsx)(e.mo,{children:"^"})]}),(0,i.jsx)(e.mi,{children:"t"})]}),(0,i.jsxs)(e.msubsup,{children:[(0,i.jsx)(e.mo,{stretchy:"false",children:"}"}),(0,i.jsxs)(e.mrow,{children:[(0,i.jsx)(e.mi,{children:"t"}),(0,i.jsx)(e.mo,{children:"="}),(0,i.jsx)(e.mn,{children:"1"})]}),(0,i.jsx)(e.mi,{children:"T"})]}),(0,i.jsx)(e.mo,{children:"="}),(0,i.jsxs)(e.msub,{children:[(0,i.jsx)(e.mi,{children:"f"}),(0,i.jsx)(e.mi,{children:"\u03B8"})]}),(0,i.jsx)(e.mo,{stretchy:"false",children:"("}),(0,i.jsx)(e.mi,{children:"V"}),(0,i.jsx)(e.mo,{separator:"true",children:","}),(0,i.jsxs)(e.msub,{children:[(0,i.jsx)(e.mi,{children:"p"}),(0,i.jsxs)(e.msub,{children:[(0,i.jsx)(e.mi,{children:"t"}),(0,i.jsx)(e.mn,{children:"0"})]})]}),(0,i.jsx)(e.mo,{stretchy:"false",children:")"})]}),(0,i.jsx)(e.annotation,{encoding:"application/x-tex",children:"\\{\\hat{M}_t\\}_{t=1}^{T}=f_\\theta(V,p_{t_0})"})]})})}),(0,i.jsxs)(e.span,{className:"katex-html","aria-hidden":"true",children:[(0,i.jsxs)(e.span,{className:"base",children:[(0,i.jsx)(e.span,{className:"strut",style:{height:"1.1968em",verticalAlign:"-0.25em"}}),(0,i.jsx)(e.span,{className:"mopen",children:"{"}),(0,i.jsxs)(e.span,{className:"mord",children:[(0,i.jsx)(e.span,{className:"mord accent",children:(0,i.jsx)(e.span,{className:"vlist-t",children:(0,i.jsx)(e.span,{className:"vlist-r",children:(0,i.jsxs)(e.span,{className:"vlist",style:{height:"0.9468em"},children:[(0,i.jsxs)(e.span,{style:{top:"-3em"},children:[(0,i.jsx)(e.span,{className:"pstrut",style:{height:"3em"}}),(0,i.jsx)(e.span,{className:"mord mathnormal",style:{marginRight:"0.109em"},children:"M"})]}),(0,i.jsxs)(e.span,{style:{top:"-3.2523em"},children:[(0,i.jsx)(e.span,{className:"pstrut",style:{height:"3em"}}),(0,i.jsx)(e.span,{className:"accent-body",style:{left:"-0.1667em"},children:(0,i.jsx)(e.span,{className:"mord",children:"^"})})]})]})})})}),(0,i.jsx)(e.span,{className:"msupsub",children:(0,i.jsxs)(e.span,{className:"vlist-t vlist-t2",children:[(0,i.jsxs)(e.span,{className:"vlist-r",children:[(0,i.jsx)(e.span,{className:"vlist",style:{height:"0.2806em"},children:(0,i.jsxs)(e.span,{style:{top:"-2.55em",marginLeft:"-0.109em",marginRight:"0.05em"},children:[(0,i.jsx)(e.span,{className:"pstrut",style:{height:"2.7em"}}),(0,i.jsx)(e.span,{className:"sizing reset-size6 size3 mtight",children:(0,i.jsx)(e.span,{className:"mord mathnormal mtight",children:"t"})})]})}),(0,i.jsx)(e.span,{className:"vlist-s",children:"\u200B"})]}),(0,i.jsx)(e.span,{className:"vlist-r",children:(0,i.jsx)(e.span,{className:"vlist",style:{height:"0.15em"},children:(0,i.jsx)(e.span,{})})})]})})]}),(0,i.jsxs)(e.span,{className:"mclose",children:[(0,i.jsx)(e.span,{className:"mclose",children:"}"}),(0,i.jsx)(e.span,{className:"msupsub",children:(0,i.jsxs)(e.span,{className:"vlist-t vlist-t2",children:[(0,i.jsxs)(e.span,{className:"vlist-r",children:[(0,i.jsxs)(e.span,{className:"vlist",style:{height:"0.8913em"}
