{"version":2,"relationshipURLBase":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p39.html","relationshipResolution":"Resolve relative relationship URLs against relationshipURLBase; complete records and absolute URLs are in blocks.","title":"两个同类对象，需要两份身份","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p39.html","markdown":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p39.md","pages":[{"id":"instance-versus-class-map","title":"从同类像素中分出两份对象身份","unit":"怎样让每一份轮廓属于一个明确对象？","unitId":"instance-mask-identity","chapter":"II","number":1},{"id":"set-matching-lab20-costs","title":"选读 LAB20：沿真实输入核对交叉匹配","unit":"选读：输出组织与训练分配还可以怎样改变？","unitId":"spatial-training-extensions","chapter":"II","number":2}],"aliases":[{"oldId":"explanation","targetId":"instance-versus-class-map","operation":"split-explanation-range"},{"oldId":"b-4d23eebbe635b1d919f7d8fc","targetId":"b-4d23eebbe635b1d919f7d8fc","page":"instance-versus-class-map","operation":"move","beforeSHA256":"d71cf15e707c8a3674b4c90a599efee783cb84da5b8c8236a1d01c3add7b4958","afterSHA256":"d71cf15e707c8a3674b4c90a599efee783cb84da5b8c8236a1d01c3add7b4958","reason":"Same source body in authored semantic page"},{"oldId":"b-0412a3bb64e1ace807f26bd6","targetId":"b-0412a3bb64e1ace807f26bd6","page":"instance-versus-class-map","operation":"move","beforeSHA256":"3a13e7e9a0b2407c1dd6b294484c71eef52e61d88d24e2935400ab9c406b500f","afterSHA256":"3a13e7e9a0b2407c1dd6b294484c71eef52e61d88d24e2935400ab9c406b500f","reason":"Same source body in authored semantic page"},{"oldId":"b-c26c26751bb6d4781f96ffd3","targetId":"b-c26c26751bb6d4781f96ffd3","page":"instance-versus-class-map","operation":"move","beforeSHA256":"711e1c9e1612c33ddf45e0a8f3dde8414e5dc22c8e70bd3ae8f1db7fd11f7b86","afterSHA256":"711e1c9e1612c33ddf45e0a8f3dde8414e5dc22c8e70bd3ae8f1db7fd11f7b86","reason":"Same source body in authored semantic page"},{"oldId":"b-57f74fcc8a17019f50203bae","targetId":"b-57f74fcc8a17019f50203bae","page":"instance-versus-class-map","operation":"rewrite","beforeSHA256":"1980fdcfae762c642452b9ad2dd4dfca2a80605b993d5f5ad2eeb2d656662a1a","afterSHA256":"4810660828940d374c07b4e630634bbd1406dcea869ff3ec9c7bc772bc69db42","reason":"Same source identity with figure semantics or reading context enriched"},{"oldId":"b-da000b1998e8c35c1aef81ac","targetId":"b-da000b1998e8c35c1aef81ac","page":"instance-versus-class-map","operation":"move","beforeSHA256":"786ce10a6b32a492259a3dccd0a54e61cd41d28a34382fb7b5386c5f03b7ad69","afterSHA256":"786ce10a6b32a492259a3dccd0a54e61cd41d28a34382fb7b5386c5f03b7ad69","reason":"Same source body in authored semantic page"},{"oldId":"b-41aabf18f2b542a5f6425433","targetId":"b-41aabf18f2b542a5f6425433","page":"instance-versus-class-map","operation":"move","beforeSHA256":"c05ef7ef11d553d5e7172bb4312790b351c4a6161519fb73371065d2a290d1c7","afterSHA256":"c05ef7ef11d553d5e7172bb4312790b351c4a6161519fb73371065d2a290d1c7","reason":"Same source body in authored semantic page"},{"oldId":"b-abfb2f9a193a67b53b9878ca","targetId":"b-abfb2f9a193a67b53b9878ca","page":"instance-versus-class-map","operation":"move","beforeSHA256":"367067e2a804b2adca4a44a30e5daddedf52eca68328d0291a72c6190be006f7","afterSHA256":"367067e2a804b2adca4a44a30e5daddedf52eca68328d0291a72c6190be006f7","reason":"Same source body in authored semantic page"},{"oldId":"b-927cc6b5a3ed04540ca1f759","targetId":"b-927cc6b5a3ed04540ca1f759","page":"instance-versus-class-map","operation":"move","beforeSHA256":"62cd3c3e7472a4976a8c101608c68d04df6eb55e6ed8b1cdfe8b808e43846633","afterSHA256":"62cd3c3e7472a4976a8c101608c68d04df6eb55e6ed8b1cdfe8b808e43846633","reason":"Same source body in authored semantic page"},{"oldId":"b-d50ce0a78976580fa14adf75","targetId":"b-d50ce0a78976580fa14adf75","page":"instance-versus-class-map","operation":"move","beforeSHA256":"b77c69f411f0cec66906efa93953096b72fee0517dd04ceb2eb4f2c14fca2c6c","afterSHA256":"b77c69f411f0cec66906efa93953096b72fee0517dd04ceb2eb4f2c14fca2c6c","reason":"Same source body in authored semantic page"},{"oldId":"b-720ee824f7d183f4483a6bfa","targetId":"b-720ee824f7d183f4483a6bfa","page":"instance-versus-class-map","operation":"move","beforeSHA256":"df469b42a008cfb7ecd8a88fd7f7836c4bc0779d31028b4938705dc288908ff6","afterSHA256":"df469b42a008cfb7ecd8a88fd7f7836c4bc0779d31028b4938705dc288908ff6","reason":"Same source body in authored semantic page"},{"oldId":"b-09055f5a69416b665e247340","targetId":"b-09055f5a69416b665e247340","page":"instance-versus-class-map","operation":"move","beforeSHA256":"ea717067584e344418a930c49862e5bea35e54c3e75453a5dc35674aa7112bb5","afterSHA256":"ea717067584e344418a930c49862e5bea35e54c3e75453a5dc35674aa7112bb5","reason":"Same source body in authored semantic