{"version":2,"relationshipURLBase":"https://codingai-lec03.pages.dev/course/notebook-blocks/d1a5e426d75643fcb95866b2042b4f18.html","relationshipResolution":"Resolve relative relationship URLs against relationshipURLBase; complete records and absolute URLs are in blocks.","title":"先预测采样结果，再判断指标名称","url":"https://codingai-lec03.pages.dev/course/notebook-blocks/d1a5e426d75643fcb95866b2042b4f18.html","markdown":"https://codingai-lec03.pages.dev/course/notebook-blocks/d1a5e426d75643fcb95866b2042b4f18.md","pages":[{"id":"reading-opening","title":"先预测采样结果，再判断指标名称","unit":"先预测采样结果，再判断指标名称","unitId":"notebook-blocks/d1a5e426d75643fcb95866b2042b4f18","chapter":"course","number":1}],"aliases":[{"oldId":"b-5e8b93706946e5096e38d1aa","targetId":"b-5e8b93706946e5096e38d1aa","page":"reading-opening","operation":"move","beforeSHA256":"f6dfd09a50caab375f02e0abcc3ff7a454ea579ed8eef0c35b95c95b0619c304","afterSHA256":"f6dfd09a50caab375f02e0abcc3ff7a454ea579ed8eef0c35b95c95b0619c304","reason":"Same source body in authored semantic page"},{"oldId":"b-22da985247e24bb81c7bb379","targetId":"b-22da985247e24bb81c7bb379","page":"reading-opening","operation":"move","beforeSHA256":"7b920e8a16cbfec89ba88e9ad2e57250eaacd76fa3677a00f549a7588f32e503","afterSHA256":"7b920e8a16cbfec89ba88e9ad2e57250eaacd76fa3677a00f549a7588f32e503","reason":"Same source body in authored semantic page"},{"oldId":"b-78b61b6be51132ca74128a88","targetId":"b-78b61b6be51132ca74128a88","page":"reading-opening","operation":"move","beforeSHA256":"e91f24e4e4ca40e2099a0d81d30db6568f4bf2074b499eeee6e3959617bd97ac","afterSHA256":"e91f24e4e4ca40e2099a0d81d30db6568f4bf2074b499eeee6e3959617bd97ac","reason":"Same source body in authored semantic page"},{"oldId":"b-cf305a3b1bd13da4bf91bb01","targetId":"b-cf305a3b1bd13da4bf91bb01","page":"reading-opening","operation":"move","beforeSHA256":"6f0b115c3c32032428a4c0d5ec4791e16d6b63aa3d643457bed758bf6af2b765","afterSHA256":"6f0b115c3c32032428a4c0d5ec4791e16d6b63aa3d643457bed758bf6af2b765","reason":"Same source body in authored semantic page"},{"oldId":"b-64b7a81660a98d9e7240c021","targetId":"b-64b7a81660a98d9e7240c021","page":"reading-opening","operation":"move","beforeSHA256":"5e5de37a286901ae99e730dcfd55ef7555bc75708621a2de2271f9ec46a48fc7","afterSHA256":"5e5de37a286901ae99e730dcfd55ef7555bc75708621a2de2271f9ec46a48fc7","reason":"Same source body in authored semantic page"},{"oldId":"b-6c1c8483d45ac7276118fd76","targetId":"b-6c1c8483d45ac7276118fd76","page":"reading-opening","operation":"move","beforeSHA256":"3e79e6ee37bc2651602606740b6349a62a1d4e121af0d4849b58b3c20d43aadf","afterSHA256":"3e79e6ee37bc2651602606740b6349a62a1d4e121af0d4849b58b3c20d43aadf","reason":"Same source body in authored semantic