1,children:[(0,i.jsxs)(e.span,{style:{top:"-2.453em",marginLeft:"0em",marginRight:"0.05em"},children:[(0,i.jsx)(e.span,{className:"pstrut",style:{height:"2.7em"}}),(0,i.jsx)(e.span,{className:"sizing reset-size6 size3 mtight",children:(0,i.jsxs)(e.span,{className:"mord mtight",children:[(0,i.jsx)(e.span,{className:"mord mathnormal mtight",children:"t"}),(0,i.jsx)(e.span,{className:"mrel mtight",children:"="}),(0,i.jsx)(e.span,{className:"mord mtight",children:"1"})]})})]}),(0,i.jsxs)(e.span,{style:{top:"-3.113em",marginRight:"0.05em"},children:[(0,i.jsx)(e.span,{className:"pstrut",style:{height:"2.7em"}}),(0,i.jsx)(e.span,{className:"sizing reset-size6 size3 mtight",children:(0,i.jsx)(e.span,{className:"mord mtight",children:(0,i.jsx)(e.span,{className:"mord mathnormal mtight",style:{marginRight:"0.1389em"},children:"T"})})})]})]}),(0,i.jsx)(e.span,{className:"vlist-s",children:"\u200B"})]}),(0,i.jsx)(e.span,{className:"vlist-r",children:(0,i.jsx)(e.span,{className:"vlist",style:{height:"0.247em"},children:(0,i.jsx)(e.span,{})})})]})})]}),(0,i.jsx)(e.span,{className:"mspace",style:{marginRight:"0.2778em"}}),(0,i.jsx)(e.span,{className:"mrel",children:"="}),(0,i.jsx)(e.span,{className:"mspace",style:{marginRight:"0.2778em"}})]}),(0,i.jsxs)(e.span,{className:"base",children:[(0,i.jsx)(e.span,{className:"strut",style:{height:"1.0001em",verticalAlign:"-0.2501em"}}),(0,i.jsxs)(e.span,{className:"mord",children:[(0,i.jsx)(e.span,{className:"mord mathnormal",style:{marginRight:"0.1076em"},children:"f"}),(0,i.jsx)(e.span,{className:"msupsub",children:(0,i.jsxs)(e.span,{className:"vlist-t vlist-t2",children:[(0,i.jsxs)(e.span,{className:"vlist-r",children:[(0,i.jsx)(e.span,{className:"vlist",style:{height:"0.3361em"},children:(0,i.jsxs)(e.span,{style:{top:"-2.55em",marginLeft:"-0.1076em",marginRight:"0.05em"},children:[(0,i.jsx)(e.span,{className:"pstrut",style:{height:"2.7em"}}),(0,i.jsx)(e.span,{className:"sizing reset-size6 size3 mtight",children:(0,i.jsx)(e.span,{className:"mord mathnormal mtight",style:{marginRight:"0.0278em"},children:"\u03B8"})})]})}),(0,i.jsx)(e.span,{className:"vlist-s",children:"\u200B"})]}),(0,i.jsx)(e.span,{className:"vlist-r",children:(0,i.jsx)(e.span,{className:"vlist",style:{height:"0.15em"},children:(0,i.jsx)(e.span,{})})})]})})]}),(0,i.jsx)(e.span,{className:"mopen",children:"("}),(0,i.jsx)(e.span,{className:"mord mathnormal",style:{marginRight:"0.2222em"},children:"V"}),(0,i.jsx)(e.span,{className:"mpunct",children:","}),(0,i.jsx)(e.span,{className:"mspace",style:{marginRight:"0.1667em"}}),(0,i.jsxs)(e.span,{className:"mord",children:[(0,i.jsx)(e.span,{className:"mord mathnormal",children:"p"}),(0,i.jsx)(e.span,{className:"msupsub",children:(0,i.jsxs)(e.span,{className:"vlist-t vlist-t2",children:[(0,i.jsxs)(e.span,{className:"vlist-r",children:[(0,i.jsx)(e.span,{className:"vlist",style:{height:"0.2806em"},children:(0,i.jsxs)(e.span,{style:{top:"-2.55em",marginLeft:"0em",marginRight:"0.05em"},children:[(0,i.jsx)(e.span,{className:"pstrut",style:{height:"2.7em"}}),(0,i.jsx)(e.span,{className:"sizing reset-size6 size3 mtight",children:(0,i.jsx)(e.span,{className:"mord mtight",children:(0,i.jsxs)(e.span,{className:"mord mtight",children:[(0,i.jsx)(e.span,{className:"mord mathnormal mtight",children:"t"}),(0,i.jsx)(e.span,{className:"msupsub",children:(0,i.jsxs)(e.span,{className:"vlist-t vlist-t2",children:[(0,i.jsxs)(e.span,{className:"vlist-r",children:[(0,i.jsx)(e.span,{className:"vlist",style:{height:"0.3173em"},children:(0,i.jsxs)(e.span,{style:{top:"-2.357em",marginLeft:"0em",marginRight:"0.0714em"},children:[(0,i.jsx)(e.span,{className:"pstrut",style:{height:"2.5em"}}),(0,i.jsx)(e.span,{className:"sizing reset-size3 size1 mtight",children:(0,i.jsx)(e.span,{className:"mord mtight",children:"0"})})]})}),(0,i.jsx)(e.span,{className:"vlist-s",children:"\u200B"})]}),(0,i.jsx)(e.span,{className:"vlist-r",children:(0,i.jsx)(e.span,{className:"vlist",style:{height:"0.143em"},children:(0,i.jsx)(e.span,{})})})]})})]})})})]})}),(0,i.jsx)(e.span,{className:"vlist-s",children:"\u200B"})]}),(0,i.jsx)(e.span,{className:"vlist-r",children:(0,i.jsx)(e.span,{className:"vlist",style:{height:"0.2501em"},children:(0,i.jsx)(e.span,{})})})]})})]}),(0,i.jsx)(e.span,{className:"mclose",children:")"})]})]})]})}
