{"version":2,"relationshipURLBase":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p63.html","relationshipResolution":"Resolve relative relationship URLs against relationshipURLBase; complete records and absolute URLs are in blocks.","title":"GLIP：不只说相关，还要指出区域","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p63.html","markdown":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p63.md","pages":[{"id":"grounding-target-reference","title":"沿坐标选出目标，而不是参照物","unit":"一句话怎样落实为正确对象的区域？","unitId":"grounding-object-contract","chapter":"III","number":1},{"id":"grounding-lab15-interventions","title":"从输入干预追到实际返回框","unit":"一句话怎样落实为正确对象的区域？","unitId":"grounding-object-contract","chapter":"III","number":2},{"id":"glip-language-localization","title":"从透明规则走向对象级学习方案","unit":"一句话怎样落实为正确对象的区域？","unitId":"grounding-object-contract","chapter":"III","number":3}],"aliases":[{"oldId":"explanation","targetId":"grounding-target-reference","operation":"split-explanation-range"},{"oldId":"b-6a2c9f1e2fecffd286c7f773","targetId":"b-6a2c9f1e2fecffd286c7f773","page":"grounding-target-reference","operation":"move","beforeSHA256":"d71cf15e707c8a3674b4c90a599efee783cb84da5b8c8236a1d01c3add7b4958","afterSHA256":"d71cf15e707c8a3674b4c90a599efee783cb84da5b8c8236a1d01c3add7b4958","reason":"Same source body in authored semantic page"},{"oldId":"b-4a71a6c50e2c28eb72319e46","targetId":"b-4a71a6c50e2c28eb72319e46","page":"grounding-target-reference","operation":"move","beforeSHA256":"108379ff132b0353c733334f16ac13aa69384e7e3a164e62dd9b6e10040c347d","afterSHA256":"108379ff132b0353c733334f16ac13aa69384e7e3a164e62dd9b6e10040c347d","reason":"Same source body in authored semantic page"},{"oldId":"b-c9497c1c9165483e4bdb23bd","targetId":"b-c9497c1c9165483e4bdb23bd","page":"grounding-target-reference","operation":"move","beforeSHA256":"7d8bc1408f3f30ddf3278f32dbb54c475cf47f3b9b5ab00d4d7f1ef8f718627d","afterSHA256":"7d8bc1408f3f30ddf3278f32dbb54c475cf47f3b9b5ab00d4d7f1ef8f718627d","reason":"Same source body in authored semantic page"},{"oldId":"b-ee065de60c5e5e93124fbcf9","targetId":"b-ee065de60c5e5e93124fbcf9","page":"grounding-target-reference","operation":"rewrite","beforeSHA256":"526397dd17c987aba79c15dfb28c91182a616e117a6a124cf995cfaa5229fe99","afterSHA256":"618182430fc6c88b7c5aaa9ce1051810811379d808448bdd4f87e1384af7d520","reason":"Same source identity with figure semantics or reading context enriched"},{"oldId":"b-921d1aedcef0d8ff26ecab90","targetId":"b-921d1aedcef0d8ff26ecab90","page":"grounding-target-reference","operation":"move","beforeSHA256":"786ce10a6b32a492259a3dccd0a54e61cd41d28a34382fb7b5386c5f03b7ad69","afterSHA256":"786ce10a6b32a492259a3dccd0a54e61cd41d28a34382fb7b5386c5f03b7ad69","reason":"Same source body in authored semantic page"},{"oldId":"b-18f3a24b83bc2247925b24cf","targetId":"b-18f3a24b83bc2247925b24cf","page":"grounding-target-reference","operation":"move","beforeSHA256":"bb6cd502133938027abc411012d2a99effb7d476854b0ef7e94c4085f29ce046","afterSHA256":"bb6cd502133938027abc411012d2a99effb7d476854b0ef7e94c4085f29ce046","reason":"Same source body in authored semantic page"},{"oldId":"b-61edd5270bb03d38a412f436","targetId":"b-61edd5270bb03d38a412f436","page":"grounding-target-reference","operation":"move","beforeSHA256":"f7b7211d8e394c04e6a50891f2d186f5b58e89a45d28fc7dd9506700a24092ae","afterSHA256":"f7b7211d8e394c04e6a50891f2d186f5b58e89a45d28fc7dd9506700a24092ae","reason":"Same source body in authored semantic page"},{"oldId":"b-8aa65492c7aa4aa160d80f7a","targetId":"b-8aa65492c7aa4aa160d80f7a","page":"grounding-target-reference","operation":"move","beforeSHA256":"ecc2d4cc35754ab1239dd38d2bcab0d3de0a2b3e1a2f7a7a6055bad2349326fd","afterSHA256":"ecc2d4cc35754ab1239dd38d2bcab0d3de0a2b3e1a2f7a7a6055bad2349326fd","reason":"Same source body in authored semantic page"},{"oldId":"b-aa29cdf5e421c266f91a739e","targetId":"b-aa29cdf5e421c266f91a739e","page":"grounding-target-reference","operation":"move","beforeSHA256":"fa64c7e04210b5383ba8acd1b35e411888d7e97d955ce692b290f61d15fc651f","afterSHA256":"fa64c7e04210b5383ba8acd1b35e411888d7e97d955ce692b290f61d15fc651f","reason":"Same source body in authored semantic page"},{"oldId":"b-506963aed02c282eadab012e","targetId":"b-506963aed02c282eadab012e","page":"grounding-target-reference","operation":"move","beforeSHA256":"a7e6186a3879cf00b7b4e6691c1a5fe72292766a0c2120f0667b376036a914b6","afterSHA256":"a7e6186a3879cf00b7b4e6691c1a5fe72292766a0c2120f0667b376036a914b6","reason":"Same source body in authored semantic page"},{"oldId":"b-3d7c4519e6825a4bc457c63e","targetId":"b-3d7c4519e6825a4bc457c63e","page":"grounding-target-reference","operation":"move","beforeSHA256":"374d2bfb115c9901efd89d4182bd0a05ca1076977308b072a18ea0d33bcb6d4a","afterSHA256":"374d2bfb115c9901efd89d4182bd0a05ca1076977308b072a18ea0d33bcb6d4a","reason":"Same source body in authored semantic page"},{"oldId":"b-ceb7e39d586b201cb5f6e9fd","targetId":"b-ceb7e39d586b201cb5f6e9fd","page":"grounding-target-reference","operation":"move","beforeSHA256":"6846eb169f6eea9b81d65738eb8d5da77c3c3ff35cd5cb3cd8ac9606c3e53ee7","afterSHA256":"6846eb169f6eea9b81d65738eb8d5da77c3c3ff35cd5cb3cd8ac9606c3e53ee7","reason":"Same source body in authored semantic page"},{"oldId":"b-97ebb336efaf78bd0c446c58","targetId":"b-97ebb336efaf78bd0c446c58","page":"grounding-target-reference","operation":"move","beforeSHA256":"d28ae5ad8854a8867c06ae0995df510cbc1a835b6039f9172788343269235477","afterSHA256":"d28ae5ad8854a8867c06ae0995df510cbc1a835b6039f9172788343269235477","reason":"Same source body in authored semantic page"},{"oldId":"b-fbd1d981831d1e9d3ea87d2b","targetId":"b-fbd1d981831d1e9d3ea87d2b","page":"grounding-target-reference","operation":"move","beforeSHA256":"3a685b8b95c708b86b245f39cdbde5d8c6936c29273c99261590abb9422d04f1","afterSHA256":"3a685b8b95c708b86b245f39cdbde5d8c6936c29273c99261590abb9422d04f1","reason":"Same source body in authored semantic page"},{"oldId":"b-17fcb7df9471d529d60daf62","targetId":"b-17fcb7df9471d529d60daf62","page":"grounding-target-reference","operation":"move","beforeSHA256":"875ade1b239910a6741fd477b16f4f1e09b84b25bd4c731ef99b391ab597d300","afterSHA256":"875ade1b239910a6741fd477b16f4f1e09b84b25bd4c731ef99b391ab597d300","reason":"Same source body in authored semantic page"},{"oldId":"b-a99c507166df44bf48000e50","targetId":"b-a99c507166df44bf48000e50","page":"grounding-target-reference","operation":"move","beforeSHA256":"7768902d58ef57b20c88ac26d2fec7ae259c1fcfaee5c150e2530364af2e5027","afterSHA256":"7768902d58ef57b20c88ac26d2fec7ae259c1fcfaee5c150e2530364af2e5027","reason":"Same source body in authored semantic page"},{"oldId":"b-8c732a78745b7f13f26979da","targetId":"b-8c732a78745b7f13f26979da","page":"grounding-target-reference","operation":"move","beforeSHA256":"9417752cc706088cad5cf42f725b2bbbd10f191fcbaa05ffd8202581eeca17ee","afterSHA256":"9417752cc706088cad5cf42f725b2bbbd10f191fcbaa05ffd8202581eeca17ee","reason":"Same source body in authored semantic page"},{"oldId":"b-d67e4e9281001cc6c05e1314","targetId":"b-d67e4e9281001cc6c05e1314","page":"grounding-target-reference","operation":"move","beforeSHA256":"2508a2657e8a66706205f014cdb1116ab56d8dd9eb786410309cd124d243eb03","afterSHA256":"2508a2657e8a66706205f014cdb1116ab56d8dd9eb786410309cd124d243eb03","reason":"Same source body in authored semantic page"},{"oldId":"b-332609e5fd09e2c21c9bd95f","targetId":"b-332609e5fd09e2c21c9bd95f","page":"grounding-target-reference","operation":"move","beforeSHA256":"bca26edd21c30a247e6d973495e11bc2f557af02d1b0ff794c9123e1dc0c4123","afterSHA256":"bca26edd21c30a247e6d973495e11bc2f557af02d1b0ff794c9123e1dc0c4123","reason":"Same source body in authored semantic page"},{"oldId":"b-830446b5c23c16f0cb906b5e","targetId":"b-830446b5c23c16f0cb906b5e","page":"grounding-lab15-interventions","operation":"move","beforeSHA256":"b919cfcf83f000f7f330246ddb469148384c48ca59f8b48b976b98d36652228e","afterSHA256":"b919cfcf83f000f7f330246ddb469148384c48ca59f8b48b976b98d36652228e","reason":"Same