page"},{"oldId":"b-948d065222c99129232d9e0b","targetId":"b-948d065222c99129232d9e0b","page":"reading-opening","operation":"move","beforeSHA256":"e473b366baa2429f354f50c456cb6c5b42b683bd2e19bcd7f9188a205d7e10ce","afterSHA256":"e473b366baa2429f354f50c456cb6c5b42b683bd2e19bcd7f9188a205d7e10ce","reason":"Same source body in authored semantic page"},{"oldId":"b-285e66b31e01105e99b1ad41","targetId":"b-285e66b31e01105e99b1ad41","page":"reading-opening","operation":"move","beforeSHA256":"e18a781a50c0cb125aa3eaa447621068cd99a2f03498a9f29c96561b27e06a4b","afterSHA256":"e18a781a50c0cb125aa3eaa447621068cd99a2f03498a9f29c96561b27e06a4b","reason":"Same source body in authored semantic page"},{"oldId":"block-d1a5e426d75643fcb95866b2042b4f18--先预测采样结果再判断指标名称","targetId":"b-64b7a81660a98d9e7240c021","operation":"retain-source-identity"}],"blocks":[{"id":"b-5e8b93706946e5096e38d1aa","type":"p","scope":"document","url":"https://codingai-lec03.pages.dev/course/notebook-blocks/d1a5e426d75643fcb95866b2042b4f18.html#b-5e8b93706946e5096e38d1aa","content":"[LAB 11 · 从 RGB 到像素：读出、FCN 与边界](/course/notebooks/9e268706471944e3a59e6f568f9ce9a3.html)","sha256":"f6dfd09a50caab375f02e0abcc3ff7a454ea579ed8eef0c35b95c95b0619c304","chapter":{"id":"course","title":"先预测采样结果，再判断指标名称"},"unit":{"id":"notebook-blocks/d1a5e426d75643fcb95866b2042b4f18","title":"先预测采样结果，再判断指标名称"},"page":{"id":"reading-opening","title":"先预测采样结果，再判断指标名称","number":1},"sectionTitle":"先预测采样结果，再判断指标名称","currentQuestion":"先预测采样结果，再判断指标名称","sourceSlides":[],"knowledgeBlock":null,"formulas":[],"previousBlock":null,"nextBlock":"b-22da985247e24bb81c7bb379","prerequisites":[],"relatedBlocks":[]},{"id":"b-22da985247e24bb81c7bb379","type":"h1","scope":"document","url":"https://codingai-lec03.pages.dev/course/notebook-blocks/d1a5e426d75643fcb95866b2042b4f18.html#b-22da985247e24bb81c7bb379","content":"# 先预测采样结果，再判断指标名称","sha256":"7b920e8a16cbfec89ba88e9ad2e57250eaacd76fa3677a00f549a7588f32e503","chapter":{"id":"course","title":"先预测采样结果，再判断指标名称"},"unit":{"id":"notebook-blocks/d1a5e426d75643fcb95866b2042b4f18","title":"先预测采样结果，再判断指标名称"},"page":{"id":"reading-opening","title":"先预测采样结果，再判断指标名称","number":1},"sectionTitle":"先预测采样结果，再判断指标名称","currentQuestion":"先预测采样结果，再判断指标名称","sourceSlides":[],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-5e8b93706946e5096e38d1aa","nextBlock":"b-78b61b6be51132ca74128a88","prerequisites":[{"id":"b-5e8b93706946e5096e38d1aa","url":"#b-5e8b93706946e5096e38d1aa","role":"problem-context"}],"relatedBlocks":[]},{"id":"b-78b61b6be51132ca74128a88","type":"p","scope":"document","url":"https://codingai-lec03.pages.dev/course/notebook-blocks/d1a5e426d75643fcb95866b2042b4f18.html#b-78b61b6be51132ca74128a88","content":"按问题继续","sha256":"e91f24e4e4ca40e2099a0d81d30db6568f4bf2074b499eeee6e3959617bd97ac","chapter":{"id":"course","title":"先预测采样结果，再判断指标名称"},"unit":{"id":"notebook-blocks/d1a5e426d75643fcb95866b2042b4f18","title":"先预测采样结果，再判断指标名称"},"page":{"id":"reading-opening","title":"先预测采样结果，再判断指标名称","number":1},"sectionTitle":