page"},{"oldId":"b-4f1991c943910c30887aa02d","targetId":"b-4f1991c943910c30887aa02d","page":"instance-versus-class-map","operation":"move","beforeSHA256":"0b72d1aeae2101b989af7a9f866b1e58740515b358603751717dfb93c0137160","afterSHA256":"0b72d1aeae2101b989af7a9f866b1e58740515b358603751717dfb93c0137160","reason":"Same source body in authored semantic page"},{"oldId":"b-a9ecbcc64a15a15c7e344e26","targetId":"b-a9ecbcc64a15a15c7e344e26","page":"set-matching-lab20-costs","operation":"move","beforeSHA256":"e22ba271dbf8de903d354b548af45e31549397c87cf88aa2ff55aab69ec47377","afterSHA256":"e22ba271dbf8de903d354b548af45e31549397c87cf88aa2ff55aab69ec47377","reason":"Same source body in authored semantic page"},{"oldId":"b-9257aab2323ee2dbc2d775c0","targetId":"b-9257aab2323ee2dbc2d775c0","page":"set-matching-lab20-costs","operation":"move","beforeSHA256":"8c4a4fcaec0e6c0eb856799d47828d49dcd70cbeddd2d724b59395207d95348e","afterSHA256":"8c4a4fcaec0e6c0eb856799d47828d49dcd70cbeddd2d724b59395207d95348e","reason":"Same source body in authored semantic page"},{"oldId":"b-656da9b2b7c7130e194f0a3a","targetId":"b-656da9b2b7c7130e194f0a3a","page":"set-matching-lab20-costs","operation":"move","beforeSHA256":"666afb23ae32517dfc3373217aa07120f72ac5e338277cf55cef59ec35e29c96","afterSHA256":"666afb23ae32517dfc3373217aa07120f72ac5e338277cf55cef59ec35e29c96","reason":"Same source body in authored semantic page"},{"oldId":"b-02509d3458e83596789f8ade","targetId":"b-02509d3458e83596789f8ade","page":"set-matching-lab20-costs","operation":"move","beforeSHA256":"00cdf3ea105cff573d3b5e30cd194b7c8bb22cc8b37fda235912d4f9ef325925","afterSHA256":"00cdf3ea105cff573d3b5e30cd194b7c8bb22cc8b37fda235912d4f9ef325925","reason":"Same source body in authored semantic page"},{"oldId":"b-9e42d2818afe818cac5ceb58","targetId":"b-9e42d2818afe818cac5ceb58","page":"set-matching-lab20-costs","operation":"move","beforeSHA256":"2c901bf98cd806f477e314f0897e9351b3172089a6eab29d4f2569837c1837e6","afterSHA256":"2c901bf98cd806f477e314f0897e9351b3172089a6eab29d4f2569837c1837e6","reason":"Same source body in authored semantic page"},{"oldId":"b-85268d7825036025b7af04df","targetId":"b-85268d7825036025b7af04df","page":"set-matching-lab20-costs","operation":"move","beforeSHA256":"cf44fcc62782fb4f6bdc42c16c5187d57556a03cb7ce6065e7bd9c36d8e7d733","afterSHA256":"cf44fcc62782fb4f6bdc42c16c5187d57556a03cb7ce6065e7bd9c36d8e7d733","reason":"Same source body in authored semantic page"},{"oldId":"b-df1402a24d369554ba2a5099","targetId":"b-df1402a24d369554ba2a5099","page":"set-matching-lab20-costs","operation":"move","beforeSHA256":"b8b69250624fd8c15e6f5db8690409848aaec027fc3b89222c6665c3044c9586","afterSHA256":"b8b69250624fd8c15e6f5db8690409848aaec027fc3b89222c6665c3044c9586","reason":"Same source body in authored semantic page"},{"oldId":"instance-versus-class-map-title","targetId":"b-a64098fd9b9f4366f7ce73ac","operation":"retain-source-identity"},{"oldId":"set-matching-lab20-costs-title","targetId":"b-dd495b78e581774c587f1841","operation":"retain-source-identity"},{"oldId":"set-matching-lab20-costs-bridge","targetId":"b-47bb2366f2357ac7518b002b","operation":"retain-source-identity"}],"blocks":[{"id":"b-a64098fd9b9f4366f7ce73ac","type":"h2","scope":"instance-versus-class-map","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p39.html#b-a64098fd9b9f4366f7ce73ac","content":"## 从同类像素中分出两份对象身份","sha256":"34a17a8e4150518eababa5b75d0f432506b2642254b86f0060d828ec876a82b2","chapter":{"id":"II","title":"区域、像素与实例"},"unit":{"id":"instance-mask-identity","title":"怎样让每一份轮廓属于一个明确对象？"},"page":{"id":"instance-versus-class-map","title":"从同类像素中分出两份对象身份","number":1},"sectionTitle":"从同类像素中分出两份对象身份","currentQuestion":"类别地图还缺哪种身份，区域 mask 如何补上它？","sourceSlides":["l3-r2-p39"],"knowledgeBlock":null,"formulas":[],"previousBlock":null,"nextBlock":"b-4d23eebbe635b1d919f7d8fc","prerequisites":[{"id":"b-c26c26751bb6d4781f96ffd3","url":"#b-c26c26751bb6d4781f96ffd3","role":"problem-context"}],"relatedBlocks":[{"id":"b-57f74fcc8a17019f50203bae","url":"#b-57f74fcc8a17019f50203bae","relation":"figure-and-explanation"}]},{"id":"b-4d23eebbe635b1d919f7d8fc","type":"h2","scope":"explanation","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p39.html#b-4d23eebbe635b1d919f7d8fc","content":"## 解释、算例与证据","sha256":"d71cf15e707c8a3674b4c90a599efee783cb84da5b8c8236a1d01c3add7b4958","chapter":{"id":"II","title":"区域、像素与实例"},"unit":{"id":"instance-mask-identity","title":"怎样让每一份轮廓属于一个明确对象？"