source body in authored semantic page"},{"oldId":"b-25b6cd118879b490a311e463","targetId":"b-25b6cd118879b490a311e463","page":"grounding-lab15-interventions","operation":"move","beforeSHA256":"ec817964628fe547da360e42935f3ab5e1d4ab0a413223243d9b9bdd08b46be5","afterSHA256":"ec817964628fe547da360e42935f3ab5e1d4ab0a413223243d9b9bdd08b46be5","reason":"Same source body in authored semantic page"},{"oldId":"b-c54caf119e7797bf6e8d13cb","targetId":"b-c54caf119e7797bf6e8d13cb","page":"grounding-lab15-interventions","operation":"rewrite","beforeSHA256":"73bea769aec3dcf143e6309aa7486e9552aae00c14ac4750b7956401686e39f3","afterSHA256":"0de466cf5672ac86c9dd4214b39ae872dd46ee32d4256b3291c661a7a495c361","reason":"Same source identity with figure semantics or reading context enriched"},{"oldId":"b-d8b9555619e7c4e2dc3548f0","targetId":"b-d8b9555619e7c4e2dc3548f0","page":"grounding-lab15-interventions","operation":"move","beforeSHA256":"55fb5f3349686e772ea73d62b88bc243f3fdb3efd45df8c2454262f799c64329","afterSHA256":"55fb5f3349686e772ea73d62b88bc243f3fdb3efd45df8c2454262f799c64329","reason":"Same source body in authored semantic page"},{"oldId":"b-6ec0eaa5d6ba03a6ac8bb3f0","targetId":"b-6ec0eaa5d6ba03a6ac8bb3f0","page":"grounding-lab15-interventions","operation":"move","beforeSHA256":"0b5941126b2133f8855a3f934fc08044d64ac2370383c8794b2060a33380f994","afterSHA256":"70aab124a4de4ccb1629c768a54fc062777fd23a1d521943a55fc0d99ea993f7","reason":"Same source body in authored semantic page"},{"oldId":"b-830b8f7e40668ab1598dd62e","targetId":"b-830b8f7e40668ab1598dd62e","page":"grounding-lab15-interventions","operation":"move","beforeSHA256":"2c856e57e1c37c3d70093e2c5de7c9890f3a1bfe364eef8129588edb5dd7a622","afterSHA256":"2c856e57e1c37c3d70093e2c5de7c9890f3a1bfe364eef8129588edb5dd7a622","reason":"Same source body in authored semantic page"},{"oldId":"b-7fbed2e85325174c9948694f","targetId":"b-7fbed2e85325174c9948694f","page":"glip-language-localization","operation":"move","beforeSHA256":"f7f5551cc7a0af3daebec2868febbf8d91ddd869713f104db0b7c9920e28a535","afterSHA256":"f7f5551cc7a0af3daebec2868febbf8d91ddd869713f104db0b7c9920e28a535","reason":"Same source body in authored semantic page"},{"oldId":"b-314b2f66ec6db241fbc09f45","targetId":"b-314b2f66ec6db241fbc09f45","page":"glip-language-localization","operation":"move","beforeSHA256":"24baff1cf87b2de7ca97ba1336e8c33d4560ab39283b2ef8c7c1c36dba4b50fc","afterSHA256":"24baff1cf87b2de7ca97ba1336e8c33d4560ab39283b2ef8c7c1c36dba4b50fc","reason":"Same source body in authored semantic page"},{"oldId":"b-069046fefcdf06847c400fb2","targetId":"b-069046fefcdf06847c400fb2","page":"glip-language-localization","operation":"move","beforeSHA256":"80dfe0ea15830ea4065fc5c1744a90f46c42c9fa95ba56ae9d2715668a597d60","afterSHA256":"80dfe0ea15830ea4065fc5c1744a90f46c42c9fa95ba56ae9d2715668a597d60","reason":"Same source body in authored semantic page"},{"oldId":"b-3d6cd84caa9d9ba96e8d6987","targetId":"b-3d6cd84caa9d9ba96e8d6987","page":"glip-language-localization","operation":"move","beforeSHA256":"bb4028ef924f2a8d235dbf0f1ba9b2b386877b47814234f5a6e6b5a0f191e153","afterSHA256":"bb4028ef924f2a8d235dbf0f1ba9b2b386877b47814234f5a6e6b5a0f191e153","reason":"Same source body in authored semantic page"},{"oldId":"b-dcef8d2d7d7fa2e97ec5f919","targetId":"b-dcef8d2d7d7fa2e97ec5f919","page":"glip-language-localization","operation":"move","beforeSHA256":"acb849912de164fda2c6b1f36086f33e94a4c1f8fc5b5957a303c674c13a844f","afterSHA256":"acb849912de164fda2c6b1f36086f33e94a4c1f8fc5b5957a303c674c13a844f","reason":"Same source body in authored semantic page"},{"oldId":"b-1d9b0885f2a4a66578c780f7","targetId":"b-1d9b0885f2a4a66578c780f7","page":"glip-language-localization","operation":"move","beforeSHA256":"cf44fcc62782fb4f6bdc42c16c5187d57556a03cb7ce6065e7bd9c36d8e7d733","afterSHA256":"cf44fcc62782fb4f6bdc42c16c5187d57556a03cb7ce6065e7bd9c36d8e7d733","reason":"Same source body in authored semantic page"},{"oldId":"b-7cc00c2dd5b32b41f709450a","targetId":"b-7cc00c2dd5b32b41f709450a","page":"glip-language-localization","operation":"move","beforeSHA256":"ba9e0813debefcdf122ce5a17bf25680c0faa25ecf1a61ee6459f109e5776998","afterSHA256":"ba9e0813debefcdf122ce5a17bf25680c0faa25ecf1a61ee6459f109e5776998","reason":"Same source body in authored semantic page"},{"oldId":"grounding-target-reference-title","targetId":"b-38db5e20dd0fb0aa3c7d86c0","operation":"retain-source-identity"},{"oldId":"grounding-lab15-interventions-title","targetId":"b-f161c449ecedb21ef539390a","operation":"retain-source-identity"},{"oldId":"grounding-lab15-interventions-bridge","targetId":"b-360162cda01cb1cddbf43ec2","operation":"retain-source-identity"},{"oldId":"glip-language-localization-title","targetId":"b-d50a51f982259222c44dc219","operation":"retain-source-identity"},{"oldId":"glip-language-localization-bridge","targetId":"b-1c989e608aeddcda3cce73c3","operation":"retain-source-identity"}],"blocks":[{"id":"b-38db5e20dd0fb0aa3c7d86c0","type":"h2","scope":"grounding-target-reference","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p63.html#b-38db5e20dd0fb0aa3c7d86c0","content":"## 沿坐标选出目标，而不是参照物","sha256":"fc7ac1126caeca61befb7125c2e6d58c1ee02db82c84340997345fdbe53e1497","chapter":{"id":"III","title":"视觉与语言"},"unit":{"id":"grounding-object-contract","title":"一句话怎样落实为正确对象的区域？"},"page":{"id":"grounding-target-reference","title":"沿坐标选出目标，而不是参照物","number":1},"sectionTitle":"沿坐标选出目标，而不是参照物","currentQuestion":"目标、参照物与关系怎样共同约束定位输出？","sourceSlides":["l3-r2-p63"],"knowledgeBlock":null,"formulas":[],"previousBlock":null,"nextBlock":"b-6a2c9f1e2fecffd286c7f773","prerequisites":[{"id":"b-c9497c1c9165483e4bdb23bd","url":"#b-c9497c1c9165483e4bdb23bd","role":"problem-context"}],"relatedBlocks":[{"id":"b-ee065de60c5e5e93124fbcf9","url":"#b-ee065de60c5e5e93124fbcf9","relation":"figure-and-explanation"}]},{"id":"b-6a2c9f1e2fecffd286c7f773","type":"h2","scope":"explanation","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p63.html#b-6a2c9f1e2fecffd286c7f773","content":"## 解释、算例与证据","sha256":"d71cf15e707c8a3674b4c90a599efee783cb84da5b8c8236a1d01c3add7b4958","chapter":{"id":"III","title":"视觉与语言"},"unit":{"id":"grounding-object-contract","title":"一句话怎样落实为正确对象的区域？"},"page":{"id":"grounding-target-reference","title":"沿坐标选出目标，而不是参照物","number":1},"sectionTitle":"沿坐标选出目标，而不是参照物","currentQuestion":"目标、参照物与关系怎样共同约束定位输出？","sourceSlides":["l3-r2-p63"],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-38db5e20dd0fb0aa3c7d86c0","nextBlock":"b-4a71a6c50e2c28eb72319e46","prerequisites":[{"id":"b-38db5e20dd0fb0aa3c7d86c0","url":"#b-38db5e20dd0fb0aa3c7d86c0","role":"section-title"},{"id":"b-c9497c1c9165483e4bdb23bd","url":"#b-c9497c1c9165483e4bdb23bd","role":"problem-context"}],"relatedBlocks":[{"id":"b-ee065de60c5e5e93124fbcf9","url":"#b-ee065de60c5e5e93124fbcf9","relation":"figure-and-explanation"}]},{"id":"b-4a71a6c50e2c28eb72319e46","type":"h1","scope":"document","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p63.html#b-4a71a6c50e2c28eb72319e46","content":"# P63 · GLIP：不只说相关，还要指出区域","sha256":"108379ff132b0353c733334f16ac13aa69384e7e3a164e62dd9b6e10040c347d","chapter":{"id":"III","title":"视觉与语言"},"unit":{"id":"grounding-object-contract","title":"一句话怎样落实为正确对象的区域？"},"page":{"id":"grounding-target-reference","title":"沿坐标选出目标，而不是参照物","number":1},"sectionTitle":"沿坐标选出目标，而不是参照物","currentQuestion":"目标、参照物与关系怎样共同约束定位输出？","sourceSlides":["l3-r2-p63"],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-6a2c9f1e2fecffd286c7f773","nextBlock":"b-c9497c1c9165483e4bdb23bd","prerequisites":[{"id":"b-6a2c9f1e2fecffd286c7f773","url":"#b-6a2c9f1e2fecffd286c7f773","role":"section-title"},{"id":"b-c9497c1c9165483e4bdb23bd","url":"#b-c9497c1c9165483e4bdb23bd","role":"problem-context"}],"relatedBlocks":[{"id":"b-ee065de60c5e5e93124fbcf9","url":"#b-ee065de60c5e5e93124fbcf9","relation":"figure-and-explanation"}]},{"id":"b-c9497c1c9165483e4bdb23bd","type":"p","scope":"document","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p63.html#b-c9497c1c9165483e4bdb23bd","content":"语言条件落实为空间输出时，需要明确目标、参照物和关系。