"先预测采样结果，再判断指标名称","currentQuestion":"先预测采样结果，再判断指标名称","sourceSlides":[],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-22da985247e24bb81c7bb379","nextBlock":"b-cf305a3b1bd13da4bf91bb01","prerequisites":[{"id":"b-22da985247e24bb81c7bb379","url":"#b-22da985247e24bb81c7bb379","role":"section-title"},{"id":"b-5e8b93706946e5096e38d1aa","url":"#b-5e8b93706946e5096e38d1aa","role":"problem-context"}],"relatedBlocks":[]},{"id":"b-cf305a3b1bd13da4bf91bb01","type":"p","scope":"document","url":"https://codingai-lec03.pages.dev/course/notebook-blocks/d1a5e426d75643fcb95866b2042b4f18.html#b-cf305a3b1bd13da4bf91bb01","content":"前景/背景/mIoU口径、RGB到mask训练、FCN与1×1读出及边界","sha256":"6f0b115c3c32032428a4c0d5ec4791e16d6b63aa3d643457bed758bf6af2b765","chapter":{"id":"course","title":"先预测采样结果，再判断指标名称"},"unit":{"id":"notebook-blocks/d1a5e426d75643fcb95866b2042b4f18","title":"先预测采样结果，再判断指标名称"},"page":{"id":"reading-opening","title":"先预测采样结果，再判断指标名称","number":1},"sectionTitle":"先预测采样结果，再判断指标名称","currentQuestion":"先预测采样结果，再判断指标名称","sourceSlides":[],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-78b61b6be51132ca74128a88","nextBlock":"b-64b7a81660a98d9e7240c021","prerequisites":[{"id":"b-22da985247e24bb81c7bb379","url":"#b-22da985247e24bb81c7bb379","role":"section-title"},{"id":"b-5e8b93706946e5096e38d1aa","url":"#b-5e8b93706946e5096e38d1aa","role":"problem-context"}],"relatedBlocks":[]},{"id":"b-64b7a81660a98d9e7240c021","type":"h2","scope":"document","url":"https://codingai-lec03.pages.dev/course/notebook-blocks/d1a5e426d75643fcb95866b2042b4f18.html#b-64b7a81660a98d9e7240c021","content":"## 先预测采样结果，再判断指标名称","sha256":"5e5de37a286901ae99e730dcfd55ef7555bc75708621a2de2271f9ec46a48fc7","chapter":{"id":"course","title":"先预测采样结果，再判断指标名称"},"unit":{"id":"notebook-blocks/d1a5e426d75643fcb95866b2042b4f18","title":"先预测采样结果，再判断指标名称"},"page":{"id":"reading-opening","title":"先预测采样结果，再判断指标名称","number":1},"sectionTitle":"先预测采样结果，再判断指标名称","currentQuestion":"先预测采样结果，再判断指标名称","sourceSlides":[],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-cf305a3b1bd13da4bf91bb01","nextBlock":"b-6c1c8483d45ac7276118fd76","prerequisites":[{"id":"b-22da985247e24bb81c7bb379","url":"#b-22da985247e24bb81c7bb379","role":"section-title"},{"id":"b-5e8b93706946e5096e38d1aa","url":"#b-5e8b93706946e5096e38d1aa","role":"problem-context"}],"relatedBlocks":[]},{"id":"b-6c1c8483d45ac7276118fd76","type":"p","scope":"document","url":"https://codingai-lec03.pages.dev/course/notebook-blocks/d1a5e426d75643fcb95866b2042b4f18.html#b-6c1c8483d45ac7276118fd76","content":"8×8 真值有 16 个前景、48 个背景。每个 2×2 小块求平均得到 4×4 覆盖率；最近邻扩大回 8×8，再以 ≥.5 二值化，会把半覆盖小块全部判前景，因此预测前景有 24 个像素。前景交集 16、并集 24，$IoU_{fg}=2/3$；背景交集 40、并集 48，$IoU_{bg}=5/6$。