1),"\n",(0,i.jsxs)(e.p,{children:["\u6642\u9593\u65B9\u5411\u306E\u6574\u5408\u6027\u306F\u3001memory bank ",(0,i.jsxs)(e.span,{className:"katex",children:[(0,i.jsx)(e.span,{className:"katex-mathml",children:(0,i.jsx)(e.math,{xmlns:"http://www.w3.org/1998/Math/MathML",children:(0,i.jsxs)(e.semantics,{children:[(0,i.jsx)(e.mrow,{children:(0,i.jsxs)(e.msub,{children:[(0,i.jsx)(e.mi,{mathvariant:"script",children:"M"}),(0,i.jsx)(e.mi,{children:"t"})]})}),(0,i.jsx)(e.annotation,{encoding:"application/x-tex",children:"\\mathcal{M}_t"})]})})}),(0,i.jsx)(e.span,{className:"katex-html","aria-hidden":"true",children:(0,i.jsxs)(e.span,{className:"base",children:[(0,i.jsx)(e.span,{className:"strut",style:{height:"0.8333em",verticalAlign:"-0.15em"}}),(0,i.jsxs)(e.span,{className:"mord",children:[(0,i.jsx)(e.span,{className:"mord mathcal",children:"M"}),(0,i.jsx)(e.span,{className:"msupsub",children:(0,i.jsxs)(e.span,{className:"vlist-t vlist-t2",children:[(0,i.jsxs)(e.span,{className:"vlist-r",children:[(0,i.jsx)(e.span,{className:"vlist",style:{height:"0.2806em"},children:(0,i.jsxs)(e.span,{style:{top:"-2.55em",marginLeft:"0em",marginRight:"0.05em"},children:[(0,i.jsx)(e.span,{className:"pstrut",style:{height:"2.7em"}}),(0,i.jsx)(e.span,{className:"sizing reset-size6 size3 mtight",children:(0,i.jsx)(e.span,{className:"mord mathnormal mtight",children:"t"})})]})}),(0,i.jsx)(e.span,{className:"vlist-s",children:"\u200B"})]}),(0,i.jsx)(e.span,{className:"vlist-r",children:(0,i.jsx)(e.span,{className:"vlist",style:{height:"0.15em"},children:(0,i.jsx)(e.span,{})})})]})})]})]})})]})," \u306B\u904E\u53BB\u306E mask \u60C5\u5831\u3092\u84C4\u7A4D\u3059\u308B\u3053\u3068\u3067\u4FDD\u305F\u308C\u307E\u3059\u3002"]}),"\n",(0,i.jsx)(e.span,{className:"katex-display",children:(0,i.jsxs)(e.span,{className:"katex",children:[(0,i.jsx)(e.span,{className:"katex-mathml",children:(0,i.jsx)(e.math,{xmlns:"http://www.w3.org/1998/Math/MathML",display:"block",children:(0,i.jsxs)(e.semantics,{children:[(0,i.jsxs)(e.mrow,{children:[(0,i.jsxs)(e.msub,{children:[(0,i.jsx)(e.mi,{mathvariant:"script",children:"M"}),(0,i.jsx)(e.mi,{children:"t"})]}),(0,i.jsx)(e.mo,{children:"="}),(0,i.jsxs)(e.mrow,{children:[(0,i.jsx)(e.mi,{mathvariant:"normal",children:"U"}),(0,i.jsx)(e.mi,{mathvariant:"normal",children:"p"}),(0,i.jsx)(e.mi,{mathvariant:"normal",children:"d"}),(0,i.jsx)(e.mi,{mathvariant:"normal",children:"a"}),(0,i.jsx)(e.mi,{mathvariant:"normal",children:"t"}),(0,i.jsx)(e.mi,{mathvariant:"normal",children:"e"})]}),(0,i.jsx)(e.mo,{stretchy:"false",children:"("}),(0,i.jsxs)(e.msub,{children:[(0,i.jsx)(e.mi,{mathvariant:"script",children:"M"}),(0,i.jsxs)(e.mrow,{children:[(0,i.jsx)(e.mi,{children:"t"}),(0,i.jsx)(e.mo,{children:"\u2212"}),(0,i.jsx)(e.mn,{children:"1"})]})]}),(0,i.jsx)(e.mo,{separator:"true",children:","}),(0,i.jsxs)(e.msub,{children:[(0,i.jsxs)(