GLIP 提供对象级学习和迁移的研究路线；LAB15 提供可逐像素追踪的规则实验。两者的层级不同，但都帮助判断系统到底把框放在谁身上，再与后续掩码处理连接。","sha256":"7d8bc1408f3f30ddf3278f32dbb54c475cf47f3b9b5ab00d4d7f1ef8f718627d","chapter":{"id":"III","title":"视觉与语言"},"unit":{"id":"grounding-object-contract","title":"一句话怎样落实为正确对象的区域？"},"page":{"id":"grounding-target-reference","title":"沿坐标选出目标，而不是参照物","number":1},"sectionTitle":"沿坐标选出目标，而不是参照物","currentQuestion":"目标、参照物与关系怎样共同约束定位输出？","sourceSlides":["l3-r2-p63"],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-4a71a6c50e2c28eb72319e46","nextBlock":"b-ee065de60c5e5e93124fbcf9","prerequisites":[{"id":"b-4a71a6c50e2c28eb72319e46","url":"#b-4a71a6c50e2c28eb72319e46","role":"section-title"}],"relatedBlocks":[{"id":"b-ee065de60c5e5e93124fbcf9","url":"#b-ee065de60c5e5e93124fbcf9","relation":"figure-and-explanation"}]},{"id":"b-ee065de60c5e5e93124fbcf9","type":"figure","scope":"document","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p63.html#b-ee065de60c5e5e93124fbcf9","content":"图内标签与数值：指代：拿球的蓝衣人；“蓝衣人”；“拿球的”；图文交互 + 区域预测；短语对应与定位监督；训练目标落在局部区域；输出：与短语对应的框；GLIP 把词语与空间定位放进同一学习任务；教学框展示任务接口，不是 GLIP 运行结果；区域交互和定位学习不能由整图分数直接代替。\n\n蓝衣人在右侧持球场景，红衣人、蓝衣人与球都有候选框；短语和定位监督指向蓝衣人区域。\n\n- 指代：拿球的蓝衣人\n- “蓝衣人”\n- “拿球的”\n- 图文交互 + 区域预测\n- 短语对应与定位监督\n- 训练目标落在局部区域\n- 输出：与短语对应的框\n- GLIP 把词语与空间定位放进同一学习任务\n- 教学框展示任务接口，不是 GLIP 运行结果；区域交互和定位学习不能由整图分数直接代替。蓝衣人在右侧持球场景，红衣人、蓝衣人与球都有候选框；短语和定位监督指向蓝衣人区域。 [查看对应图解状态](/#/l3-r2-p63/0)[查看完整图解](/course/figures/l3-r2-p63-0.html)","sha256":"618182430fc6c88b7c5aaa9ce1051810811379d808448bdd4f87e1384af7d520","chapter":{"id":"III","title":"视觉与语言"},"unit":{"id":"grounding-object-contract","title":"一句话怎样落实为正确对象的区域？"},"page":{"id":"grounding-target-reference","title":"沿坐标选出目标，而不是参照物","number":1},"sectionTitle":"沿坐标选出目标，而不是参照物","currentQuestion":"目标、参照物与关系怎样共同约束定位输出？","sourceSlides":["l3-r2-p63"],"knowledgeBlock":null,"formulas":[],"figure":{"caption":"蓝衣人在右侧持球场景，红衣人、蓝衣人与球都有候选框；短语和定位监督指向蓝衣人区域。 查看对应图解状态","labels":["指代：拿球的蓝衣人","“蓝衣人”","“拿球的”","图文交互 + 区域预测","短语对应与定位监督","训练目标落在局部区域","输出：与短语对应的框","GLIP 把词语与空间定位放进同一学习任务","教学框展示任务接口，不是 GLIP 运行结果；区域交互和定位学习不能由整图分数直接代替。"],"description":"蓝衣人在右侧持球场景，红衣人、蓝衣人与球都有候选框；短语和定位监督指向蓝衣人区域。指代：拿球的蓝衣人“蓝衣人”“拿球的”图文交互 + 区域预测短语对应与定位监督训练目标落在局部区域输出：与短语对应的框GLIP 把词语与空间定位放进同一学习任务教学框展示任务接口，不是 GLIP 运行结果；区域交互和定位学习不能由整图分数直接代替。","evidenceIdentity":"蓝衣人在右侧持球场景，红衣人、蓝衣人与球都有候选框；短语和定位监督指向蓝衣人区域。 查看对应图解状态","images":[{"src":"/assets/display/ab33341ce4021f383054cfc5.webp","alt":""}]},"previousBlock":"b-c9497c1c9165483e4bdb23bd","nextBlock":"b-921d1aedcef0d8ff26ecab90","prerequisites":[{"id":"b-4a71a6c50e2c28eb72319e46","url":"#b-4a71a6c50e2c28eb72319e46","role":"section-title"},{"id":"b-c9497c1c9165483e4bdb23bd","url":"#b-c9497c1c9165483e4bdb23bd","role":"problem-context"}],"relatedBlocks":[]},{"id":"b-921d1aedcef0d8ff26ecab90","type":"h2","scope":"narrative","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p63.html#b-921d1aedcef0d8ff26ecab90","content":"## 课堂讲述","sha256":"786ce10a6b32a492259a3dccd0a54e61cd41d28a34382fb7b5386c5f03b7ad69","chapter":{"id":"III","title":"视觉与语言"},"unit":{"id":"grounding-object-contract","title":"一句话怎样落实为正确对象的区域？"},"page":{"id":"grounding-target-reference","title":"沿坐标选出目标，而不是参照物","number":1},"sectionTitle":"沿坐标选出目标，而不是参照物","currentQuestion":"目标、参照物与关系怎样共同约束定位输出？","sourceSlides":["l3-r2-p63"],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-ee065de60c5e5e93124fbcf9","nextBlock":"b-18f3a24b83bc2247925b24cf","prerequisites":[{"id":"b-4a71a6c50e2c28eb72319e46","url":"#b-4a71a6c50e2c28eb72319e46","role":"section-title"},{"id":"b-c9497c1c9165483e4bdb23bd","url":"#b-c9497c1c9165483e4bdb23bd","role":"problem-context"}],"relatedBlocks":[{"id":"b-ee065de60c5e5e93124fbcf9","url":"#b-ee065de60c5e5e93124fbcf9","relation":"figure-and-explanation"}]},{"id":"b-18f3a24b83bc2247925b24cf","type":"p","scope":"narrative","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p63.html#b-18f3a24b83bc2247925b24cf","content":"GLIP 把语言与区域感知更直接地结合。文字中的概念和短语，不只与整幅图比较，还要与区域建立对应，并通过定位相关的监督学习输出位置。\n因此，它不是简单把 CLIP 的整图分数画成一个框。局部关系、训练数据以及图文交互都进入了方法。我们在第二部分学到的区域组织，在这里遇到了更开放的语言条件。\n这条技术线说明，语言可以改变视觉任务的组织方式，但不会自动消除空间任务。要指出“拿球的蓝衣人”，既要理解文字提出了什么，也要找到画面中的对应区域。接下来再看另一个方向：如果输出不是框，而是一段根据图片作出的回答呢？","sha256":"bb6cd502133938027abc411012d2a99effb7d476854b0ef7e94c4085f29ce046","chapter":{"id":"III","title":"视觉与语言"},"unit":{"id":"grounding-object-contract","title":"一句话怎样落实为正确对象的区域？"},"page":{"id":"grounding-target-reference","title":"沿坐标选出目标，而不是参照物","number":1},"sectionTitle":"沿坐标选出目标，而不是参照物","currentQuestion":"目标、参照物与关系怎样共同约束定位输出？","sourceSlides":["l3-r2-p63"],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-921d1aedcef0d8ff26ecab90","nextBlock":"b-61edd5270bb03d38a412f436","prerequisites":[{"id":"b-921d1aedcef0d8ff26ecab90","url":"#b-921d1aedcef0d8ff26ecab90","role":"section-title"},{"id":"b-c9497c1c9165483e4bdb23bd","url":"#b-c9497c1c9165483e4bdb23bd","role":"problem-context"}],"relatedBlocks":[{"id":"b-ee065de60c5e5e93124fbcf9","url":"#b-ee065de60c5e5e93124fbcf9","relation":"figure-and-explanation"}]},{"id":"b-61edd5270bb03d38a412f436","type":"h3","scope":"grounding-target-reference-relation","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p63.html#b-61edd5270bb03d38a412f436","content":"### “蓝杯左边的杯子”：目标、参照物、关系三者分开","sha256":"f7b7211d8e394c04e6a50891f2d186f5b58e89a45d28fc7dd9506700a24092ae","chapter":{"id":"III","title":"视觉与语言"},"unit":{"id":"grounding-object-contract","title":"一句话怎样落实为正确对象的区域？"},"page":{"id":"grounding-target-reference","title":"沿坐标选出目标，而不是参照物","number":1},"sectionTitle":"“蓝杯左边的杯子”：目标、参照物、关系三者分开","currentQuestion":"目标、参照物与关系怎样共同约束定位输出？","sourceSlides":["l3-r2-p63"],"knowledgeBlock":"grounding-target-reference-relation","formulas":[],"previousBlock":"b-18f3a24b83bc2247925b24cf","nextBlock":"b-8aa65492c7aa4aa160d80f7a","prerequisites":[{"id":"b-921d1aedcef0d8ff26ecab90","url":"#b-921d1aedcef0d8ff26ecab90","role":"section-title"},{"id":"b-c9497c1c9165483e4bdb23bd","url":"#b-c9497c1c9165483e4bdb23bd","role":"problem-context"}],"relatedBlocks":[{"id":"b-ee065de60c5e5e93124fbcf9","url":"#b-ee065de60c5e5e93124fbcf9","relation":"figure-and-explanation"},{"id":"b-8aa65492c7aa4aa160d80f7a","url":"#b-8aa65492c7aa4aa160d80f7a","relation":"same-reasoning"},{"id":"b-aa29cdf5e421c266f91a739e","url":"#b-aa29cdf5e421c266f91a739e","relation":"same-reasoning"},{"id":"b-506963aed02c282eadab012e","url":"#b-506963aed02c282eadab012e","relation":"same-reasoning"},{"id":"b-3d7c4519e6825a4bc457c63e","url":"#b-3d7c4519e6825a4bc457c63e","relation":"same-reasoning"}]},{"id":"b-8aa65492c7aa4aa160d80f7a","type":"p","scope":"grounding-target-reference-relation","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p63.html#b-8aa65492c7aa4aa160d80f7a","content":"短语 `cup left of blue cup` 中，目标类别是 cup，参照物是 blue cup，关系是 left of。看到 blue 就返回蓝杯，找到了参照物，却没有完成目标选择。","sha256":"ecc2d4cc35754ab1239dd38d2bcab0d3de0a2b3e1a2f7a7a6055bad2349326fd","chapter":{"id":"III","title":"视觉与语言"},"unit":{"id":"grounding-object-contract","title":"一句话怎样落实为正确对象的区域？"},"page":{"id":"grounding-target-reference","title":"沿坐标选出目标，而不是参照物","number":1},"sectionTitle":"“蓝杯左边的杯子”：目标、参照物、关系三者分开","currentQuestion":"目标、参照物与关系怎样共同约束定位输出？","sourceSlides":["l3-r2-p63"],"knowledgeBlock":"grounding-target-reference-relation","formulas":[],"previousBlock":"b-61edd5270bb03d38a412f436","nextBlock":"b-aa29cdf5e421c266f91a739e","prerequisites":[{"id":"b-61edd5270bb03d38a412f436","url":"#b-61edd5270bb03d38a412f436","role":"section-title"},{"id":"b-c9497c1c9165483e4bdb23bd","url":"#b-c9497c1c9165483e4bdb23bd","role":"problem-context"}],"relatedBlocks":[{"id":"b-ee065de60c5e5e93124fbcf9","url":"#b-ee065de60c5e5e93124fbcf9","relation":"figure-and-explanation"},{"id":"b-61edd5270bb03d38a412f436","url":"#b-61edd5270bb03d38a412f436","relation":"same-reasoning"},{"id":"b-aa29cdf5e421c266f91a739e","url":"#b-aa29cdf5e421c266f91a739e","relation":"same-reasoning"},{"id":"b-506963aed02c282eadab012e","url":"#b-506963aed02c282eadab012e","relation":"same-reasoning"},{"id":"b-3d7c4519e6825a4bc457c63e","url":"#b-3d7c4519e6825a4bc457c63e","relation":"same-reasoning"}]},{"id":"b-aa29cdf5e421c266f91a739e","type":"p","scope":"grounding-target-reference-relation","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p63.html#b-aa29cdf5e421c266f91a739e","content":"LAB15 的受控画布中，左红杯框为 $[15,19,50,58]$，中心横坐标 32.5；右蓝杯框为 $[109,19,144,58]$，中心横坐标 126.5；书的框为 $[60,65,90,80]$。