两类平均为 .75，8 个像素错，准确率为 .875。","sha256":"3e79e6ee37bc2651602606740b6349a62a1d4e121af0d4849b58b3c20d43aadf","chapter":{"id":"course","title":"先预测采样结果，再判断指标名称"},"unit":{"id":"notebook-blocks/d1a5e426d75643fcb95866b2042b4f18","title":"先预测采样结果，再判断指标名称"},"page":{"id":"reading-opening","title":"先预测采样结果，再判断指标名称","number":1},"sectionTitle":"先预测采样结果，再判断指标名称","currentQuestion":"先预测采样结果，再判断指标名称","sourceSlides":[],"knowledgeBlock":null,"formulas":[{"latex":"IoU_{fg}=2/3","definitionContext":"先预测采样结果，再判断指标名称"},{"latex":"IoU_{bg}=5/6","definitionContext":"先预测采样结果，再判断指标名称"}],"previousBlock":"b-64b7a81660a98d9e7240c021","nextBlock":"b-948d065222c99129232d9e0b","prerequisites":[{"id":"b-64b7a81660a98d9e7240c021","url":"#b-64b7a81660a98d9e7240c021","role":"section-title"},{"id":"b-5e8b93706946e5096e38d1aa","url":"#b-5e8b93706946e5096e38d1aa","role":"problem-context"}],"relatedBlocks":[]},{"id":"b-948d065222c99129232d9e0b","type":"p","scope":"document","url":"https://codingai-lec03.pages.dev/course/notebook-blocks/d1a5e426d75643fcb95866b2042b4f18.html#b-948d065222c99129232d9e0b","content":"双线性插值在四个邻近覆盖率之间加权，`align_corners=False` 明确规定坐标对应。它不凭空恢复原边界。本例阈值仍为 .5：双线性只有 6 个错误像素，像素准确率 .90625，却仍然得到前景 IoU=2/3；误差位置、前景集合大小变化会使不同指标给出不同读数。它的背景 IoU=.88462，两类 mIoU=.77564。不能把“更平滑”直接改写成“知道了真实轮廓”。","sha256":"e473b366baa2429f354f50c456cb6c5b42b683bd2e19bcd7f9188a205d7e10ce","chapter":{"id":"course","title":"先预测采样结果，再判断指标名称"},"unit":{"id":"notebook-blocks/d1a5e426d75643fcb95866b2042b4f18","title":"先预测采样结果，再判断指标名称"},"page":{"id":"reading-opening","title":"先预测采样结果，再判断指标名称","number":1},"sectionTitle":"先预测采样结果，再判断指标名称","currentQuestion":"先预测采样结果，再判断指标名称","sourceSlides":[],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-6c1c8483d45ac7276118fd76","nextBlock":"b-285e66b31e01105e99b1ad41","prerequisites":[{"id":"b-64b7a81660a98d9e7240c021","url":"#b-64b7a81660a98d9e7240c021","role":"section-title"},{"id":"b-5e8b93706946e5096e38d1aa","url":"#b-5e8b93706946e5096e38d1aa","role":"problem-context"}],"relatedBlocks":[]},{"id":"b-285e66b31e01105e99b1ad41","type":"p","scope":"document","url":"https://codingai-lec03.pages.dev/course/notebook-blocks/d1a5e426d75643fcb95866b2042b4f18.html#b-285e66b31e01105e99b1ad41","content":"对 RGB 或 logits 的双线性插值可以混合实值。对类别 mask，编号 1、2 不是数轴上可平均的强度：几何变换应采用最近邻保持离散标签；ignore=255 也必须保持身份。下面的训练增强只水平翻转，对图像和 mask 执行同一个随机选择。","sha256":"e18a781a50c0cb125aa3eaa447621068cd99a2f03498a9f29c96561b27e06a4b","chapter":{"id":"course","title":"先预测采样结果，再判断指标名称"},"unit":{"id":"notebook-blocks/d1a5e426d75643fcb95866b2042b4f18","title":"先预测采样结果，再判断指标名称"},"page":{"id":"reading-opening","title":"先预测采样结果，再判断指标名称","number":1},"sectionTitle":"先预测采样结果，再判断指标名称","currentQuestion":"先预测采样结果，再判断指标名称","sourceSlides":[],"knowledgeBlock":null,"formulas":[],"previousBlock":"b-948d065222c99129232d9e0b","nextBlock":null,"prerequisites":[{"id":"b-64b7a81660a98d9e7240c021","url":"#b-64b7a81660a98d9e7240c021","role":"section-title"},{"id":"b-5e8b93706946e5096e38d1aa","url":"#b-5e8b93706946e5096e38d1aa","role":"problem-context"}],"relatedBlocks":[]}]}