e.mover,{accent:"true",children:[(0,i.jsx)(e.mi,{children:"M"}),(0,i.jsx)(e.mo,{children:"^"})]}),(0,i.jsxs)(e.mrow,{children:[(0,i.jsx)(e.mi,{children:"t"}),(0,i.jsx)(e.mo,{children:"\u2212"}),(0,i.jsx)(e.mn,{children:"1"})]})]}),(0,i.jsx)(e.mo,{separator:"true",children:","}),(0,i.jsxs)(e.msub,{children:[(0,i.jsx)(e.mi,{mathvariant:"bold",children:"f"}),(0,i.jsxs)(e.mrow,{children:[(0,i.jsx)(e.mi,{children:"t"}),(0,i.jsx)(e.mo,{children:"\u2212"}),(0,i.jsx)(e.mn,{children:"1"})]})]}),(0,i.jsx)(e.mo,{stretchy:"false",children:")"})]}),(0,i.jsx)(e.annotation,{encoding:"application/x-tex",children:"\\mathcal{M}_t=\\mathrm{Update}(\\mathcal{M}_{t-1},\\hat{M}_{t-1},\\mathbf{f}_{t-1})"})]})})}),(0,i.jsxs)(e.span,{className:"katex-html","aria-hidden":"true",children:[(0,i.jsxs)(e.span,{className:"base",children:[(0,i.jsx)(e.span,{className:"strut",style:{height:"0.8333em",verticalAlign:"-0.15em"}}),(0,i.jsxs)(e.span,{className:"mord",children:[(0,i.jsx)(e.span,{className:"mord mathcal",children:"M"}),(0,i.jsx)(e.span,{className:"msupsub",children:(0,i.jsxs)(e.span,{className:"vlist-t vlist-t2",children:[(0,i.jsxs)(e.span,{className:"vlist-r",children:[(0,i.jsx)(e.span,{className:"vlist",style:{height:"0.2806em"}
1,children:(0,i.jsxs)(e.span,{style:{top:"-2.55em",marginLeft:"0em",marginRight:"0.05em"},children:[(0,i.jsx)(e.span,{className:"pstrut",style:{height:"2.7em"}}),(0,i.jsx)(e.span,{className:"sizing reset-size6 size3 mtight",children:(0,i.jsx)(e.span,{className:"mord mathnormal mtight",children:"t"})})]})}),(0,i.jsx)(e.span,{className:"vlist-s",children:"\u200B"})]}),(0,i.jsx)(e.span,{className:"vlist-r",children:(0,i.jsx)(e.span,{className:"vlist",style:{height:"0.15em"},children:(0,i.jsx)(e.span,{})})})]})})]}),(0,i.jsx)(e.span,{className:"mspace",style:{marginRight:"0.2778em"}}),(0,i.jsx)(e.span,{className:"mrel",children:"="}),(0,i.jsx)(e.span,{className:"mspace",style:{marginRight:"0.2778em"}})]}),(0,i.jsxs)(e.span,{className:"base",children:[(0,i.jsx)(e.span,{className:"strut",style:{height:"1.1968em",verticalAlign:"-0.25em"}}),(0,i.jsx)(e.span,{className:"mord",children:(0,i.jsx)(e.span,{className:"mord mathrm",children:"Update"})}),(0,i.jsx)(e.span,{className:"mopen",children:"("}),(0,i.jsxs)(e.span,{className:"mord",children:[(0,i.jsx)(e.span,{className:"mord mathcal",children:"M"}),(0,i.jsx)(e.span,{className:"msupsub",children:(0,i.jsxs)(e.span,{className:"vlist-t vlist-t2",children:[(0,i.jsxs)(e.span,{className:"vlist-r",children:[(0,i.jsx)(e.span,{className:"vlist",style:{height:"0.3011em"},children:(0,i.jsxs)(e.span,{style:{top:"-2.55em",marginLeft:"0em",marginRight:"0.05em"},children:[(0,i.jsx)(e.span,{className:"pstrut",style:{height:"2.7em"}}),(0,i.jsx)(e.span,{className:"sizing reset-size6 size3 mtight",children:(0,i.jsxs)(e.span,{className:"mord mtight",children:[(0,i.jsx)(e.span,{className:"mord mathnormal mtight",children:"t"}),(0,i.jsx)(e.span,{className:"mbin mtight",children:"\u2212"}),(0,i.jsx)(e.span,{className:"mord mtight",children:"1"})]})})]})}),(0,i.jsx)(e.span,{className:"vlist-s",children:"\u200B"})]}),(0,i.jsx)(e.span,{className:"vlist-r",children:(0,i.jsx)(e.span,{className:"vlist",style:{height:"0.2083em"},children:(0,i.jsx)(e.span,{})})})]})})]}),(0,i.jsx)(e.span,{className:"mpunct",children:","}),(0,i.jsx)(e.span,{className:"mspace",style:{marginRight:"0.1667em"}}),(0,i.jsxs)(e.span,{className:"mord",children:[(0,i.jsx)(e.span,{className:"mord