本例定义 left-of 为目标中心横坐标小于参照中心，且目标不是参照物本身。","sha256":"fa64c7e04210b5383ba8acd1b35e411888d7e97d955ce692b290f61d15fc651f","chapter":{"id":"III","title":"视觉与语言"},"unit":{"id":"grounding-object-contract","title":"一句话怎样落实为正确对象的区域？"},"page":{"id":"grounding-target-reference","title":"沿坐标选出目标，而不是参照物","number":1},"sectionTitle":"“蓝杯左边的杯子”：目标、参照物、关系三者分开","currentQuestion":"目标、参照物与关系怎样共同约束定位输出？","sourceSlides":["l3-r2-p63"],"knowledgeBlock":"grounding-target-reference-relation","formulas":[{"latex":"[15,19,50,58]","definitionContext":"短语 `cup left of blue cup` 中，目标类别是 cup，参照物是 blue cup，关系是 left of。看到 blue 就返回蓝杯，找到了参照物，却没有完成目标选择。\nLAB15 的受控画布中，左红杯框为 $[15,19,50,58]$，中心横坐标 32.5；右蓝杯框为 $[109,19,144,58]$，中心横坐标 126.5；书的框为 $[60,65,90,80]$。本例定义 left-of 为目标中心横坐标小于参照中心，且目标不是参照物本身。\n先找唯一蓝杯，再从其他杯子中筛选：红杯满足 $32.5<126.5$，应返回红杯框。书也在蓝杯左侧，却不满足 cup 类别约束。若蓝杯缺失、出现多个蓝杯或存在多个满足关系的目标，程序必须有明确的无目标/多义处理，不能随意返回最高颜色分的对象。\n这里说明的是一套显式关系语义。“完全在左侧”可以改用 $x_{max}(a)<x_{min}(r)$，与中心比较在重叠情况下不同。把语言问题落实为区域时，要先解释采用哪种语义，再谈模型是否正确。图上的坐标约定是半开边界框，不是逐像素包含端点的面积公式。"},{"latex":"[109,19,144,58]","definitionContext":"短语 `cup left of blue cup` 中，目标类别是 cup，参照物是 blue cup，关系是 left of。看到 blue 就返回蓝杯，找到了参照物，却没有完成目标选择。\nLAB15 的受控画布中，左红杯框为 $[15,19,50,58]$，中心横坐标 32.5；右蓝杯框为 $[109,19,144,58]$，中心横坐标 126.5；书的框为 $[60,65,90,80]$。本例定义 left-of 为目标中心横坐标小于参照中心，且目标不是参照物本身。\n先找唯一蓝杯，再从其他杯子中筛选：红杯满足 $32.5<126.5$，应返回红杯框。书也在蓝杯左侧，却不满足 cup 类别约束。若蓝杯缺失、出现多个蓝杯或存在多个满足关系的目标，程序必须有明确的无目标/多义处理，不能随意返回最高颜色分的对象。\n这里说明的是一套显式关系语义。“完全在左侧”可以改用 $x_{max}(a)<x_{min}(r)$，与中心比较在重叠情况下不同。把语言问题落实为区域时，要先解释采用哪种语义，再谈模型是否正确。图上的坐标约定是半开边界框，不是逐像素包含端点的面积公式。"},{"latex":"[60,65,90,80]","definitionContext":"短语 `cup left of blue cup` 中，目标类别是 cup，参照物是 blue cup，关系是 left of。看到 blue 就返回蓝杯，找到了参照物，却没有完成目标选择。\nLAB15 的受控画布中，左红杯框为 $[15,19,50,58]$，中心横坐标 32.5；右蓝杯框为 $[109,19,144,58]$，中心横坐标 126.5；书的框为 $[60,65,90,80]$。本例定义 left-of 为目标中心横坐标小于参照中心，且目标不是参照物本身。\n先找唯一蓝杯，再从其他杯子中筛选：红杯满足 $32.5<126.5$，应返回红杯框。书也在蓝杯左侧，却不满足 cup 类别约束。若蓝杯缺失、出现多个蓝杯或存在多个满足关系的目标，程序必须有明确的无目标/多义处理，不能随意返回最高颜色分的对象。\n这里说明的是一套显式关系语义。“完全在左侧”可以改用 $x_{max}(a)<x_{min}(r)$，与中心比较在重叠情况下不同。把语言问题落实为区域时，要先解释采用哪种语义，再谈模型是否正确。图上的坐标约定是半开边界框，不是逐像素包含端点的面积公式。"}],"previousBlock":"b-8aa65492c7aa4aa160d80f7a","nextBlock":"b-506963aed02c282eadab012e","prerequisites":[{"id":"b-61edd5270bb03d38a412f436","url":"#b-61edd5270bb03d38a412f436","role":"section-title"},{"id":"b-c9497c1c9165483e4bdb23bd","url":"#b-c9497c1c9165483e4bdb23bd","role":"problem-context"}],"relatedBlocks":[{"id":"b-ee065de60c5e5e93124fbcf9","url":"#b-ee065de60c5e5e93124fbcf9","relation":"figure-and-explanation"},{"id":"b-61edd5270bb03d38a412f436","url":"#b-61edd5270bb03d38a412f436","relation":"same-reasoning"},{"id":"b-8aa65492c7aa4aa160d80f7a","url":"#b-8aa65492c7aa4aa160d80f7a","relation":"same-reasoning"},{"id":"b-506963aed02c282eadab012e","url":"#b-506963aed02c282eadab012e","relation":"same-reasoning"},{"id":"b-3d7c4519e6825a4bc457c63e","url":"#b-3d7c4519e6825a4bc457c63e","relation":"same-reasoning"}]},{"id":"b-506963aed02c282eadab012e","type":"p","scope":"grounding-target-reference-relation","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p63.html#b-506963aed02c282eadab012e","content":"先找唯一蓝杯，再从其他杯子中筛选：红杯满足 $32.5<126.5$，应返回红杯框。书也在蓝杯左侧，却不满足 cup 类别约束。若蓝杯缺失、出现多个蓝杯或存在多个满足关系的目标，程序必须有明确的无目标/多义处理，不能随意返回最高颜色分的对象。","sha256":"a7e6186a3879cf00b7b4e6691c1a5fe72292766a0c2120f0667b376036a914b6","chapter":{"id":"III","title":"视觉与语言"},"unit":{"id":"grounding-object-contract","title":"一句话怎样落实为正确对象的区域？"},"page":{"id":"grounding-target-reference","title":"沿坐标选出目标，而不是参照物","number":1},"sectionTitle":"“蓝杯左边的杯子”：目标、参照物、关系三者分开","currentQuestion":"目标、参照物与关系怎样共同约束定位输出？","sourceSlides":["l3-r2-p63"],"knowledgeBlock":"grounding-target-reference-relation","formulas":[{"latex":"32.5<126.5","definitionContext":"短语 `cup left of blue cup` 中，目标类别是 cup，参照物是 blue cup，关系是 left of。看到 blue 就返回蓝杯，找到了参照物，却没有完成目标选择。\nLAB15 的受控画布中，左红杯框为 $[15,19,50,58]$，中心横坐标 32.5；右蓝杯框为 $[109,19,144,58]$，中心横坐标 126.5；书的框为 $[60,65,90,80]$。本例定义 left-of 为目标中心横坐标小于参照中心，且目标不是参照物本身。\n先找唯一蓝杯，再从其他杯子中筛选：红杯满足 $32.5<126.5$，应返回红杯框。书也在蓝杯左侧，却不满足 cup 类别约束。若蓝杯缺失、出现多个蓝杯或存在多个满足关系的目标，程序必须有明确的无目标/多义处理，不能随意返回最高颜色分的对象。\n这里说明的是一套显式关系语义。“完全在左侧”可以改用 $x_{max}(a)<x_{min}(r)$，与中心比较在重叠情况下不同。把语言问题落实为区域时，要先解释采用哪种语义，再谈模型是否正确。图上的坐标约定是半开边界框，不是逐像素包含端点的面积公式。"}],"previousBlock":"b-aa29cdf5e421c266f91a739e","nextBlock":"b-3d7c4519e6825a4bc457c63e","prerequisites":[{"id":"b-61edd5270bb03d38a412f436","url":"#b-61edd5270bb03d38a412f436","role":"section-title"},{"id":"b-c9497c1c9165483e4bdb23bd","url":"#b-c9497c1c9165483e4bdb23bd","role":"problem-context"}],"relatedBlocks":[{"id":"b-ee065de60c5e5e93124fbcf9","url":"#b-ee065de60c5e5e93124fbcf9","relation":"figure-and-explanation"},{"id":"b-61edd5270bb03d38a412f436","url":"#b-61edd5270bb03d38a412f436","relation":"same-reasoning"},{"id":"b-8aa65492c7aa4aa160d80f7a","url":"#b-8aa65492c7aa4aa160d80f7a","relation":"same-reasoning"},{"id":"b-aa29cdf5e421c266f91a739e","url":"#b-aa29cdf5e421c266f91a739e","relation":"same-reasoning"},{"id":"b-3d7c4519e6825a4bc457c63e","url":"#b-3d7c4519e6825a4bc457c63e","relation":"same-reasoning"}]},{"id":"b-3d7c4519e6825a4bc457c63e","type":"p","scope":"grounding-target-reference-relation","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p63.html#b-3d7c4519e6825a4bc457c63e","content":"这里说明的是一套显式关系语义。“完全在左侧”可以改用 $x_{max}(a)<x_{min}(r)$，与中心比较在重叠情况下不同。把语言问题落实为区域时，要先解释采用哪种语义，再谈模型是否正确。图上的坐标约定是半开边界框，不是逐像素包含端点的面积公式。","sha256":"374d2bfb115c9901efd89d4182bd0a05ca1076977308b072a18ea0d33bcb6d4a","chapter":{"id":"III","title":"视觉与语言"},"unit":{"id":"grounding-object-contract","title":"一句话怎样落实为正确对象的区域？"},"page":{"id":"grounding-target-reference","title":"沿坐标选出目标，而不是参照物","number":1},"sectionTitle":"“蓝杯左边的杯子”：目标、参照物、关系三者分开","currentQuestion":"目标、参照物与关系怎样共同约束定位输出？","sourceSlides":["l3-r2-p63"],"knowledgeBlock":"grounding-target-reference-relation","formulas":[{"latex":"x_{max}(a)<x_{min}(r)","definitionContext":"短语 `cup left of blue cup` 中，目标类别是 cup，参照物是 blue cup，关系是 left of。看到 blue 就返回蓝杯，找到了参照物，却没有完成目标选择。\nLAB15 的受控画布中，左红杯框为 $[15,19,50,58]$，中心横坐标 32.5；右蓝杯框为 $[109,19,144,58]$，中心横坐标 126.5；书的框为 $[60,65,90,80]$。本例定义 left-of 为目标中心横坐标小于参照中心，且目标不是参照物本身。\n先找唯一蓝杯，再从其他杯子中筛选：红杯满足 $32.5<126.5$，应返回红杯框。书也在蓝杯左侧，却不满足 cup 类别约束。若蓝杯缺失、出现多个蓝杯或存在多个满足关系的目标，程序必须有明确的无目标/多义处理，不能随意返回最高颜色分的对象。\n这里说明的是一套显式关系语义。“完全在左侧”可以改用 $x_{max}(a)<x_{min}(r)$，与中心比较在重叠情况下不同。把语言问题落实为区域时，要先解释采用哪种语义，再谈模型是否正确。图上的坐标约定是半开边界框，不是逐像素包含端点的面积公式。"}],"previousBlock":"b-506963aed02c282eadab012e","nextBlock":"b-197cdb14ab29bb4e1d673d6f","prerequisites":[{"id":"b-61edd5270bb03d38a412f436","url":"#b-61edd5270bb03d38a412f436","role":"section-title"},{"id":"b-c9497c1c9165483e4bdb23bd","url":"#b-c9497c1c9165483e4bdb23bd","role":"problem-context"}],"relatedBlocks":[{"id":"b-ee065de60c5e5e93124fbcf9","url":"#b-ee065de60c5e5e93124fbcf9","relation":"figure-and-explanation"},{"id":"b-61edd5270bb03d38a412f436","url":"#b-61edd5270bb03d38a412f436","relation":"same-reasoning"},{"id":"b-8aa65492c7aa4aa160d80f7a","url":"#b-8aa65492c7aa4aa160d80f7a","relation":"same-reasoning"},{"id":"b-aa29cdf5e421c266f91a739e","url":"#b-aa29cdf5e421c266f91a739e","relation":"same-reasoning"},{"id":"b-506963aed02c282eadab012e","url":"#b-506963aed02c282eadab012e","relation":"same-reasoning"}]},{"id":"b-197cdb14ab29bb4e1d673d6f","type":"h4","scope":"retained-reading","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p63.html#b-197cdb14ab29bb4e1d673d6f","content":"#### 这条线索的展开与后续阅读","sha256":"48b8229f7559b36ed4d5628416e3991e5c73293fbce43374db67386c8736f3bc","chapter":{"id":"III","title":"视觉与语言"},"unit":{"id":"grounding-object-contract","title":"一句话怎样落实为正确对象的区域？"