},"page":{"id":"instance-versus-class-map","title":"从同类像素中分出两份对象身份","number":1},"sectionTitle":"从同类像素中分出两份对象身份","currentQuestion":"类别地图还缺哪种身份，区域 mask 如何补上它？","sourceSlides":["l3-r2-p39"],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-a64098fd9b9f4366f7ce73ac","nextBlock":"b-0412a3bb64e1ace807f26bd6","prerequisites":[{"id":"b-a64098fd9b9f4366f7ce73ac","url":"#b-a64098fd9b9f4366f7ce73ac","role":"section-title"},{"id":"b-c26c26751bb6d4781f96ffd3","url":"#b-c26c26751bb6d4781f96ffd3","role":"problem-context"}],"relatedBlocks":[{"id":"b-57f74fcc8a17019f50203bae","url":"#b-57f74fcc8a17019f50203bae","relation":"figure-and-explanation"}]},{"id":"b-0412a3bb64e1ace807f26bd6","type":"h1","scope":"document","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p39.html#b-0412a3bb64e1ace807f26bd6","content":"# P39 · 两个同类对象，需要两份身份","sha256":"3a13e7e9a0b2407c1dd6b294484c71eef52e61d88d24e2935400ab9c406b500f","chapter":{"id":"II","title":"区域、像素与实例"},"unit":{"id":"instance-mask-identity","title":"怎样让每一份轮廓属于一个明确对象？"},"page":{"id":"instance-versus-class-map","title":"从同类像素中分出两份对象身份","number":1},"sectionTitle":"从同类像素中分出两份对象身份","currentQuestion":"类别地图还缺哪种身份，区域 mask 如何补上它？","sourceSlides":["l3-r2-p39"],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-4d23eebbe635b1d919f7d8fc","nextBlock":"b-c26c26751bb6d4781f96ffd3","prerequisites":[{"id":"b-4d23eebbe635b1d919f7d8fc","url":"#b-4d23eebbe635b1d919f7d8fc","role":"section-title"},{"id":"b-c26c26751bb6d4781f96ffd3","url":"#b-c26c26751bb6d4781f96ffd3","role":"problem-context"}],"relatedBlocks":[{"id":"b-57f74fcc8a17019f50203bae","url":"#b-57f74fcc8a17019f50203bae","relation":"figure-and-explanation"}]},{"id":"b-c26c26751bb6d4781f96ffd3","type":"p","scope":"document","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p39.html#b-c26c26751bb6d4781f96ffd3","content":"同类像素与同一实例是两个概念。实例到类别的映射可多对一，连通块也不可靠地等于对象数。此页提出掩码必须随对象身份分别组织，为区域路线与像素路线的汇合提供具体理由。","sha256":"711e1c9e1612c33ddf45e0a8f3dde8414e5dc22c8e70bd3ae8f1db7fd11f7b86","chapter":{"id":"II","title":"区域、像素与实例"},"unit":{"id":"instance-mask-identity","title":"怎样让每一份轮廓属于一个明确对象？"},"page":{"id":"instance-versus-class-map","title":"从同类像素中分出两份对象身份","number":1},"sectionTitle":"从同类像素中分出两份对象身份","currentQuestion":"类别地图还缺哪种身份，区域 mask 如何补上它？","sourceSlides":["l3-r2-p39"],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-0412a3bb64e1ace807f26bd6","nextBlock":"b-57f74fcc8a17019f50203bae","prerequisites":[{"id":"b-0412a3bb64e1ace807f26bd6","url":"#b-0412a3bb64e1ace807f26bd6","role":"section-title"}],"relatedBlocks":[{"id":"b-57f74fcc8a17019f50203bae","url":"#b-57f74fcc8a17019f50203bae","relation":"figure-and-explanation"}]},{"id":"b-57f74fcc8a17019f50203bae","type":"figure","scope":"document","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p39.html#b-57f74fcc8a17019f50203bae","content":"图内标签与数值：Lecture 03 课堂图解 P39；同一底图，只改变标注问题；语义：两只都属于“杯子”；实例：#1 与 #2 分开；同色编码类别；不同色编码对象 ID；可见轮廓、杯柄孔洞和书遮挡保留；同类像素正确，还不等于交付了实例身份。\n\n同一底图并排叠加语义和实例 mask，杯柄孔洞和书遮挡保留；颜色分别编码类别和实例 ID。\n\n- 同一底图，只改变标注问题\n- 语义：两只都属于“杯子”\n- 实例：#1 与 #2 分开\n- 同色编码类别\n- 不同色编码对象 ID\n- 可见轮廓、杯柄孔洞和书遮挡保留；同类像素正确，还不等于交付了实例身份。同一底图并排叠加语义和实例 mask，杯柄孔洞和书遮挡保留；颜色分别编码类别和实例 ID。 [查看对应图解状态](/#/l3-r2-p39/0)[查看完整图解](/course/figures/l3-r2-p39-0.html)","sha256":"4810660828940d374c07b4e630634bbd1406dcea869ff3ec9c7bc772bc69db42","chapter":{"id":"II","title":"区域、像素与实例"},"unit":{"id":"instance-mask-identity","title":"怎样让每一份轮廓属于一个明确对象？"},"page":{"id":"instance-versus-class-map","title":"从同类像素中分出两份对象身份","number":1},"sectionTitle":"从同类像素中分出两份对象身份","currentQuestion":"类别地图还缺哪种身份，区域 mask 如何补上它？","sourceSlides":["l3-r2-p39"],"knowledgeBlock":null,"formulas":[],"figure":{"caption":"同一底图并排叠加语义和实例 mask，杯柄孔洞和书遮挡保留；颜色分别编码类别和实例 ID。 查看对应图解状态","labels":["同一底图，只改变标注问题","语义：两只都属于“杯子”","实例：#1 与 #2 分开","同色编码类别","不同色编码对象 ID","可见轮廓、杯柄孔洞和书遮挡保留；同类像素正确，还不等于交付了实例身份。"],"description":"同一底图并排叠加语义和实例 mask，杯柄孔洞和书遮挡保留；颜色分别编码类别和实例 ID。同一底图，只改变标注问题语义：两只都属于“杯子”实例：#1 与 #2 分开同色编码类别不同色编码对象 ID可见轮廓、杯柄孔洞和书遮挡保留；同类像素正确，还不等于交付了实例身份。","evidenceIdentity":"同一底图并排叠加语义和实例 mask，杯柄孔洞和书遮挡保留；颜色分别编码类别和实例 ID。 查看对应图解状态","images":[{"src":"/assets/display/8ced939a1fd55047fbf73ea5.webp","alt":""},{"src":"/assets/display/8ced939a1fd55047fbf73ea5.webp","alt":""}]},"previousBlock":"b-c26c26751bb6d4781f96ffd3","nextBlock":"b-da000b1998e8c35c1aef81ac","prerequisites":[{"id":"b-0412a3bb64e1ace807f26bd6","url":"#b-0412a3bb64e1ace807f26bd6","role":"section-title"},{"id":"b-c26c26751bb6d4781f96ffd3","url":"#b-c26c26751bb6d4781f96ffd3","role":"problem-context"}],"relatedBlocks":[]},{"id":"b-da000b1998e8c35c1aef81ac","type":"h2","scope":"narrative","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p39.html#b-da000b1998e8c35c1aef81ac","content":"## 课堂讲述","sha256":"786ce10a6b32a492259a3dccd0a54e61cd41d28a34382fb7b5386c5f03b7ad69","chapter":{"id":"II","title":"区域、像素与实例"},"unit":{"id":"instance-mask-identity","title":"怎样让每一份轮廓属于一个明确对象？"