accent",children:(0,i.jsx)(e.span,{className:"vlist-t",children:(0,i.jsx)(e.span,{className:"vlist-r",children:(0,i.jsxs)(e.span,{className:"vlist",style:{height:"0.9468em"},children:[(0,i.jsxs)(e.span,{style:{top:"-3em"},children:[(0,i.jsx)(e.span,{className:"pstrut",style:{height:"3em"}}),(0,i.jsx)(e.span,{className:"mord mathnormal",style:{marginRight:"0.109em"},children:"M"})]}),(0,i.jsxs)(e.span,{style:{top:"-3.2523em"},children:[(0,i.jsx)(e.span,{className:"pstrut",style:{height:"3em"}}),(0,i.jsx)(e.span,{className:"accent-body",style:{left:"-0.1667em"},children:(0,i.jsx)(e.span,{className:"mord",children:"^"})})]})]})})})}),(0,i.jsx)(e.span,{className:"msupsub",children:(0,i.jsxs)(e.span,{className:"vlist-t vlist-t2",children:[(0,i.jsxs)(e.span,{className:"vlist-r",children:[(0,i.jsx)(e.span,{className:"vlist",style:{height:"0.3011em"},children:(0,i.jsxs)(e.span,{style:{top:"-2.55em",marginLeft:"-0.109em",marginRight:"0.05em"},children:[(0,i.jsx)(e.span,{className:"pstrut",style:{height:"2.7em"}}),(0,i.jsx)(e.span,{className:"sizing reset-size6 size3 mtight",children:(0,i.jsxs)(e.span,{className:"mord mtight",children:[(0,i.jsx)(e.span,{className:"mord mathnormal mtight",children:"t"}),(0,i.jsx)(e.span,{className:"mbin mtight",children:"\u2212"}),(0,i.jsx)(e.span,{className:"mord mtight",children:"1"})]})})]})}),(0,i.jsx)(e.span,{className:"vlist-s",children:"\u200B"})]}),(0,i.jsx)(e.span,{className:"vlist-r",children:(0,i.jsx)(e.span,{className:"vlist",style:{height:"0.2083em"},children:(0,i.jsx)(e.span,{})})})]})})]}),(0,i.jsx)(e.span,{className:"mpunct",children:","}),(0,i.jsx)(e.span,{className:"mspace",style:{marginRight:"0.1667em"}}),(0,i.jsxs)(e.span,{className:"mord",children:[(0,i.jsx)(e.span,{className:"mord mathbf",style:{marginRight:"0.109em"},children:"f"}),(0,i.jsx)(e.span,{className:"msupsub",children:(0,i.jsxs)(e.span,{className:"vlist-t vlist-t2",children:[(0,i.jsxs)(e.span,{className:"vlist-r",children:[(0,i.jsx)(e.span,{className:"vlist",style:{height:"0.3011em"}
1,children:(0,i.jsxs)(e.span,{style:{top:"-2.55em",marginLeft:"-0.109em",marginRight:"0.05em"},children:[(0,i.jsx)(e.span,{className:"pstrut",style:{height:"2.7em"}}),(0,i.jsx)(e.span,{className:"sizing reset-size6 size3 mtight",children:(0,i.jsxs)(e.span,{className:"mord mtight",children:[(0,i.jsx)(e.span,{className:"mord mathnormal mtight",children:"t"}),(0,i.jsx)(e.span,{className:"mbin mtight",children:"\u2212"}),(0,i.jsx)(e.span,{className:"mord mtight",children:"1"})]})})]})}),(0,i.jsx)(e.span,{className:"vlist-s",children:"\u200B"})]}),(0,i.jsx)(e.span,{className:"vlist-r",children:(0,i.jsx)(e.span,{className:"vlist",style:{height:"0.2083em"},children:(0,i.jsx)(e.span,{})})})]})})]}),(0,i.jsx)(e.span,{className:"mclose",children:")"})]})]})]})}),"\n",(0,i.jsxs)(e.p,{children:["\u3053\u3053\u3067\u3001",(0,i.jsxs)(e.span,{className:"katex",children:[(0,i.jsx)(e.span,{className:"katex-mathml",children:(0,i.jsx)(e.math,{xmlns:"http://www.w3.org/1998/Math/MathML",children:(0,i.jsxs)(e.semantics,{children:[(0,i.jsx)(e.mrow,{children:(0,i.jsxs)(e.msub,{children:[(0,i.jsx)(e.mi,{mathvariant:"bold",children:"f"}),(0,i.jsxs)(e.mrow,{children:[(0,i.jsx)(e.mi,{children:"t"}),(0,i.jsx)(e.mo,{children:"\u2212"}),(0,i.jsx)(e.mn,{children:"1"})]})]})}),(0,i.jsx)(e.annotation,{encoding:"application/x-tex",children:"\\mathbf{f}_{t-1}"})]})})}),(0,i.jsx)(e.span,{className:"katex-html","aria-hidden":"true",children:(0,i.jsxs)(e.span,{className:"base",children:[(0,i.jsx)(e.span,{className:"strut",style:{height:"0.9028em",verticalAlign:"-0.2083em"}}),(0,i.jsxs)(e.span,{className:"mord",children:[(0,i.jsx)(e.span,{className:"mord