},"page":{"id":"grounding-target-reference","title":"沿坐标选出目标，而不是参照物","number":1},"sectionTitle":"沿坐标选出目标，而不是参照物","currentQuestion":"目标、参照物与关系怎样共同约束定位输出？","sourceSlides":["l3-r2-p63"],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-3d7c4519e6825a4bc457c63e","nextBlock":"b-ceb7e39d586b201cb5f6e9fd","prerequisites":[{"id":"b-61edd5270bb03d38a412f436","url":"#b-61edd5270bb03d38a412f436","role":"section-title"},{"id":"b-c9497c1c9165483e4bdb23bd","url":"#b-c9497c1c9165483e4bdb23bd","role":"problem-context"}],"relatedBlocks":[{"id":"b-ee065de60c5e5e93124fbcf9","url":"#b-ee065de60c5e5e93124fbcf9","relation":"figure-and-explanation"}]},{"id":"b-ceb7e39d586b201cb5f6e9fd","type":"ul","scope":"retained-reading","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p63.html#b-ceb7e39d586b201cb5f6e9fd","content":"- [GLIP：让区域对上词语，让框落在正确位置](/supplement.html?return=l3-r2-p63/0#/l3-glip/0) 本页说明\n- [先检查问题指向了哪个区域](/supplement.html?return=l3-r2-p63/0#/l3-visual-grounding/0) 本页说明\n- [语言指定区域需要新的对应监督｜问题从哪里出现](/supplement.html?return=l3-r2-p63/0#/l3-grounding-glip-problem/0) 本页说明\n- [语言指定区域需要新的对应监督｜核心改动怎样工作](/supplement.html?return=l3-r2-p63/0#/l3-grounding-glip-mechanism/0) 本页说明\n- [语言指定区域需要新的对应监督｜一起算一个完整例子](/supplement.html?return=l3-r2-p63/0#/l3-grounding-glip-example/0) 本页说明\n- [语言指定区域需要新的对应监督｜怎样检查这项解释](/supplement.html?return=l3-r2-p63/0#/l3-grounding-glip-evidence/0) 本页说明\n- [语言指定区域需要新的对应监督｜什么结论还不能推出](/supplement.html?return=l3-r2-p63/0#/l3-grounding-glip-boundary/0) 本页说明","sha256":"6846eb169f6eea9b81d65738eb8d5da77c3c3ff35cd5cb3cd8ac9606c3e53ee7","chapter":{"id":"III","title":"视觉与语言"},"unit":{"id":"grounding-object-contract","title":"一句话怎样落实为正确对象的区域？"},"page":{"id":"grounding-target-reference","title":"沿坐标选出目标，而不是参照物","number":1},"sectionTitle":"沿坐标选出目标，而不是参照物","currentQuestion":"目标、参照物与关系怎样共同约束定位输出？","sourceSlides":["l3-r2-p63"],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-197cdb14ab29bb4e1d673d6f","nextBlock":"b-97ebb336efaf78bd0c446c58","prerequisites":[{"id":"b-197cdb14ab29bb4e1d673d6f","url":"#b-197cdb14ab29bb4e1d673d6f","role":"section-title"},{"id":"b-c9497c1c9165483e4bdb23bd","url":"#b-c9497c1c9165483e4bdb23bd","role":"problem-context"}],"relatedBlocks":[{"id":"b-ee065de60c5e5e93124fbcf9","url":"#b-ee065de60c5e5e93124fbcf9","relation":"figure-and-explanation"}]},{"id":"b-97ebb336efaf78bd0c446c58","type":"li","scope":"retained-reading","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p63.html#b-97ebb336efaf78bd0c446c58","content":"- [GLIP：让区域对上词语，让框落在正确位置](/supplement.html?return=l3-r2-p63/0#/l3-glip/0) 本页说明","sha256":"d28ae5ad8854a8867c06ae0995df510cbc1a835b6039f9172788343269235477","chapter":{"id":"III","title":"视觉与语言"},"unit":{"id":"grounding-object-contract","title":"一句话怎样落实为正确对象的区域？"},"page":{"id":"grounding-target-reference","title":"沿坐标选出目标，而不是参照物","number":1},"sectionTitle":"沿坐标选出目标，而不是参照物","currentQuestion":"目标、参照物与关系怎样共同约束定位输出？","sourceSlides":["l3-r2-p63"],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-ceb7e39d586b201cb5f6e9fd","nextBlock":"b-fbd1d981831d1e9d3ea87d2b","prerequisites":[{"id":"b-197cdb14ab29bb4e1d673d6f","url":"#b-197cdb14ab29bb4e1d673d6f","role":"section-title"},{"id":"b-c9497c1c9165483e4bdb23bd","url":"#b-c9497c1c9165483e4bdb23bd","role":"problem-context"}],"relatedBlocks":[{"id":"b-ee065de60c5e5e93124fbcf9","url":"#b-ee065de60c5e5e93124fbcf9","relation":"figure-and-explanation"}]},{"id":"b-fbd1d981831d1e9d3ea87d2b","type":"li","scope":"retained-reading","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p63.html#b-fbd1d981831d1e9d3ea87d2b","content":"- [先检查问题指向了哪个区域](/supplement.html?return=l3-r2-p63/0#/l3-visual-grounding/0) 本页说明","sha256":"3a685b8b95c708b86b245f39cdbde5d8c6936c29273c99261590abb9422d04f1","chapter":{"id":"III","title":"视觉与语言"},"unit":{"id":"grounding-object-contract","title":"一句话怎样落实为正确对象的区域？"},"page":{"id":"grounding-target-reference","title":"沿坐标选出目标，而不是参照物","number":1},"sectionTitle":"沿坐标选出目标，而不是参照物","currentQuestion":"目标、参照物与关系怎样共同约束定位输出？","sourceSlides":["l3-r2-p63"],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-97ebb336efaf78bd0c446c58","nextBlock":"b-17fcb7df9471d529d60daf62","prerequisites":[{"id":"b-197cdb14ab29bb4e1d673d6f","url":"#b-197cdb14ab29bb4e1d673d6f","role":"section-title"},{"id":"b-c9497c1c9165483e4bdb23bd","url":"#b-c9497c1c9165483e4bdb23bd","role":"problem-context"}],"relatedBlocks":[{"id":"b-ee065de60c5e5e93124fbcf9","url":"#b-ee065de60c5e5e93124fbcf9","relation":"figure-and-explanation"}]},{"id":"b-17fcb7df9471d529d60daf62","type":"li","scope":"retained-reading","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p63.html#b-17fcb7df9471d529d60daf62","content":"- [语言指定区域需要新的对应监督｜问题从哪里出现](/supplement.html?return=l3-r2-p63/0#/l3-grounding-glip-problem/0) 本页说明","sha256":"875ade1b239910a6741fd477b16f4f1e09b84b25bd4c731ef99b391ab597d300","chapter":{"id":"III","title":"视觉与语言"},"unit":{"id":"grounding-object-contract","title":"一句话怎样落实为正确对象的区域？"},"page":{"id":"grounding-target-reference","title":"沿坐标选出目标，而不是参照物","number":1},"sectionTitle":"沿坐标选出目标，而不是参照物","currentQuestion":"目标、参照物与关系怎样共同约束定位输出？","sourceSlides":["l3-r2-p63"],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-fbd1d981831d1e9d3ea87d2b","nextBlock":"b-a99c507166df44bf48000e50","prerequisites":[{"id":"b-197cdb14ab29bb4e1d673d6f","url":"#b-197cdb14ab29bb4e1d673d6f","role":"section-title"},{"id":"b-c9497c1c9165483e4bdb23bd","url":"#b-c9497c1c9165483e4bdb23bd","role":"problem-context"}],"relatedBlocks":[{"id":"b-ee065de60c5e5e93124fbcf9","url":"#b-ee065de60c5e5e93124fbcf9","relation":"figure-and-explanation"}]},{"id":"b-a99c507166df44bf48000e50","type":"li","scope":"retained-reading","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p63.html#b-a99c507166df44bf48000e50","content":"- [语言指定区域需要新的对应监督｜核心改动怎样工作](/supplement.html?return=l3-r2-p63/0#/l3-grounding-glip-mechanism/0) 本页说明","sha256":"7768902d58ef57b20c88ac26d2fec7ae259c1fcfaee5c150e2530364af2e5027","chapter":{"id":"III","title":"视觉与语言"},"unit":{"id":"grounding-object-contract","title":"一句话怎样落实为正确对象的区域？"},"page":{"id":"grounding-target-reference","title":"沿坐标选出目标，而不是参照物","number":1},"sectionTitle":"沿坐标选出目标，而不是参照物","currentQuestion":"目标、参照物与关系怎样共同约束定位输出？","sourceSlides":["l3-r2-p63"],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-17fcb7df9471d529d60daf62","nextBlock":"b-8c732a78745b7f13f26979da","prerequisites":[{"id":"b-197cdb14ab29bb4e1d673d6f","url":"#b-197cdb14ab29bb4e1d673d6f","role":"section-title"},{"id":"b-c9497c1c9165483e4bdb23bd","url":"#b-c9497c1c9165483e4bdb23bd","role":"problem-context"}],"relatedBlocks":[{"id":"b-ee065de60c5e5e93124fbcf9","url":"#b-ee065de60c5e5e93124fbcf9","relation":"figure-and-explanation"}]},{"id":"b-8c732a78745b7f13f26979da","type":"li","scope":"retained-reading","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p63.html#b-8c732a78745b7f13f26979da","content":"- [语言指定区域需要新的对应监督｜一起算一个完整例子](/supplement.html?return=l3-r2-p63/0#/l3-grounding-glip-example/0) 本页说明","sha256":"9417752cc706088cad5cf42f725b2bbbd10f191fcbaa05ffd8202581eeca17ee","chapter":{"id":"III","title":"视觉与语言"},"unit":{"id":"grounding-object-contract","title":"一句话怎样落实为正确对象的区域？"},"page":{"id":"grounding-target-reference","title":"沿坐标选出目标，而不是参照物","number":1},"sectionTitle":"沿坐标选出目标，而不是参照物","currentQuestion":"目标、参照物与关系怎样共同约束定位输出？","sourceSlides":["l3-r2-p63"],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-a99c507166df44bf48000e50","nextBlock":"b-d67e4e9281001cc6c05e1314","prerequisites":[{"id":"b-197cdb14ab29bb4e1d673d6f","url":"#b-197cdb14ab29bb4e1d673d6f","role":"section-title"},{"id":"b-c9497c1c9165483e4bdb23bd","url":"#b-c9497c1c9165483e4bdb23bd","role":"problem-context"}],"relatedBlocks":[{"id":"b-ee065de60c5e5e93124fbcf9","url":"#b-ee065de60c5e5e93124fbcf9","relation":"figure-and-explanation"}]},{"id":"b-d67e4e9281001cc6c05e1314","type":"li","scope":"retained-reading","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p63.html#b-d67e4e9281001cc6c05e1314","content":"- [语言指定区域需要新的对应监督｜怎样检查这项解释](/supplement.html?return=l3-r2-p63/0#/l3-grounding-glip-evidence/0) 本页说明","sha256":"2508a2657e8a66706205f014cdb1116ab56d8dd9eb786410309cd124d243eb03","chapter":{"id":"III","title":"视觉与语言"},"unit":{"id":"grounding-object-contract","title":"一句话怎样落实为正确对象的区域？"