},"page":{"id":"instance-versus-class-map","title":"从同类像素中分出两份对象身份","number":1},"sectionTitle":"从同类像素中分出两份对象身份","currentQuestion":"类别地图还缺哪种身份，区域 mask 如何补上它？","sourceSlides":["l3-r2-p39"],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-57f74fcc8a17019f50203bae","nextBlock":"b-41aabf18f2b542a5f6425433","prerequisites":[{"id":"b-0412a3bb64e1ace807f26bd6","url":"#b-0412a3bb64e1ace807f26bd6","role":"section-title"},{"id":"b-c26c26751bb6d4781f96ffd3","url":"#b-c26c26751bb6d4781f96ffd3","role":"problem-context"}],"relatedBlocks":[{"id":"b-57f74fcc8a17019f50203bae","url":"#b-57f74fcc8a17019f50203bae","relation":"figure-and-explanation"}]},{"id":"b-41aabf18f2b542a5f6425433","type":"p","scope":"narrative","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p39.html#b-41aabf18f2b542a5f6425433","content":"现在看两只相邻的杯子。即使所有杯子像素都判对了，系统若只交付“杯子”这一类别，也还没有告诉我们每个像素属于哪一个对象。对语义类别来说，这未必是主要错误；对实例任务来说，答案就不够了。\n实例分割要保留“这是哪一个对象”的身份。不能只把所有同类像素放进一个桶，也不能只提供外接框。\n这让区域路线和像素路线有了自然的汇合点：先用区域组织不同对象，再在每个区域内部预测属于这个实例的像素。我们前面花时间讲区域表示和空间预测，到了这里就不再是两段独立知识。Mask R-CNN 把这两种要求放到了一套结构中。","sha256":"c05ef7ef11d553d5e7172bb4312790b351c4a6161519fb73371065d2a290d1c7","chapter":{"id":"II","title":"区域、像素与实例"},"unit":{"id":"instance-mask-identity","title":"怎样让每一份轮廓属于一个明确对象？"},"page":{"id":"instance-versus-class-map","title":"从同类像素中分出两份对象身份","number":1},"sectionTitle":"从同类像素中分出两份对象身份","currentQuestion":"类别地图还缺哪种身份，区域 mask 如何补上它？","sourceSlides":["l3-r2-p39"],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-da000b1998e8c35c1aef81ac","nextBlock":"b-abfb2f9a193a67b53b9878ca","prerequisites":[{"id":"b-da000b1998e8c35c1aef81ac","url":"#b-da000b1998e8c35c1aef81ac","role":"section-title"},{"id":"b-c26c26751bb6d4781f96ffd3","url":"#b-c26c26751bb6d4781f96ffd3","role":"problem-context"}],"relatedBlocks":[{"id":"b-57f74fcc8a17019f50203bae","url":"#b-57f74fcc8a17019f50203bae","relation":"figure-and-explanation"}]},{"id":"b-abfb2f9a193a67b53b9878ca","type":"h3","scope":"instance-identity-map","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p39.html#b-abfb2f9a193a67b53b9878ca","content":"### 从实例到类别容易，反过来不唯一","sha256":"367067e2a804b2adca4a44a30e5daddedf52eca68328d0291a72c6190be006f7","chapter":{"id":"II","title":"区域、像素与实例"},"unit":{"id":"instance-mask-identity","title":"怎样让每一份轮廓属于一个明确对象？"},"page":{"id":"instance-versus-class-map","title":"从同类像素中分出两份对象身份","number":1},"sectionTitle":"从实例到类别容易，反过来不唯一","currentQuestion":"类别地图还缺哪种身份，区域 mask 如何补上它？","sourceSlides":["l3-r2-p39"],"knowledgeBlock":"instance-identity-map","formulas":[],"previousBlock":"b-41aabf18f2b542a5f6425433","nextBlock":"b-927cc6b5a3ed04540ca1f759","prerequisites":[{"id":"b-da000b1998e8c35c1aef81ac","url":"#b-da000b1998e8c35c1aef81ac","role":"section-title"},{"id":"b-c26c26751bb6d4781f96ffd3","url":"#b-c26c26751bb6d4781f96ffd3","role":"problem-context"}],"relatedBlocks":[{"id":"b-57f74fcc8a17019f50203bae","url":"#b-57f74fcc8a17019f50203bae","relation":"figure-and-explanation"},{"id":"b-927cc6b5a3ed04540ca1f759","url":"#b-927cc6b5a3ed04540ca1f759","relation":"same-reasoning"},{"id":"b-d50ce0a78976580fa14adf75","url":"#b-d50ce0a78976580fa14adf75","relation":"same-reasoning"},{"id":"b-720ee824f7d183f4483a6bfa","url":"#b-720ee824f7d183f4483a6bfa","relation":"same-reasoning"},{"id":"b-09055f5a69416b665e247340","url":"#b-09055f5a69416b665e247340","relation":"same-reasoning"}]},{"id":"b-927cc6b5a3ed04540ca1f759","type":"p","scope":"instance-identity-map","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p39.html#b-927cc6b5a3ed04540ca1f759","content":"把像素标为杯子类别，只声明“这里属于杯子集合”。实例标注还声明“这里属于杯一，那里属于杯二”。两只杯子即使同色、接触甚至相互遮挡，实例任务仍需给出两份对象身份。","sha256":"62cd3c3e7472a4976a8c101608c68d04df6eb55e6ed8b1cdfe8b808e43846633","chapter":{"id":"II","title":"区域、像素与实例"},"unit":{"id":"instance-mask-identity","title":"怎样让每一份轮廓属于一个明确对象？"},"page":{"id":"instance-versus-class-map","title":"从同类像素中分出两份对象身份","number":1},"sectionTitle":"从实例到类别容易，反过来不唯一","currentQuestion":"类别地图还缺哪种身份，区域 mask 如何补上它？","sourceSlides":["l3-r2-p39"],"knowledgeBlock":"instance-identity-map","formulas":[],"previousBlock":"b-abfb2f9a193a67b53b9878ca","nextBlock":"b-d50ce0a78976580fa14adf75","prerequisites":[{"id":"b-abfb2f9a193a67b53b9878ca","url":"#b-abfb2f9a193a67b53b9878ca","role":"section-title"},{"id":"b-c26c26751bb6d4781f96ffd3","url":"#b-c26c26751bb6d4781f96ffd3","role":"problem-context"}],"relatedBlocks":[{"id":"b-57f74fcc8a17019f50203bae","url":"#b-57f74fcc8a17019f50203bae","relation":"figure-and-explanation"},{"id":"b-abfb2f9a193a67b53b9878ca","url":"#b-abfb2f9a193a67b53b9878ca","relation":"same-reasoning"},{"id":"b-d50ce0a78976580fa14adf75","url":"#b-d50ce0a78976580fa14adf75","relation":"same-reasoning"},{"id":"b-720ee824f7d183f4483a6bfa","url":"#b-720ee824f7d183f4483a6bfa","relation":"same-reasoning"},{"id":"b-09055f5a69416b665e247340","url":"#b-09055f5a69416b665e247340","relation":"same-reasoning"}]},{"id":"b-d50ce0a78976580fa14adf75","type":"p","scope":"instance-identity-map","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p39.html#b-d50ce0a78976580fa14adf75","content":"考虑一行标签 `[1,1,2,2]`，其中 1、2 是两只杯子的实例号。