mathbf",style:{marginRight:"0.109em"},children:"f"}),(0,i.jsx)(e.span,{className:"msupsub",children:(0,i.jsxs)(e.span,{className:"vlist-t vlist-t2",children:[(0,i.jsxs)(e.span,{className:"vlist-r",children:[(0,i.jsx)(e.span,{className:"vlist",style:{height:"0.3011em"},children:(0,i.jsxs)(e.span,{style:{top:"-2.55em",marginLeft:"-0.109em",marginRight:"0.05em"},children:[(0,i.jsx)(e.span,{className:"pstrut",style:{height:"2.7em"}}),(0,i.jsx)(e.span,{className:"sizing reset-size6 size3 mtight",children:(0,i.jsxs)(e.span,{className:"mord mtight",children:[(0,i.jsx)(e.span,{className:"mord mathnormal mtight",children:"t"}),(0,i.jsx)(e.span,{className:"mbin mtight",children:"\u2212"}),(0,i.jsx)(e.span,{className:"mord mtight",children:"1"})]})})]})}),(0,i.jsx)(e.span,{className:"vlist-s",children:"\u200B"})]}),(0,i.jsx)(e.span,{className:"vlist-r",children:(0,i.jsx)(e.span,{className:"vlist",style:{height:"0.2083em"},children:(0,i.jsx)(e.span,{})})})]})})]})]})})]})," \u306F frame \u306E visual feature \u3067\u3059\u3002\u6B21\u30D5\u30EC\u30FC\u30E0\u306E\u4E88\u6E2C\u306F\u3001\u73FE\u5728\u30D5\u30EC\u30FC\u30E0\u306E feature \u3068 memory \u3092 attention \u3067\u53C2\u7167\u3057\u306A\u304C\u3089\u884C\u3044\u307E\u3059\u3002"]}),"\n",(0,i.jsx)(e.span,{className:"katex-display",children:(0,i.jsxs)(e.span,{className:"katex",children:[(0,i.jsx)(e.span,{className:"katex-mathml",children:(0,i.jsx)(e.math,{xmlns:"http://www.w3.org/1998/Math/MathML",display:"block",children:(0,i.jsxs)(e.semantics,{children:[(0,i.jsxs)(e.mrow,{children:[(0,i.jsxs)(e.msub,{children:[(0,i.jsxs)(e.mover,{accent:"true",children:[(0,i.jsx)(e.mi,{children:"M"}),(0,i.jsx)(e.mo,{children:"^"})]}),(0,i.jsx)(e.mi,{children:"t"})]}),(0,i.jsx)(e.mo,{children:"="}),(0,i.jsxs)(e.msub,{children:[(0,i.jsx)(e.mi,{children:"g"}),(0,i.jsx)(e.mi,{children:"\u03B8"})]}),(0,i.jsx)(e.mo,{stretchy:"false",children:"("}),(0,i.jsxs)(e.msub,{children:[(0,i.jsx)(e.mi,{mathvariant:"bold",children:"f"}),(0,i.jsx)(e.mi,{children:"t"})]}),(0,i.jsx)(e.mo,{separator:"true",children:","}),(0,i.jsxs)(e.msub,{children:[(0,i.jsx)(e.mi,{mathvariant:"script",children:"M"}),(0,i.jsx)(e.mi,{children:"t"})]}),(0,i.jsx)(e.mo,{stretchy:"false",children:")"})]}),(0,i.jsx)(e.annotation,{encoding:"application/x-tex",children:"\\hat{M}_t=g_\\theta(\\mathbf{f}_t,\\mathcal{M}_t)"})]})})}),(0,i.jsxs)(e.span,{className:"katex-html","aria-hidden":"true",children:[(0,i.jsxs)(e.span,{className:"base",children:[(0,i.jsx)(e.span,{className:"strut",style:{height:"1.0968em",verticalAlign:"-0.15em"}}),(0,i.jsxs)(e.span,{className:"mord",children:[(0,i.jsx)(e.span,{className:"mord accent",children:(0,i.jsx)(e.span,{className:"vlist-t",children:(0,i.jsx)(e.span,{className:"vlist-r",children:(0,i.jsxs)(e.span,{className:"vlist",style:{height:"0.9468em"}