},"page":{"id":"grounding-target-reference","title":"沿坐标选出目标，而不是参照物","number":1},"sectionTitle":"沿坐标选出目标，而不是参照物","currentQuestion":"目标、参照物与关系怎样共同约束定位输出？","sourceSlides":["l3-r2-p63"],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-8c732a78745b7f13f26979da","nextBlock":"b-332609e5fd09e2c21c9bd95f","prerequisites":[{"id":"b-197cdb14ab29bb4e1d673d6f","url":"#b-197cdb14ab29bb4e1d673d6f","role":"section-title"},{"id":"b-c9497c1c9165483e4bdb23bd","url":"#b-c9497c1c9165483e4bdb23bd","role":"problem-context"}],"relatedBlocks":[{"id":"b-ee065de60c5e5e93124fbcf9","url":"#b-ee065de60c5e5e93124fbcf9","relation":"figure-and-explanation"}]},{"id":"b-332609e5fd09e2c21c9bd95f","type":"li","scope":"retained-reading","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p63.html#b-332609e5fd09e2c21c9bd95f","content":"- [语言指定区域需要新的对应监督｜什么结论还不能推出](/supplement.html?return=l3-r2-p63/0#/l3-grounding-glip-boundary/0) 本页说明","sha256":"bca26edd21c30a247e6d973495e11bc2f557af02d1b0ff794c9123e1dc0c4123","chapter":{"id":"III","title":"视觉与语言"},"unit":{"id":"grounding-object-contract","title":"一句话怎样落实为正确对象的区域？"},"page":{"id":"grounding-target-reference","title":"沿坐标选出目标，而不是参照物","number":1},"sectionTitle":"沿坐标选出目标，而不是参照物","currentQuestion":"目标、参照物与关系怎样共同约束定位输出？","sourceSlides":["l3-r2-p63"],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-d67e4e9281001cc6c05e1314","nextBlock":null,"prerequisites":[{"id":"b-197cdb14ab29bb4e1d673d6f","url":"#b-197cdb14ab29bb4e1d673d6f","role":"section-title"},{"id":"b-c9497c1c9165483e4bdb23bd","url":"#b-c9497c1c9165483e4bdb23bd","role":"problem-context"}],"relatedBlocks":[{"id":"b-ee065de60c5e5e93124fbcf9","url":"#b-ee065de60c5e5e93124fbcf9","relation":"figure-and-explanation"}]},{"id":"b-f161c449ecedb21ef539390a","type":"h2","scope":"grounding-lab15-interventions","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p63.html#b-f161c449ecedb21ef539390a","content":"## 从输入干预追到实际返回框","sha256":"51fefcbc6ba7dacda763983b5bcb963667f3fad7d99713458ee05f6e01ff940b","chapter":{"id":"III","title":"视觉与语言"},"unit":{"id":"grounding-object-contract","title":"一句话怎样落实为正确对象的区域？"},"page":{"id":"grounding-lab15-interventions","title":"从输入干预追到实际返回框","number":2},"sectionTitle":"从输入干预追到实际返回框","currentQuestion":"目标、参照物与关系怎样共同约束定位输出？","sourceSlides":["l3-r2-p63"],"knowledgeBlock":null,"formulas":[],"previousBlock":null,"nextBlock":"b-360162cda01cb1cddbf43ec2","prerequisites":[{"id":"b-360162cda01cb1cddbf43ec2","url":"#b-360162cda01cb1cddbf43ec2","role":"problem-context"}],"relatedBlocks":[]},{"id":"b-360162cda01cb1cddbf43ec2","type":"p","scope":"grounding-lab15-interventions","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p63.html#b-360162cda01cb1cddbf43ec2","content":"目标语义已定义；现在用保存的像素规则系统逐项检查输入怎样改变候选和答案。","sha256":"321b45e629fa324edc1fc103679cfb3d3c940f167954ccbc1458332aee3c442a","chapter":{"id":"III","title":"视觉与语言"},"unit":{"id":"grounding-object-contract","title":"一句话怎样落实为正确对象的区域？"},"page":{"id":"grounding-lab15-interventions","title":"从输入干预追到实际返回框","number":2},"sectionTitle":"从输入干预追到实际返回框","currentQuestion":"目标、参照物与关系怎样共同约束定位输出？","sourceSlides":["l3-r2-p63"],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-f161c449ecedb21ef539390a","nextBlock":"b-830446b5c23c16f0cb906b5e","prerequisites":[{"id":"b-f161c449ecedb21ef539390a","url":"#b-f161c449ecedb21ef539390a","role":"section-title"}],"relatedBlocks":[]},{"id":"b-830446b5c23c16f0cb906b5e","type":"h3","scope":"grounding-pixel-interventions","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p63.html#b-830446b5c23c16f0cb906b5e","content":"### 已保存的指代实验：从像素提框后再应用关系","sha256":"b919cfcf83f000f7f330246ddb469148384c48ca59f8b48b976b98d36652228e","chapter":{"id":"III","title":"视觉与语言"},"unit":{"id":"grounding-object-contract","title":"一句话怎样落实为正确对象的区域？"},"page":{"id":"grounding-lab15-interventions","title":"从输入干预追到实际返回框","number":2},"sectionTitle":"已保存的指代实验：从像素提框后再应用关系","currentQuestion":"目标、参照物与关系怎样共同约束定位输出？","sourceSlides":["l3-r2-p63"],"knowledgeBlock":"grounding-pixel-interventions","formulas":[],"previousBlock":"b-360162cda01cb1cddbf43ec2","nextBlock":"b-25b6cd118879b490a311e463","prerequisites":[{"id":"b-f161c449ecedb21ef539390a","url":"#b-f161c449ecedb21ef539390a","role":"section-title"},{"id":"b-360162cda01cb1cddbf43ec2","url":"#b-360162cda01cb1cddbf43ec2","role":"problem-context"}],"relatedBlocks":[{"id":"b-25b6cd118879b490a311e463","url":"#b-25b6cd118879b490a311e463","relation":"same-reasoning"},{"id":"b-c54caf119e7797bf6e8d13cb","url":"#b-c54caf119e7797bf6e8d13cb","relation":"same-reasoning"},{"id":"b-d8b9555619e7c4e2dc3548f0","url":"#b-d8b9555619e7c4e2dc3548f0","relation":"same-reasoning"},{"id":"b-6ec0eaa5d6ba03a6ac8bb3f0","url":"#b-6ec0eaa5d6ba03a6ac8bb3f0","relation":"same-reasoning"},{"id":"b-830b8f7e40668ab1598dd62e","url":"#b-830b8f7e40668ab1598dd62e","relation":"same-reasoning"}]},{"id":"b-25b6cd118879b490a311e463","type":"p","scope":"grounding-pixel-interventions","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p63.html#b-25b6cd118879b490a311e463","content":"LAB15 的当前实现先从 RGB 图像按颜色阈值提取连通域，得到候选框；再解析目标、参照物与关系。它是真正读取像素的规则系统，类别由有限颜色约定提供，没有训练神经 grounding 模型。这个身份既不应被贬为“没读图”，也不应被升级为“复现 GLIP”。","sha256":"ec817964628fe547da360e42935f3ab5e1d4ab0a413223243d9b9bdd08b46be5","chapter":{"id":"III","title":"视觉与语言"},"unit":{"id":"grounding-object-contract","title":"一句话怎样落实为正确对象的区域？"},"page":{"id":"grounding-lab15-interventions","title":"从输入干预追到实际返回框","number":2},"sectionTitle":"已保存的指代实验：从像素提框后再应用关系","currentQuestion":"目标、参照物与关系怎样共同约束定位输出？","sourceSlides":["l3-r2-p63"],"knowledgeBlock":"grounding-pixel-interventions","formulas":[],"previousBlock":"b-830446b5c23c16f0cb906b5e","nextBlock":"b-c54caf119e7797bf6e8d13cb","prerequisites":[{"id":"b-830446b5c23c16f0cb906b5e","url":"#b-830446b5c23c16f0cb906b5e","role":"section-title"},{"id":"b-360162cda01cb1cddbf43ec2","url":"#b-360162cda01cb1cddbf43ec2","role":"problem-context"}],"relatedBlocks":[{"id":"b-830446b5c23c16f0cb906b5e","url":"#b-830446b5c23c16f0cb906b5e","relation":"same-reasoning"},{"id":"b-c54caf119e7797bf6e8d13cb","url":"#b-c54caf119e7797bf6e8d13cb","relation":"same-reasoning"},{"id":"b-d8b9555619e7c4e2dc3548f0","url":"#b-d8b9555619e7c4e2dc3548f0","relation":"same-reasoning"},{"id":"b-6ec0eaa5d6ba03a6ac8bb3f0","url":"#b-6ec0eaa5d6ba03a6ac8bb3f0","relation":"same-reasoning"},{"id":"b-830b8f7e40668ab1598dd62e","url":"#b-830b8f7e40668ab1598dd62e","relation":"same-reasoning"}]},{"id":"b-c54caf119e7797bf6e8d13cb","type":"table","scope":"grounding-pixel-interventions","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p63.html#b-c54caf119e7797bf6e8d13cb","content":"| 实际输入改变 | 问句 | 保存的输出 |\n| --- | --- | --- |\n| 原图左红、右蓝 | 蓝杯左边的杯子 | 左红杯 |\n| 灰块遮住左杯 | 同上 | 无唯一可见目标 |\n| 只交换两杯 RGB 色值 | 同上 | 无目标，蓝杯已在最左 |\n| 保持交换图，只把 left 改 right | 蓝杯右边的杯子 | 右红杯 |","sha256":"0de466cf5672ac86c9dd4214b39ae872dd46ee32d4256b3291c661a7a495c361","chapter":{"id":"III","title":"视觉与语言"},"unit":{"id":"grounding-object-contract","title":"一句话怎样落实为正确对象的区域？"