将它们都映射到类别杯子=1，就得到 `[1,1,1,1]`。另一个实例划分 `[1,2,2,2]` 经同样映射也得到同一类别行。因此类别图不能唯一反推出实例边界。","sha256":"b77c69f411f0cec66906efa93953096b72fee0517dd04ceb2eb4f2c14fca2c6c","chapter":{"id":"II","title":"区域、像素与实例"},"unit":{"id":"instance-mask-identity","title":"怎样让每一份轮廓属于一个明确对象？"},"page":{"id":"instance-versus-class-map","title":"从同类像素中分出两份对象身份","number":1},"sectionTitle":"从实例到类别容易，反过来不唯一","currentQuestion":"类别地图还缺哪种身份，区域 mask 如何补上它？","sourceSlides":["l3-r2-p39"],"knowledgeBlock":"instance-identity-map","formulas":[],"previousBlock":"b-927cc6b5a3ed04540ca1f759","nextBlock":"b-720ee824f7d183f4483a6bfa","prerequisites":[{"id":"b-abfb2f9a193a67b53b9878ca","url":"#b-abfb2f9a193a67b53b9878ca","role":"section-title"},{"id":"b-c26c26751bb6d4781f96ffd3","url":"#b-c26c26751bb6d4781f96ffd3","role":"problem-context"}],"relatedBlocks":[{"id":"b-57f74fcc8a17019f50203bae","url":"#b-57f74fcc8a17019f50203bae","relation":"figure-and-explanation"},{"id":"b-abfb2f9a193a67b53b9878ca","url":"#b-abfb2f9a193a67b53b9878ca","relation":"same-reasoning"},{"id":"b-927cc6b5a3ed04540ca1f759","url":"#b-927cc6b5a3ed04540ca1f759","relation":"same-reasoning"},{"id":"b-720ee824f7d183f4483a6bfa","url":"#b-720ee824f7d183f4483a6bfa","relation":"same-reasoning"},{"id":"b-09055f5a69416b665e247340","url":"#b-09055f5a69416b665e247340","relation":"same-reasoning"}]},{"id":"b-720ee824f7d183f4483a6bfa","type":"p","scope":"instance-identity-map","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p39.html#b-720ee824f7d183f4483a6bfa","content":"连通域可以在“每个对象互不接触”的特定条件下帮忙计数，但一旦两只杯子相接，一个连通块可含两个对象；反过来同一被遮挡的对象还可能产生不连通的可见部分。对象身份不是连通性规则自动保证的属性。","sha256":"df469b42a008cfb7ecd8a88fd7f7836c4bc0779d31028b4938705dc288908ff6","chapter":{"id":"II","title":"区域、像素与实例"},"unit":{"id":"instance-mask-identity","title":"怎样让每一份轮廓属于一个明确对象？"},"page":{"id":"instance-versus-class-map","title":"从同类像素中分出两份对象身份","number":1},"sectionTitle":"从实例到类别容易，反过来不唯一","currentQuestion":"类别地图还缺哪种身份，区域 mask 如何补上它？","sourceSlides":["l3-r2-p39"],"knowledgeBlock":"instance-identity-map","formulas":[],"previousBlock":"b-d50ce0a78976580fa14adf75","nextBlock":"b-09055f5a69416b665e247340","prerequisites":[{"id":"b-abfb2f9a193a67b53b9878ca","url":"#b-abfb2f9a193a67b53b9878ca","role":"section-title"},{"id":"b-c26c26751bb6d4781f96ffd3","url":"#b-c26c26751bb6d4781f96ffd3","role":"problem-context"}],"relatedBlocks":[{"id":"b-57f74fcc8a17019f50203bae","url":"#b-57f74fcc8a17019f50203bae","relation":"figure-and-explanation"},{"id":"b-abfb2f9a193a67b53b9878ca","url":"#b-abfb2f9a193a67b53b9878ca","relation":"same-reasoning"},{"id":"b-927cc6b5a3ed04540ca1f759","url":"#b-927cc6b5a3ed04540ca1f759","relation":"same-reasoning"},{"id":"b-d50ce0a78976580fa14adf75","url":"#b-d50ce0a78976580fa14adf75","relation":"same-reasoning"},{"id":"b-09055f5a69416b665e247340","url":"#b-09055f5a69416b665e247340","relation":"same-reasoning"}]},{"id":"b-09055f5a69416b665e247340","type":"p","scope":"instance-identity-map","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p39.html#b-09055f5a69416b665e247340","content":"区域路线给每个候选一份身份，像素路线给精细轮廓。把它们结合，就可以在“这个候选对象”之内预测局部 mask。这是进入 Mask R-CNN 的具体需求，而不是为了再增加一种模型名称。","sha256":"ea717067584e344418a930c49862e5bea35e54c3e75453a5dc35674aa7112bb5","chapter":{"id":"II","title":"区域、像素与实例"},"unit":{"id":"instance-mask-identity","title":"怎样让每一份轮廓属于一个明确对象？"},"page":{"id":"instance-versus-class-map","title":"从同类像素中分出两份对象身份","number":1},"sectionTitle":"从实例到类别容易，反过来不唯一","currentQuestion":"类别地图还缺哪种身份，区域 mask 如何补上它？","sourceSlides":["l3-r2-p39"],"knowledgeBlock":"instance-identity-map","formulas":[],"previousBlock":"b-720ee824f7d183f4483a6bfa","nextBlock":"b-f88908456f7c55a6e91f93a2","prerequisites":[{"id":"b-abfb2f9a193a67b53b9878ca","url":"#b-abfb2f9a193a67b53b9878ca","role":"section-title"},{"id":"b-c26c26751bb6d4781f96ffd3","url":"#b-c26c26751bb6d4781f96ffd3","role":"problem-context"}],"relatedBlocks":[{"id":"b-57f74fcc8a17019f50203bae","url":"#b-57f74fcc8a17019f50203bae","relation":"figure-and-explanation"},{"id":"b-abfb2f9a193a67b53b9878ca","url":"#b-abfb2f9a193a67b53b9878ca","relation":"same-reasoning"},{"id":"b-927cc6b5a3ed04540ca1f759","url":"#b-927cc6b5a3ed04540ca1f759","relation":"same-reasoning"},{"id":"b-d50ce0a78976580fa14adf75","url":"#b-d50ce0a78976580fa14adf75","relation":"same-reasoning"},{"id":"b-720ee824f7d183f4483a6bfa","url":"#b-720ee824f7d183f4483a6bfa","relation":"same-reasoning"}]},{"id":"b-f88908456f7c55a6e91f93a2","type":"h4","scope":"instance-versus-class-map","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p39.html#b-f88908456f7c55a6e91f93a2","content":"#### 图示条件与公式约定","sha256":"ef7f915c1ae63ed67bd23fecf050249273ba0efa8965d2c8d1c1c6b2d3178122","chapter":{"id":"II","title":"区域、像素与实例"},"unit":{"id":"instance-mask-identity","title":"怎样让每一份轮廓属于一个明确对象？"