1,children:[(0,i.jsxs)(e.span,{style:{top:"-3em"},children:[(0,i.jsx)(e.span,{className:"pstrut",style:{height:"3em"}}),(0,i.jsx)(e.span,{className:"mord mathnormal",style:{marginRight:"0.109em"},children:"M"})]}),(0,i.jsxs)(e.span,{style:{top:"-3.2523em"},children:[(0,i.jsx)(e.span,{className:"pstrut",style:{height:"3em"}}),(0,i.jsx)(e.span,{className:"accent-body",style:{left:"-0.1667em"},children:(0,i.jsx)(e.span,{className:"mord",children:"^"})})]})]})})})}),(0,i.jsx)(e.span,{className:"msupsub",children:(0,i.jsxs)(e.span,{className:"vlist-t vlist-t2",children:[(0,i.jsxs)(e.span,{className:"vlist-r",children:[(0,i.jsx)(e.span,{className:"vlist",style:{height:"0.2806em"},children:(0,i.jsxs)(e.span,{style:{top:"-2.55em",marginLeft:"-0.109em",marginRight:"0.05em"},children:[(0,i.jsx)(e.span,{className:"pstrut",style:{height:"2.7em"}}),(0,i.jsx)(e.span,{className:"sizing reset-size6 size3 mtight",children:(0,i.jsx)(e.span,{className:"mord mathnormal mtight",children:"t"})})]})}),(0,i.jsx)(e.span,{className:"vlist-s",children:"\u200B"})]}),(0,i.jsx)(e.span,{className:"vlist-r",children:(0,i.jsx)(e.span,{className:"vlist",style:{height:"0.15em"},children:(0,i.jsx)(e.span,{})})})]})})]}),(0,i.jsx)(e.span,{className:"mspace",style:{marginRight:"0.2778em"}}),(0,i.jsx)(e.span,{className:"mrel",children:"="}),(0,i.jsx)(e.span,{className:"mspace",style:{marginRight:"0.2778em"}})]}),(0,i.jsxs)(e.span,{className:"base",children:[(0,i.jsx)(e.span,{className:"strut",style:{height:"1em",verticalAlign:"-0.25em"}}),(0,i.jsxs)(e.span,{className:"mord",children:[(0,i.jsx)(e.span,{className:"mord mathnormal",style:{marginRight:"0.0359em"},children:"g"}),(0,i.jsx)(e.span,{className:"msupsub",children:(0,i.jsxs)(e.span,{className:"vlist-t vlist-t2",children:[(0,i.jsxs)(e.span,{className:"vlist-r",children:[(0,i.jsx)(e.span,{className:"vlist",style:{height:"0.3361em"},children:(0,i.jsxs)(e.span,{style:{top:"-2.55em",marginLeft:"-0.0359em",marginRight:"0.05em"},children:[(0,i.jsx)(e.span,{className:"pstrut",style:{height:"2.7em"}}),(0,i.jsx)(e.span,{className:"sizing reset-size6 size3 mtight",children:(0,i.jsx)(e.span,{className:"mord mathnormal mtight",style:{marginRight:"0.0278em"},children:"\u03B8"})})]})}),(0,i.jsx)(e.span,{className:"vlist-s",children:"\u200B"})]}),(0,i.jsx)(e.span,{className:"vlist-r",children:(0,i.jsx)(e.span,{className:"vlist",style:{height:"0.15em"},children:(0,i.jsx)(e.span,{})})})]})})]}),(0,i.jsx)(e.span,{className:"mopen",children:"("}),(0,i.jsxs)(e.span,{className:"mord",children:[(0,i.jsx)(e.span,{className:"mord mathbf",style:{marginRight:"0.109em"},children:"f"}),(0,i.jsx)(e.span,{className:"msupsub",children:(0,i.jsxs)(e.span,{className:"vlist-t vlist-t2",children:[(0,i.jsxs)(e.span,{className:"vlist-r",children:[(0,i.jsx)(e.span,{className:"vlist",style:{height:"0.2806em"},children:(0,i.jsxs)(e.span,{style:{top:"-2.55em",marginLeft:"-0.109em",marginRight:"0.05em"},children:[(0,i.jsx)(e.span,{className:"pstrut",style:{height:"2.7em"}}),(0,i.jsx)(e.span,{className:"sizing reset-size6 size3 mtight",children:(0,i.jsx)(e.span,{className:"mord mathnormal mtight",children:"t"})})]})}),(0,i.jsx)(e.span,{className:"vlist-s",children:"\u200B"})]}),(0,i.jsx)(e.span,{className:"vlist-r",children:(0,i.jsx)(e.span,{className:"vlist",style:{height:"0.15em"},children:(0,i.jsx)(e.span,{})})})]})})]}),(0,i.jsx)(e.span,{className:"mpunct",children:","}),(0,i.jsx)(e.span,{className:"mspace",style:{marginRight:"0.1667em"}}),(0,i.jsxs)(e.span,{className:"mord",children:[(0,i.jsx)(e.span,{className:"mord