},"page":{"id":"grounding-lab15-interventions","title":"从输入干预追到实际返回框","number":2},"sectionTitle":"已保存的指代实验：从像素提框后再应用关系","currentQuestion":"目标、参照物与关系怎样共同约束定位输出？","sourceSlides":["l3-r2-p63"],"knowledgeBlock":"grounding-pixel-interventions","formulas":[],"table":{"headers":["实际输入改变","问句","保存的输出"],"rows":[["实际输入改变","问句","保存的输出"],["原图左红、右蓝","蓝杯左边的杯子","左红杯"],["灰块遮住左杯","同上","无唯一可见目标"],["只交换两杯 RGB 色值","同上","无目标，蓝杯已在最左"],["保持交换图，只把 left 改 right","蓝杯右边的杯子","右红杯"]]},"previousBlock":"b-25b6cd118879b490a311e463","nextBlock":"b-d8b9555619e7c4e2dc3548f0","prerequisites":[{"id":"b-830446b5c23c16f0cb906b5e","url":"#b-830446b5c23c16f0cb906b5e","role":"section-title"},{"id":"b-360162cda01cb1cddbf43ec2","url":"#b-360162cda01cb1cddbf43ec2","role":"problem-context"}],"relatedBlocks":[{"id":"b-830446b5c23c16f0cb906b5e","url":"#b-830446b5c23c16f0cb906b5e","relation":"same-reasoning"},{"id":"b-25b6cd118879b490a311e463","url":"#b-25b6cd118879b490a311e463","relation":"same-reasoning"},{"id":"b-d8b9555619e7c4e2dc3548f0","url":"#b-d8b9555619e7c4e2dc3548f0","relation":"same-reasoning"},{"id":"b-6ec0eaa5d6ba03a6ac8bb3f0","url":"#b-6ec0eaa5d6ba03a6ac8bb3f0","relation":"same-reasoning"},{"id":"b-830b8f7e40668ab1598dd62e","url":"#b-830b8f7e40668ab1598dd62e","relation":"same-reasoning"}]},{"id":"b-d8b9555619e7c4e2dc3548f0","type":"p","scope":"grounding-pixel-interventions","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p63.html#b-d8b9555619e7c4e2dc3548f0","content":"遮挡是真的修改图片，不是在答案字典里删除目标；交换色值保留轮廓和背景；语言对照保持图片完全相同。代码断言干预区域外像素不变，并保存差分图。它因此能够支持这些有限条件下“图片怎样改变候选、关系怎样选择目标”的解释。","sha256":"55fb5f3349686e772ea73d62b88bc243f3fdb3efd45df8c2454262f799c64329","chapter":{"id":"III","title":"视觉与语言"},"unit":{"id":"grounding-object-contract","title":"一句话怎样落实为正确对象的区域？"},"page":{"id":"grounding-lab15-interventions","title":"从输入干预追到实际返回框","number":2},"sectionTitle":"已保存的指代实验：从像素提框后再应用关系","currentQuestion":"目标、参照物与关系怎样共同约束定位输出？","sourceSlides":["l3-r2-p63"],"knowledgeBlock":"grounding-pixel-interventions","formulas":[],"previousBlock":"b-c54caf119e7797bf6e8d13cb","nextBlock":"b-6ec0eaa5d6ba03a6ac8bb3f0","prerequisites":[{"id":"b-830446b5c23c16f0cb906b5e","url":"#b-830446b5c23c16f0cb906b5e","role":"section-title"},{"id":"b-360162cda01cb1cddbf43ec2","url":"#b-360162cda01cb1cddbf43ec2","role":"problem-context"}],"relatedBlocks":[{"id":"b-830446b5c23c16f0cb906b5e","url":"#b-830446b5c23c16f0cb906b5e","relation":"same-reasoning"},{"id":"b-25b6cd118879b490a311e463","url":"#b-25b6cd118879b490a311e463","relation":"same-reasoning"},{"id":"b-c54caf119e7797bf6e8d13cb","url":"#b-c54caf119e7797bf6e8d13cb","relation":"same-reasoning"},{"id":"b-6ec0eaa5d6ba03a6ac8bb3f0","url":"#b-6ec0eaa5d6ba03a6ac8bb3f0","relation":"same-reasoning"},{"id":"b-830b8f7e40668ab1598dd62e","url":"#b-830b8f7e40668ab1598dd62e","relation":"same-reasoning"}]},{"id":"b-6ec0eaa5d6ba03a6ac8bb3f0","type":"p","scope":"grounding-pixel-interventions","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p63.html#b-6ec0eaa5d6ba03a6ac8bb3f0","content":"![输入干预与返回框](/assets/display/969fa088b392b7fb8a846fb3.webp)","sha256":"70aab124a4de4ccb1629c768a54fc062777fd23a1d521943a55fc0d99ea993f7","chapter":{"id":"III","title":"视觉与语言"},"unit":{"id":"grounding-object-contract","title":"一句话怎样落实为正确对象的区域？"},"page":{"id":"grounding-lab15-interventions","title":"从输入干预追到实际返回框","number":2},"sectionTitle":"已保存的指代实验：从像素提框后再应用关系","currentQuestion":"目标、参照物与关系怎样共同约束定位输出？","sourceSlides":["l3-r2-p63"],"knowledgeBlock":"grounding-pixel-interventions","formulas":[],"previousBlock":"b-d8b9555619e7c4e2dc3548f0","nextBlock":"b-830b8f7e40668ab1598dd62e","prerequisites":[{"id":"b-830446b5c23c16f0cb906b5e","url":"#b-830446b5c23c16f0cb906b5e","role":"section-title"},{"id":"b-360162cda01cb1cddbf43ec2","url":"#b-360162cda01cb1cddbf43ec2","role":"problem-context"}],"relatedBlocks":[{"id":"b-830446b5c23c16f0cb906b5e","url":"#b-830446b5c23c16f0cb906b5e","relation":"same-reasoning"},{"id":"b-25b6cd118879b490a311e463","url":"#b-25b6cd118879b490a311e463","relation":"same-reasoning"},{"id":"b-c54caf119e7797bf6e8d13cb","url":"#b-c54caf119e7797bf6e8d13cb","relation":"same-reasoning"},{"id":"b-d8b9555619e7c4e2dc3548f0","url":"#b-d8b9555619e7c4e2dc3548f0","relation":"same-reasoning"},{"id":"b-830b8f7e40668ab1598dd62e","url":"#b-830b8f7e40668ab1598dd62e","relation":"same-reasoning"}]},{"id":"b-830b8f7e40668ab1598dd62e","type":"p","scope":"grounding-pixel-interventions","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p63.html#b-830b8f7e40668ab1598dd62e","content":"旧关键词规则错选蓝杯仍保留为失败基线。当前结果来自 2026-09-22 已保存运行，本轮未重跑。若以后换自然照片，颜色阈值和杯子类别约定会失效；那时才需要加载相应视觉模型，并对输入和实际框做新的可追溯评价。","sha256":"2c856e57e1c37c3d70093e2c5de7c9890f3a1bfe364eef8129588edb5dd7a622","chapter":{"id":"III","title":"视觉与语言"},"unit":{"id":"grounding-object-contract","title":"一句话怎样落实为正确对象的区域？"},"page":{"id":"grounding-lab15-interventions","title":"从输入干预追到实际返回框","number":2},"sectionTitle":"已保存的指代实验：从像素提框后再应用关系","currentQuestion":"目标、参照物与关系怎样共同约束定位输出？","sourceSlides":["l3-r2-p63"],"knowledgeBlock":"grounding-pixel-interventions","formulas":[],"previousBlock":"b-6ec0eaa5d6ba03a6ac8bb3f0","nextBlock":null,"prerequisites":[{"id":"b-830446b5c23c16f0cb906b5e","url":"#b-830446b5c23c16f0cb906b5e","role":"section-title"},{"id":"b-360162cda01cb1cddbf43ec2","url":"#b-360162cda01cb1cddbf43ec2","role":"problem-context"}],"relatedBlocks":[{"id":"b-830446b5c23c16f0cb906b5e","url":"#b-830446b5c23c16f0cb906b5e","relation":"same-reasoning"},{"id":"b-25b6cd118879b490a311e463","url":"#b-25b6cd118879b490a311e463","relation":"same-reasoning"},{"id":"b-c54caf119e7797bf6e8d13cb","url":"#b-c54caf119e7797bf6e8d13cb","relation":"same-reasoning"},{"id":"b-d8b9555619e7c4e2dc3548f0","url":"#b-d8b9555619e7c4e2dc3548f0","relation":"same-reasoning"},{"id":"b-6ec0eaa5d6ba03a6ac8bb3f0","url":"#b-6ec0eaa5d6ba03a6ac8bb3f0","relation":"same-reasoning"}]},{"id":"b-d50a51f982259222c44dc219","type":"h2","scope":"glip-language-localization","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p63.html#b-d50a51f982259222c44dc219","content":"## 从透明规则走向对象级学习方案","sha256":"f13b95e2fb7b9dd925c75d862fbfd506224c8c25bba10022da9336d4eb23c241","chapter":{"id":"III","title":"视觉与语言"},"unit":{"id":"grounding-object-contract","title":"一句话怎样落实为正确对象的区域？"},"page":{"id":"glip-language-localization","title":"从透明规则走向对象级学习方案","number":3},"sectionTitle":"从透明规则走向对象级学习方案","currentQuestion":"目标、参照物与关系怎样共同约束定位输出？","sourceSlides":["l3-r2-p63"],"knowledgeBlock":null,"formulas":[],"previousBlock":null,"nextBlock":"b-1c989e608aeddcda3cce73c3","prerequisites":[{"id":"b-1c989e608aeddcda3cce73c3","url":"#b-1c989e608aeddcda3cce73c3","role":"problem-context"}],"relatedBlocks":[]},{"id":"b-1c989e608aeddcda3cce73c3","type":"p","scope":"glip-language-localization","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p63.html#b-1c989e608aeddcda3cce73c3","content":"像素规则实验解释了任务契约；GLIP 展示语言与区域可共同学习的研究路线。","sha256":"4e9aefff1043b6dc02cea07e2ec5a69c467a4f2e43540dfa31e7c44cc7d1f28b","chapter":{"id":"III","title":"视觉与语言"},"unit":{"id":"grounding-object-contract","title":"一句话怎样落实为正确对象的区域？"},"page":{"id":"glip-language-localization","title":"从透明规则走向对象级学习方案","number":3},"sectionTitle":"从透明规则走向对象级学习方案","currentQuestion":"目标、参照物与关系怎样共同约束定位输出？","sourceSlides":["l3-r2-p63"],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-d50a51f982259222c44dc219","nextBlock":"b-7fbed2e85325174c9948694f","prerequisites":[{"id":"b-d50a51f982259222c44dc219","url":"#b-d50a51f982259222c44dc219","role":"section-title"}],"relatedBlocks":[]},{"id":"b-7fbed2e85325174c9948694f","type":"h3","scope":"glip-language-objective","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p63.html#b-7fbed2e85325174c9948694f","content":"### GLIP：把类别和短语都放到对象级定位任务中","sha256":"f7f5551cc7a0af3daebec2868febbf8d91ddd869713f104db0b7c9920e28a535","chapter":{"id":"III","title":"视觉与语言"},"unit":{"id":"grounding-object-contract","title":"一句话怎样落实为正确对象的区域？"},"page":{"id":"glip-language-localization","title":"从透明规则走向对象级学习方案","number":3},"sectionTitle":"GLIP：把类别和短语都放到对象级定位任务中","currentQuestion":"目标、参照物与关系怎样共同约束定位输出？","sourceSlides":["l3-r2-p63"],"knowledgeBlock":"glip-language-objective","formulas":[],"previousBlock":"b-1c989e608aeddcda3cce73c3","nextBlock":"b-314b2f66ec6db241fbc09f45","prerequisites":[{"id":"b-d50a51f982259222c44dc219","url":"#b-d50a51f982259222c44dc219","role":"section-title"},{"id":"b-1c989e608aeddcda3cce73c3","url":"#b-1c989e608aeddcda3cce73c3","role":"problem-context"}],"relatedBlocks":[{"id":"b-314b2f66ec6db241fbc09f45","url":"#b-314b2f66ec6db241fbc09f45","relation":"same-reasoning"},{"id":"b-069046fefcdf06847c400fb2","url":"#b-069046fefcdf06847c400fb2","relation":"same-reasoning"},{"id":"b-3d6cd84caa9d9ba96e8d6987","url":"#b-3d6cd84caa9d9ba96e8d6987","relation":"same-reasoning"},{"id":"b-dcef8d2d7d7fa2e97ec5f919","url":"#b-dcef8d2d7d7fa2e97ec5f919","relation":"same-reasoning"}]},{"id":"b-314b2f66ec6db241fbc09f45","type":"p","scope":"glip-language-objective","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p63.html#b-314b2f66ec6db241fbc09f45","content":"GLIP 将对象检测与短语定位用于共同的预训练。