},"page":{"id":"instance-versus-class-map","title":"从同类像素中分出两份对象身份","number":1},"sectionTitle":"从同类像素中分出两份对象身份","currentQuestion":"类别地图还缺哪种身份，区域 mask 如何补上它？","sourceSlides":["l3-r2-p39"],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-09055f5a69416b665e247340","nextBlock":"b-4f1991c943910c30887aa02d","prerequisites":[{"id":"b-abfb2f9a193a67b53b9878ca","url":"#b-abfb2f9a193a67b53b9878ca","role":"section-title"},{"id":"b-c26c26751bb6d4781f96ffd3","url":"#b-c26c26751bb6d4781f96ffd3","role":"problem-context"}],"relatedBlocks":[{"id":"b-57f74fcc8a17019f50203bae","url":"#b-57f74fcc8a17019f50203bae","relation":"figure-and-explanation"}]},{"id":"b-4f1991c943910c30887aa02d","type":"p","scope":"document","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p39.html#b-4f1991c943910c30887aa02d","content":"原底图两杯相邻但没有相接，因此讲稿开头改成“现在看两只相邻的杯子”。它们若接触时语义同类与实例身份的差异仍成立，此页不假装底图已经接触。","sha256":"0b72d1aeae2101b989af7a9f866b1e58740515b358603751717dfb93c0137160","chapter":{"id":"II","title":"区域、像素与实例"},"unit":{"id":"instance-mask-identity","title":"怎样让每一份轮廓属于一个明确对象？"},"page":{"id":"instance-versus-class-map","title":"从同类像素中分出两份对象身份","number":1},"sectionTitle":"从同类像素中分出两份对象身份","currentQuestion":"类别地图还缺哪种身份，区域 mask 如何补上它？","sourceSlides":["l3-r2-p39"],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-f88908456f7c55a6e91f93a2","nextBlock":null,"prerequisites":[{"id":"b-f88908456f7c55a6e91f93a2","url":"#b-f88908456f7c55a6e91f93a2","role":"section-title"},{"id":"b-c26c26751bb6d4781f96ffd3","url":"#b-c26c26751bb6d4781f96ffd3","role":"problem-context"}],"relatedBlocks":[{"id":"b-57f74fcc8a17019f50203bae","url":"#b-57f74fcc8a17019f50203bae","relation":"figure-and-explanation"}]},{"id":"b-dd495b78e581774c587f1841","type":"h2","scope":"set-matching-lab20-costs","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p39.html#b-dd495b78e581774c587f1841","content":"## 选读 LAB20：沿真实输入核对交叉匹配","sha256":"4726920945ac0b498f7801919906ebe166fc4f0bd9699b905e0a52247f1168e2","chapter":{"id":"II","title":"区域、像素与实例"},"unit":{"id":"spatial-training-extensions","title":"选读：输出组织与训练分配还可以怎样改变？"},"page":{"id":"set-matching-lab20-costs","title":"选读 LAB20：沿真实输入核对交叉匹配","number":2},"sectionTitle":"选读 LAB20：沿真实输入核对交叉匹配","currentQuestion":"在已掌握区域、像素和实例后，按问题阅读结构与匹配扩展。","sourceSlides":["l3-r2-p35","l3-r2-p39"],"knowledgeBlock":null,"formulas":[],"previousBlock":null,"nextBlock":"b-47bb2366f2357ac7518b002b","prerequisites":[{"id":"b-47bb2366f2357ac7518b002b","url":"#b-47bb2366f2357ac7518b002b","role":"problem-context"}],"relatedBlocks":[]},{"id":"b-47bb2366f2357ac7518b002b","type":"p","scope":"set-matching-lab20-costs","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p39.html#b-47bb2366f2357ac7518b002b","content":"用已保存的小算例落实上一页的对应问题，检查成本与数组顺序。","sha256":"80aed3648cc101c5871ace40ae8fba4028b3be912a2e368644520e77b31c275c","chapter":{"id":"II","title":"区域、像素与实例"},"unit":{"id":"spatial-training-extensions","title":"选读：输出组织与训练分配还可以怎样改变？"},"page":{"id":"set-matching-lab20-costs","title":"选读 LAB20：沿真实输入核对交叉匹配","number":2},"sectionTitle":"选读 LAB20：沿真实输入核对交叉匹配","currentQuestion":"在已掌握区域、像素和实例后，按问题阅读结构与匹配扩展。","sourceSlides":["l3-r2-p35","l3-r2-p39"],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-dd495b78e581774c587f1841","nextBlock":"b-a9ecbcc64a15a15c7e344e26","prerequisites":[{"id":"b-dd495b78e581774c587f1841","url":"#b-dd495b78e581774c587f1841","role":"section-title"}],"relatedBlocks":[]},{"id":"b-a9ecbcc64a15a15c7e344e26","type":"h3","scope":"set-matching-real-inputs","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p39.html#b-a9ecbcc64a15a15c7e344e26","content":"### LAB20 的两个槽位为何应交叉匹配","sha256":"e22ba271dbf8de903d354b548af45e31549397c87cf88aa2ff55aab69ec47377","chapter":{"id":"II","title":"区域、像素与实例"},"unit":{"id":"spatial-training-extensions","title":"选读：输出组织与训练分配还可以怎样改变？"},"page":{"id":"set-matching-lab20-costs","title":"选读 LAB20：沿真实输入核对交叉匹配","number":2},"sectionTitle":"LAB20 的两个槽位为何应交叉匹配","currentQuestion":"在已掌握区域、像素和实例后，按问题阅读结构与匹配扩展。","sourceSlides":["l3-r2-p35","l3-r2-p39"],"knowledgeBlock":"set-matching-real-inputs","formulas":[],"previousBlock":"b-47bb2366f2357ac7518b002b","nextBlock":"b-9257aab2323ee2dbc2d775c0","prerequisites":[{"id":"b-dd495b78e581774c587f1841","url":"#b-dd495b78e581774c587f1841","role":"section-title"},{"id":"b-47bb2366f2357ac7518b002b","url":"#b-47bb2366f2357ac7518b002b","role":"problem-context"}],"relatedBlocks":[{"id":"b-9257aab2323ee2dbc2d775c0","url":"#b-9257aab2323ee2dbc2d775c0","relation":"same-reasoning"},{"id":"b-656da9b2b7c7130e194f0a3a","url":"#b-656da9b2b7c7130e194f0a3a","relation":"same-reasoning"},{"id":"b-02509d3458e83596789f8ade","url":"#b-02509d3458e83596789f8ade","relation":"same-reasoning"},{"id":"b-9e42d2818afe818cac5ceb58","url":"#b-9e42d2818afe818cac5ceb58","relation":"same-reasoning"}]},{"id":"b-9257aab2323ee2dbc2d775c0","type":"p","scope":"set-matching-real-inputs","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p39.html#b-9257aab2323ee2dbc2d775c0","content":"当前 LAB20 的第一项预测框在右侧，类别概率更支持第二类；第二项预测在左侧，更支持第一类。