mathcal",children:"M"}),(0,i.jsx)(e.span,{className:"msupsub",children:(0,i.jsxs)(e.span,{className:"vlist-t vlist-t2",children:[(0,i.jsxs)(e.span,{className:"vlist-r",children:[(0,i.jsx)(e.span,{className:"vlist",style:{height:"0.2806em"},children:(0,i.jsxs)(e.span,{style:{top:"-2.55em",marginLeft:"0em",marginRight:"0.05em"},children:[(0,i.jsx)(e.span,{className:"pstrut",style:{height:"2.7em"}}),(0,i.jsx)(e.span,{className:"sizing reset-size6 size3 mtight",children:(0,i.jsx)(e.span,{className:"mord mathnormal mtight",children:"t"})})]})}),(0,i.jsx)(e.span,{className:"vlist-s",children:"\u200B"})]}),(0,i.jsx)(e.span,{className:"vlist-r",children:(0,i.jsx)(e.span,{className:"vlist",style:{height:"0.15em"},children:(0,i.jsx)(e.span,{})})})]})})]}),(0,i.jsx)(e.span,{className:"mclose",children:")"})]})]})]})}),"\n",(0,i.jsx)(e.p,{children:"\u3053\u306E\u5F0F\u306E\u6C17\u6301\u3061\u306F\u3001\u300Cobject \u306E\u898B\u305F\u76EE\u3068\u904E\u53BB\u306E mask \u3092 memory \u306B\u6301\u3061\u3001\u5404\u30D5\u30EC\u30FC\u30E0\u3067\u305D\u308C\u3092\u53C2\u7167\u3059\u308B\u3053\u3068\u3067\u3001occlusion \u3084\u898B\u3048\u65
1B9\u306E\u5909\u5316\u306B\u5BFE\u3057\u3066\u3082\u540C\u3058 object \u3092\u8FFD\u3044\u7D9A\u3051\u308B\u300D\u3068\u3044\u3046\u3082\u306E\u3067\u3059\u3002"}),"\n",(0,i.jsx)(e.h2,{id:"\u95A2\u9023\u30DA\u30FC\u30B8",children:"\u95A2\u9023\u30DA\u30FC\u30B8"}),"\n",(0,i.jsxs)(e.ul,{children:["\n",(0,i.jsx)(e.li,{children:(0,i.jsx)(e.a,{href:"/wiki/computer-vision/foundation-models/segment-anything/sam",children:"SAM"})}),"\n",(0,i.jsx)(e.li,{children:(0,i.jsx)(e.a,{href:"/wiki/computer-vision/foundation-models/segment-anything/sam-3",children:"SAM 3"})}),"\n",(0,i.jsx)(e.li,{children:(0,i.jsx)(e.a,{href:"/wiki/computer-vision/3d-reconstruction/segmentation-for-3d-reconstruction",children:"Segmentation for 3D Reconstruction"})}),"\n",(0,i.jsx)(e.li,{children:(0,i.jsx)(e.a,{href:"/wiki/machine-learning/generative-models/video-diffusion-models/",children:"Video Diffusion Models"})}),"\n"]}),"\n",(0,i.jsx)(e.h2,{id:"\u4E3B\u306A\u30BD\u30FC\u30B9",children:"\u4E3B\u306A\u30BD\u30FC\u30B9"}),"\n",(0,i.jsxs)(e.ul,{children:["\n",(0,i.jsxs)(e.li,{children:["SAM 2 official page: ",(0,i.jsx)(e.a,{href:"https://ai.meta.com/sam2/",children:"https://ai.meta.com/sam2/"})]}),"\n",(0,i.jsxs)(e.li,{children:["SAM 2 announcement: ",(0,i.jsx)(e.a,{href:"https://about.fb.com/news/2024/07/our-new-ai-model-can-segment-video/",children:"https://about.fb.com/news/2024/07/our-new-ai-model-can-segment-video/"})]}),"\n"]})]})}function d(s={}){let{wrapper:e}={...(0,l.R)(),...s.components};return e?(0,i.jsx)(e,{...s,children:(0,i.jsx)(h,{...s})}):h(s)}},28453(s,e,a){a.d(e,{R:()=>t,x:()=>m});var n=a(96540);let i={},l=n.createContext(i);function t(s){let e=n.useContext(l);return n.useMemo(function(){return"function"==typeof s?s(e):{...e,...s}},[e,s])}function m(s){let e;return e=s.disableParentContext?"function"==typeof s.components?s.components(i):s.components||i:t(s.components),n.createElement(l.Provider,{value:e},s.children)}}}]);

Line numbers count LF bytes from the start of the resource, as the search results do. Vendor segments are library code the classifier recognised; they are stored but not indexed. Bytes are shown as Latin1 characters, one per byte.