传统检测类别可以转写成文字概念，短语资料又提供更开放的语言—区域对应。视觉与文字发生交互，训练同时要求语义匹配和位置正确；这比给一组外部裁图计算 CLIP 整图分数多了任务结构与定位监督。","sha256":"24baff1cf87b2de7ca97ba1336e8c33d4560ab39283b2ef8c7c1c36dba4b50fc","chapter":{"id":"III","title":"视觉与语言"},"unit":{"id":"grounding-object-contract","title":"一句话怎样落实为正确对象的区域？"},"page":{"id":"glip-language-localization","title":"从透明规则走向对象级学习方案","number":3},"sectionTitle":"GLIP：把类别和短语都放到对象级定位任务中","currentQuestion":"目标、参照物与关系怎样共同约束定位输出？","sourceSlides":["l3-r2-p63"],"knowledgeBlock":"glip-language-objective","formulas":[],"previousBlock":"b-7fbed2e85325174c9948694f","nextBlock":"b-069046fefcdf06847c400fb2","prerequisites":[{"id":"b-7fbed2e85325174c9948694f","url":"#b-7fbed2e85325174c9948694f","role":"section-title"},{"id":"b-1c989e608aeddcda3cce73c3","url":"#b-1c989e608aeddcda3cce73c3","role":"problem-context"}],"relatedBlocks":[{"id":"b-7fbed2e85325174c9948694f","url":"#b-7fbed2e85325174c9948694f","relation":"same-reasoning"},{"id":"b-069046fefcdf06847c400fb2","url":"#b-069046fefcdf06847c400fb2","relation":"same-reasoning"},{"id":"b-3d6cd84caa9d9ba96e8d6987","url":"#b-3d6cd84caa9d9ba96e8d6987","relation":"same-reasoning"},{"id":"b-dcef8d2d7d7fa2e97ec5f919","url":"#b-dcef8d2d7d7fa2e97ec5f919","relation":"same-reasoning"}]},{"id":"b-069046fefcdf06847c400fb2","type":"p","scope":"glip-language-objective","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p63.html#b-069046fefcdf06847c400fb2","content":"原论文使用 2700 万 grounding 训练样本，包含人工标注和由网络图文扩展的资料；在其协议中，直接评估 COCO 得到 49.8 AP，LVIS 得到 26.9 AP，且说明预训练没有使用 COCO 图像。这里的数字是论文所报告的对象级迁移结果，不是本课程持球图的运行。它支持语言条件和区域学习能够带来可迁移检测接口这一方向。","sha256":"80dfe0ea15830ea4065fc5c1744a90f46c42c9fa95ba56ae9d2715668a597d60","chapter":{"id":"III","title":"视觉与语言"},"unit":{"id":"grounding-object-contract","title":"一句话怎样落实为正确对象的区域？"},"page":{"id":"glip-language-localization","title":"从透明规则走向对象级学习方案","number":3},"sectionTitle":"GLIP：把类别和短语都放到对象级定位任务中","currentQuestion":"目标、参照物与关系怎样共同约束定位输出？","sourceSlides":["l3-r2-p63"],"knowledgeBlock":"glip-language-objective","formulas":[],"previousBlock":"b-314b2f66ec6db241fbc09f45","nextBlock":"b-3d6cd84caa9d9ba96e8d6987","prerequisites":[{"id":"b-7fbed2e85325174c9948694f","url":"#b-7fbed2e85325174c9948694f","role":"section-title"},{"id":"b-1c989e608aeddcda3cce73c3","url":"#b-1c989e608aeddcda3cce73c3","role":"problem-context"}],"relatedBlocks":[{"id":"b-7fbed2e85325174c9948694f","url":"#b-7fbed2e85325174c9948694f","relation":"same-reasoning"},{"id":"b-314b2f66ec6db241fbc09f45","url":"#b-314b2f66ec6db241fbc09f45","relation":"same-reasoning"},{"id":"b-3d6cd84caa9d9ba96e8d6987","url":"#b-3d6cd84caa9d9ba96e8d6987","relation":"same-reasoning"},{"id":"b-dcef8d2d7d7fa2e97ec5f919","url":"#b-dcef8d2d7d7fa2e97ec5f919","relation":"same-reasoning"}]},{"id":"b-3d6cd84caa9d9ba96e8d6987","type":"p","scope":"glip-language-objective","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p63.html#b-3d6cd84caa9d9ba96e8d6987","content":"对课堂场景，“拿球的蓝衣人”要求图中多个对象、属性与关系共同约束一个框。若候选框身份错了，后接分割模型即使边界非常准确，也会分错对象。反过来，目标身份正确但边界粗糙，应检查分割与采样接口。把定位和掩码衔接时分开检查，正是第二部分仍然有用的原因。","sha256":"bb4028ef924f2a8d235dbf0f1ba9b2b386877b47814234f5a6e6b5a0f191e153","chapter":{"id":"III","title":"视觉与语言"},"unit":{"id":"grounding-object-contract","title":"一句话怎样落实为正确对象的区域？"},"page":{"id":"glip-language-localization","title":"从透明规则走向对象级学习方案","number":3},"sectionTitle":"GLIP：把类别和短语都放到对象级定位任务中","currentQuestion":"目标、参照物与关系怎样共同约束定位输出？","sourceSlides":["l3-r2-p63"],"knowledgeBlock":"glip-language-objective","formulas":[],"previousBlock":"b-069046fefcdf06847c400fb2","nextBlock":"b-dcef8d2d7d7fa2e97ec5f919","prerequisites":[{"id":"b-7fbed2e85325174c9948694f","url":"#b-7fbed2e85325174c9948694f","role":"section-title"},{"id":"b-1c989e608aeddcda3cce73c3","url":"#b-1c989e608aeddcda3cce73c3","role":"problem-context"}],"relatedBlocks":[{"id":"b-7fbed2e85325174c9948694f","url":"#b-7fbed2e85325174c9948694f","relation":"same-reasoning"},{"id":"b-314b2f66ec6db241fbc09f45","url":"#b-314b2f66ec6db241fbc09f45","relation":"same-reasoning"},{"id":"b-069046fefcdf06847c400fb2","url":"#b-069046fefcdf06847c400fb2","relation":"same-reasoning"},{"id":"b-dcef8d2d7d7fa2e97ec5f919","url":"#b-dcef8d2d7d7fa2e97ec5f919","relation":"same-reasoning"}]},{"id":"b-dcef8d2d7d7fa2e97ec5f919","type":"p","scope":"glip-language-objective","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p63.html#b-dcef8d2d7d7fa2e97ec5f919","content":"LAB15 负责透明的语言关系规则，GLIP 论文负责更一般的学习方案与研究证据。两者互相解释，不互相冒充。主讲保留这一对象级接口；完整训练配置与更广数据对照留材料阅读。","sha256":"acb849912de164fda2c6b1f36086f33e94a4c1f8fc5b5957a303c674c13a844f","chapter":{"id":"III","title":"视觉与语言"},"unit":{"id":"grounding-object-contract","title":"一句话怎样落实为正确对象的区域？"},"page":{"id":"glip-language-localization","title":"从透明规则走向对象级学习方案","number":3},"sectionTitle":"GLIP：把类别和短语都放到对象级定位任务中","currentQuestion":"目标、参照物与关系怎样共同约束定位输出？","sourceSlides":["l3-r2-p63"],"knowledgeBlock":"glip-language-objective","formulas":[],"previousBlock":"b-3d6cd84caa9d9ba96e8d6987","nextBlock":"b-1d9b0885f2a4a66578c780f7","prerequisites":[{"id":"b-7fbed2e85325174c9948694f","url":"#b-7fbed2e85325174c9948694f","role":"section-title"},{"id":"b-1c989e608aeddcda3cce73c3","url":"#b-1c989e608aeddcda3cce73c3","role":"problem-context"}],"relatedBlocks":[{"id":"b-7fbed2e85325174c9948694f","url":"#b-7fbed2e85325174c9948694f","relation":"same-reasoning"},{"id":"b-314b2f66ec6db241fbc09f45","url":"#b-314b2f66ec6db241fbc09f45","relation":"same-reasoning"},{"id":"b-069046fefcdf06847c400fb2","url":"#b-069046fefcdf06847c400fb2","relation":"same-reasoning"},{"id":"b-3d6cd84caa9d9ba96e8d6987","url":"#b-3d6cd84caa9d9ba96e8d6987","relation":"same-reasoning"}]},{"id":"b-1d9b0885f2a4a66578c780f7","type":"h2","scope":"practice","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p63.html#b-1d9b0885f2a4a66578c780f7","content":"## 沿这个问题继续","sha256":"cf44fcc62782fb4f6bdc42c16c5187d57556a03cb7ce6065e7bd9c36d8e7d733","chapter":{"id":"III","title":"视觉与语言"},"unit":{"id":"grounding-object-contract","title":"一句话怎样落实为正确对象的区域？"},"page":{"id":"glip-language-localization","title":"从透明规则走向对象级学习方案","number":3},"sectionTitle":"从透明规则走向对象级学习方案","currentQuestion":"目标、参照物与关系怎样共同约束定位输出？","sourceSlides":["l3-r2-p63"],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-dcef8d2d7d7fa2e97ec5f919","nextBlock":"b-7cc00c2dd5b32b41f709450a","prerequisites":[{"id":"b-7fbed2e85325174c9948694f","url":"#b-7fbed2e85325174c9948694f","role":"section-title"},{"id":"b-1c989e608aeddcda3cce73c3","url":"#b-1c989e608aeddcda3cce73c3","role":"problem-context"}],"relatedBlocks":[]},{"id":"b-7cc00c2dd5b32b41f709450a","type":"p","scope":"practice","url":"https://codingai-lec03.pages.dev/course/pages/l3-r2-p63.html#b-7cc00c2dd5b32b41f709450a","content":"先回看本页画面并指出输入、条件和输出，再按““蓝杯左边的杯子”：目标、参照物、关系三者分开 → 已保存的指代实验：从像素提框后再应用关系 → GLIP：把类别和短语都放到对象级定位任务中”阅读。相关实验以其已保存结果为起点，需要动手时再顺序执行 Notebook；","sha256":"ba9e0813debefcdf122ce5a17bf25680c0faa25ecf1a61ee6459f109e5776998","chapter":{"id":"III","title":"视觉与语言"},"unit":{"id":"grounding-object-contract","title":"一句话怎样落实为正确对象的区域？"},"page":{"id":"glip-language-localization","title":"从透明规则走向对象级学习方案","number":3},"sectionTitle":"从透明规则走向对象级学习方案","currentQuestion":"目标、参照物与关系怎样共同约束定位输出？","sourceSlides":["l3-r2-p63"],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-1d9b0885f2a4a66578c780f7","nextBlock":null,"prerequisites":[{"id":"b-1d9b0885f2a4a66578c780f7","url":"#b-1d9b0885f2a4a66578c780f7","role":"section-title"},{"id":"b-1c989e608aeddcda3cce73c3","url":"#b-1c989e608aeddcda3cce73c3","role":"problem-context"}],"relatedBlocks":[]}]}