真值数组却按左、右保存。如果不考虑内容，直接让槽位 0 学真值 0，就人为制造了位置和类别冲突。","sha256":"8c4a4fcaec0e6c0eb856799d47828d49dcd70cbeddd2d724b59395207d95348e","chapter":{"id":"II","title":"区域、像素与实例"},"unit":{"id":"spatial-training-extensions","title":"选读：输出组织与训练分配还可以怎样改变？"},"page":{"id":"set-matching-lab20-costs","title":"选读 LAB20：沿真实输入核对交叉匹配","number":2},"sectionTitle":"LAB20 的两个槽位为何应交叉匹配","currentQuestion":"在已掌握区域、像素和实例后，按问题阅读结构与匹配扩展。","sourceSlides":["l3-r2-p35","l3-r2-p39"],"knowledgeBlock":"set-matching-real-inputs","formulas":[],"previousBlock":"b-a9ecbcc64a15a15c7e344e26","nextBlock":"b-656da9b2b7c7130e194f0a3a","prerequisites":[{"id":"b-a9ecbcc64a15a15c7e344e26","url":"#b-a9ecbcc64a15a15c7e344e26","role":"section-title"},{"id":"b-47bb2366f2357ac7518b002b","url":"#b-47bb2366f2357ac7518b002b","role":"problem-context"}],"relatedBlocks":[{"id":"b-a9ecbcc64a15a15c7e344e26","url":"#b-a9ecbcc64a15a15c7e344e26","relation":"same-reasoning"},{"id":"b-656da9b2b7c7130e194f0a3a","url":"#b-656da9b2b7c7130e194f0a3a","relation":"same-reasoning"},{"id":"b-02509d3458e83596789f8ade","url":"#b-02509d3458e83596789f8ade","relation":"same-reasoning"},{"id":"b-9e42d2818afe818cac5ceb58","url":"#b-9e42d2818afe818cac5ceb58","relation":"same-reasoning"}]},{"id":"b-656da9b2b7c7130e194f0a3a","type":"p","scope":"set-matching-real-inputs","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p39.html#b-656da9b2b7c7130e194f0a3a","content":"代码按教学定义 $C_{ij}=-\\log p_i(c_j)+2\\lVert b_i-g_j\\rVert_1+(1-IoU)$ 构造成本，得到约 `[[5.0971,.4509],[.4779,5.6026]]`。对角总代价 10.6997，交叉 .9288，因此最佳分配为槽位 0→真值 1、槽位 1→真值 0。只交换标注或预测数组的排列，最小内容对应仍然不变。","sha256":"666afb23ae32517dfc3373217aa07120f72ac5e338277cf55cef59ec35e29c96","chapter":{"id":"II","title":"区域、像素与实例"},"unit":{"id":"spatial-training-extensions","title":"选读：输出组织与训练分配还可以怎样改变？"},"page":{"id":"set-matching-lab20-costs","title":"选读 LAB20：沿真实输入核对交叉匹配","number":2},"sectionTitle":"LAB20 的两个槽位为何应交叉匹配","currentQuestion":"在已掌握区域、像素和实例后，按问题阅读结构与匹配扩展。","sourceSlides":["l3-r2-p35","l3-r2-p39"],"knowledgeBlock":"set-matching-real-inputs","formulas":[{"latex":"C_{ij}=-\\log p_i(c_j)+2\\lVert b_i-g_j\\rVert_1+(1-IoU)","definitionContext":"当前 LAB20 的第一项预测框在右侧，类别概率更支持第二类；第二项预测在左侧，更支持第一类。真值数组却按左、右保存。如果不考虑内容，直接让槽位 0 学真值 0，就人为制造了位置和类别冲突。\n代码按教学定义 $C_{ij}=-\\log p_i(c_j)+2\\lVert b_i-g_j\\rVert_1+(1-IoU)$ 构造成本，得到约 `[[5.0971,.4509],[.4779,5.6026]]`。对角总代价 10.6997，交叉 .9288，因此最佳分配为槽位 0→真值 1、槽位 1→真值 0。只交换标注或预测数组的排列，最小内容对应仍然不变。\n此处的负对数分类项、IoU 与系数是教学选择；官方 DETR matcher 使用其明确的分类、L1、GIoU 成本，不应声称公式逐项一致。匹配找到最低代价也不意味着检测正确：即使两项预测都很差，算法仍会选一个较便宜分配。检测质量仍要用保留数据与定位/类别标准评价。\n本册另一小算例把 16 像素目标映到不同 stride 的连续跨度，属于尺度读取，与集合匹配不同。导航中把二者标为两项可单独读的短练习，学生无需为找一句机制解释穿过全部实验。"}],"previousBlock":"b-9257aab2323ee2dbc2d775c0","nextBlock":"b-02509d3458e83596789f8ade","prerequisites":[{"id":"b-a9ecbcc64a15a15c7e344e26","url":"#b-a9ecbcc64a15a15c7e344e26","role":"section-title"},{"id":"b-47bb2366f2357ac7518b002b","url":"#b-47bb2366f2357ac7518b002b","role":"problem-context"}],"relatedBlocks":[{"id":"b-a9ecbcc64a15a15c7e344e26","url":"#b-a9ecbcc64a15a15c7e344e26","relation":"same-reasoning"},{"id":"b-9257aab2323ee2dbc2d775c0","url":"#b-9257aab2323ee2dbc2d775c0","relation":"same-reasoning"},{"id":"b-02509d3458e83596789f8ade","url":"#b-02509d3458e83596789f8ade","relation":"same-reasoning"},{"id":"b-9e42d2818afe818cac5ceb58","url":"#b-9e42d2818afe818cac5ceb58","relation":"same-reasoning"}]},{"id":"b-02509d3458e83596789f8ade","type":"p","scope":"set-matching-real-inputs","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p39.html#b-02509d3458e83596789f8ade","content":"此处的负对数分类项、IoU 与系数是教学选择；官方 DETR matcher 使用其明确的分类、L1、GIoU 成本，不应声称公式逐项一致。匹配找到最低代价也不意味着检测正确：即使两项预测都很差，算法仍会选一个较便宜分配。检测质量仍要用保留数据与定位/类别标准评价。","sha256":"00cdf3ea105cff573d3b5e30cd194b7c8bb22cc8b37fda235912d4f9ef325925","chapter":{"id":"II","title":"区域、像素与实例"},"unit":{"id":"spatial-training-extensions","title":"选读：输出组织与训练分配还可以怎样改变？"},"page":{"id":"set-matching-lab20-costs","title":"选读 LAB20：沿真实输入核对交叉匹配","number":2},"sectionTitle":"LAB20 的两个槽位为何应交叉匹配","currentQuestion":"在已掌握区域、像素和实例后，按问题阅读结构与匹配扩展。","sourceSlides":["l3-r2-p35","l3-r2-p39"],"knowledgeBlock":"set-matching-real-inputs","formulas":[],"previousBlock":"b-656da9b2b7c7130e194f0a3a","nextBlock":"b-9e42d2818afe818cac5ceb58","prerequisites":[{"id":"b-a9ecbcc64a15a15c7e344e26","url":"#b-a9ecbcc64a15a15c7e344e26","role":"section-title"},{"id":"b-47bb2366f2357ac7518b002b","url":"#b-47bb2366f2357ac7518b002b","role":"problem-context"}],"relatedBlocks":[{"id":"b-a9ecbcc64a15a15c7e344e26","url":"#b-a9ecbcc64a15a15c7e344e26","relation":"same-reasoning"},{"id":"b-9257aab2323ee2dbc2d775c0","url":"#b-9257aab2323ee2dbc2d775c0","relation":"same-reasoning"},{"id":"b-656da9b2b7c7130e194f0a3a","url":"#b-656da9b2b7c7130e194f0a3a","relation":"same-reasoning"},{"id":"b-9e42d2818afe818cac5ceb58","url":"#b-9e42d2818afe818cac5ceb58","relation":"same-reasoning"}]},{"id":"b-9e42d2818afe818cac5ceb58","type":"p","scope":"set-matching-real-inputs","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p39.html#b-9e42d2818afe818cac5ceb58","content":"本册另一小算例把 16 像素目标映到不同 stride 的连续跨度，属于尺度读取，与集合匹配不同。导航中把二者标为两项可单独读的短练习，学生无需为找一句机制解释穿过全部实验。","sha256":"2c901bf98cd806f477e314f0897e9351b3172089a6eab29d4f2569837c1837e6","chapter":{"id":"II","title":"区域、像素与实例"},"unit":{"id":"spatial-training-extensions","title":"选读：输出组织与训练分配还可以怎样改变？"},"page":{"id":"set-matching-lab20-costs","title":"选读 LAB20：沿真实输入核对交叉匹配","number":2},"sectionTitle":"LAB20 的两个槽位为何应交叉匹配","currentQuestion":"在已掌握区域、像素和实例后，按问题阅读结构与匹配扩展。","sourceSlides":["l3-r2-p35","l3-r2-p39"],"knowledgeBlock":"set-matching-real-inputs","formulas":[],"previousBlock":"b-02509d3458e83596789f8ade","nextBlock":"b-85268d7825036025b7af04df","prerequisites":[{"id":"b-a9ecbcc64a15a15c7e344e26","url":"#b-a9ecbcc64a15a15c7e344e26","role":"section-title"},{"id":"b-47bb2366f2357ac7518b002b","url":"#b-47bb2366f2357ac7518b002b","role":"problem-context"}],"relatedBlocks":[{"id":"b-a9ecbcc64a15a15c7e344e26","url":"#b-a9ecbcc64a15a15c7e344e26","relation":"same-reasoning"},{"id":"b-9257aab2323ee2dbc2d775c0","url":"#b-9257aab2323ee2dbc2d775c0","relation":"same-reasoning"},{"id":"b-656da9b2b7c7130e194f0a3a","url":"#b-656da9b2b7c7130e194f0a3a","relation":"same-reasoning"},{"id":"b-02509d3458e83596789f8ade","url":"#b-02509d3458e83596789f8ade","relation":"same-reasoning"}]},{"id":"b-85268d7825036025b7af04df","type":"h2","scope":"practice","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p39.html#b-85268d7825036025b7af04df","content":"## 沿这个问题继续","sha256":"cf44fcc62782fb4f6bdc42c16c5187d57556a03cb7ce6065e7bd9c36d8e7d733","chapter":{"id":"II","title":"区域、像素与实例"},"unit":{"id":"spatial-training-extensions","title":"选读：输出组织与训练分配还可以怎样改变？"},"page":{"id":"set-matching-lab20-costs","title":"选读 LAB20：沿真实输入核对交叉匹配","number":2},"sectionTitle":"选读 LAB20：沿真实输入核对交叉匹配","currentQuestion":"在已掌握区域、像素和实例后，按问题阅读结构与匹配扩展。","sourceSlides":["l3-r2-p35","l3-r2-p39"],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-9e42d2818afe818cac5ceb58","nextBlock":"b-df1402a24d369554ba2a5099","prerequisites":[{"id":"b-a9ecbcc64a15a15c7e344e26","url":"#b-a9ecbcc64a15a15c7e344e26","role":"section-title"},{"id":"b-47bb2366f2357ac7518b002b","url":"#b-47bb2366f2357ac7518b002b","role":"problem-context"}],"relatedBlocks":[]},{"id":"b-df1402a24d369554ba2a5099","type":"p","scope":"practice","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p39.html#b-df1402a24d369554ba2a5099","content":"先读本页解释与图，再到 LAB03 改一个坐标或对象。","sha256":"b8b69250624fd8c15e6f5db8690409848aaec027fc3b89222c6665c3044c9586","chapter":{"id":"II","title":"区域、像素与实例"},"unit":{"id":"spatial-training-extensions","title":"选读：输出组织与训练分配还可以怎样改变？"},"page":{"id":"set-matching-lab20-costs","title":"选读 LAB20：沿真实输入核对交叉匹配","number":2},"sectionTitle":"选读 LAB20：沿真实输入核对交叉匹配","currentQuestion":"在已掌握区域、像素和实例后，按问题阅读结构与匹配扩展。","sourceSlides":["l3-r2-p35","l3-r2-p39"],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-85268d7825036025b7af04df","nextBlock":"b-32f68d78adc7d06d66bb29d5","prerequisites":[{"id":"b-85268d7825036025b7af04df","url":"#b-85268d7825036025b7af04df","role":"section-title"},{"id":"b-47bb2366f2357ac7518b002b","url":"#b-47bb2366f2357ac7518b002b","role":"problem-context"}],"relatedBlocks":[]},{"id":"b-32f68d78adc7d06d66bb29d5","type":"h4","scope":"practice","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p39.html#b-32f68d78adc7d06d66bb29d5","content":"#### 按问题回查的讲解与短练习（2）","sha256":"482a56d6c16f1fcc67a7310871c9513a1202b8628f3d5497506d151a2b4d5e51","chapter":{"id":"II","title":"区域、像素与实例"},"unit":{"id":"spatial-training-extensions","title":"选读：输出组织与训练分配还可以怎样改变？"},"page":{"id":"set-matching-lab20-costs","title":"选读 LAB20：沿真实输入核对交叉匹配","number":2},"sectionTitle":"选读 LAB20：沿真实输入核对交叉匹配","currentQuestion":"在已掌握区域、像素和实例后，按问题阅读结构与匹配扩展。","sourceSlides":["l3-r2-p35","l3-r2-p39"],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-df1402a24d369554ba2a5099","nextBlock":null,"prerequisites":[{"id":"b-85268d7825036025b7af04df","url":"#b-85268d7825036025b7af04df","role":"section-title"},{"id":"b-47bb2366f2357ac7518b002b","url":"#b-47bb2366f2357ac7518b002b","role":"problem-context"}],"relatedBlocks":[]}]}