{
 "nbformat": 4,
 "nbformat_minor": 5,
 "metadata": {
  "kernelspec": {
   "name": "python3",
   "display_name": "Python 3",
   "language": "python"
  },
  "language_info": {
   "name": "python"
  },
  "deepnote": {
   "projectId": "c0b185e7-d7e3-48b6-a605-8f1acdfe243c",
   "notebookId": "893bd96837a6421d8fa5c5788075bd5b",
   "name": "CLS 02 · 传统视觉：局部描述、共同词典与分类"
  },
  "title": "CLS 02 · 传统视觉：局部描述、共同词典与分类"
 },
 "cells": [
  {
   "cell_type": "markdown",
   "id": "8897ee5a11334abbad3042bb05947bb1",
   "metadata": {
    "deepnote_block_type": "text-cell-h1",
    "deepnote_block_id": "8897ee5a11334abbad3042bb05947bb1",
    "html_anchor": "cell-1"
   },
   "source": "# CLS 02 · 传统视觉：局部描述、共同词典与分类"
  },
  {
   "cell_type": "markdown",
   "id": "d9f71cc9cba3432a81dce05fd21939b0",
   "metadata": {
    "deepnote_block_type": "markdown",
    "deepnote_block_id": "d9f71cc9cba3432a81dce05fd21939b0",
    "html_anchor": "cell-2",
    "deepnote": {
     "block_id": "d9f71cc9cba3432a81dce05fd21939b0"
    }
   },
   "source": [
    "[打开传统视觉参考](https://codingai-lec02.pages.dev/course/reference.html#history) · [第二讲导读](https://codingai-lec02.pages.dev/course/guide.html) · [问卷作业 ↗](https://tally.so/r/VLroVv)（外链可右键在新标签页打开）\n",
    "\n",
    "本册是第二讲的**按需历史参考**。它解释端到端 CNN 之前和同期的一条路线：先组织局部测量，再形成共同坐标，最后用标签训练分类读出。它不插入 01→03→04→06→07 的主讲阅读链。\n",
    "\n",
    "想比较“手工表示”和 CNN，先读第 3–5 节的共同词典、空间汇总与分类反馈；想把一个描述子的维数算清，再读第 1–2 节。图中 SIFT、HOG、词袋和二维分类各承担一个接口，不把它们拼成同一套已训练系统。\n",
    "\n",
    "**动手入口**：[LAB 07 · 空间格、方向桶与传统表示](https://codingai-lec02.pages.dev/course/lab07.html)。固定局部描述与词编号，只交换空间位置，观察全局计数与分区计数分别变不变；本册末尾代码可先独立核对地址与向量。"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8a233e0f4a234fda95486be6c6dbc997",
   "metadata": {
    "deepnote_block_type": "text-cell-h2",
    "deepnote_block_id": "8a233e0f4a234fda95486be6c6dbc997",
    "html_anchor": "cell-3"
   },
   "source": [
    "## 1. 从局部参照到 128 维：SIFT 怎样让局部可比较？"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6c18c0c2399c40e98d450fd8e8b0ac6a",
   "metadata": {
    "deepnote_block_type": "markdown",
    "deepnote_block_id": "6c18c0c2399c40e98d450fd8e8b0ac6a",
    "html_anchor": "cell-4",
    "deepnote": {
     "block_id": "6c18c0c2399c40e98d450fd8e8b0ac6a"
    }
   },
   "source": [
    "**课件与图解**：[分类器之前，先设计一套更合适的图像表示](https://codingai-lec02.pages.dev/course/reference/feature-pipeline-before-classifier.html) · [相机图里，SIFT把哪些局部选作比较起点？](https://codingai-lec02.pages.dev/course/reference/sift-scale-and-orientation.html) · [打开一个局部，看SIFT怎样记录方向与粗位置](https://codingai-lec02.pages.dev/course/reference/sift-local-descriptor.html) · [传统视觉参考总入口](https://codingai-lec02.pages.dev/course/reference.html#history)\n",
    "\n",
    "**先抓住两个地址：方向桶与空间格。** 如果两个地方都出现 45° 梯度，描述子要不要把它们当作同一项？\n",
    "\n",
    "![SIFT教学格：r2c3四个明确方向箭头形成直方图，45度桶位于地址49；相邻r2c4同方向桶位于地址57](data:image/webp;base64,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)\n",
    "\n",
    "**跟着一个数走。** 方向桶算例 的独立机制格有 4 行×4 列，每格 8 个方向桶，共 128 维。r2c3 内四个梯度的方向是 0°、45°、45°、90°，强度是 1、2、1、2；把同方向强度相加，得到 **[1,3,2,0,0,0,0,0]**。图中箭头等长，尖端给方向，强度读数值；梯度方向不是轮廓切线。\n",
    "\n",
    "行列 r、c 从 1 开始，桶 b 从 0 开始。零基拼接地址为 j=((r−1)×4+c−1)×8+b。于是：\n",
    "\n",
    "| 空间格 | 方向 | 桶 b | 拼接地址 j | 该项的值 |\n",
    "|---|---|---:|---:|---:|\n",
    "| r2c3 | 45° | 1 | 49 | 3 |\n",
    "| r2c4 | 45° | 1 | 57 | 1 |\n",
    "\n",
    "**方向相同，空间格不同，向量地址也不同。** 相邻格 r2c4 的完整计票是 [0,1,0,0,0,2,0,0]。这个离散格只展开地址与计票；真实 SIFT 还含空间/方向插值、加权、截断和归一化。\n",
    "\n",
    "**把机制放回相机图。** 关键点参照页 为真实像素提取的位置、尺度和主方向建立比较参照，保存记录保留 700 个关键点，图上仅显示子集。局部对应页 对相机图旋转 24°、缩至 0.78 倍后重新提取和匹配：381 对候选中，349 对与已知变换的几何误差不超过 3 像素，主讲显示其中 14 对。描述参照提高可比性，不保证每次匹配成功；349/381 也不是分类准确率。\n",
    "\n",
    "**先预测与核对**：遮住表中地址，算出 49 和 57；把 r2c3 的一票搬到相邻格，说明哪里减、哪里增。下一节保持方向统计，却改用固定窗口和规则网格。"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "70d8c10e08e94d9c8eb1a8b534ebdaff",
   "metadata": {
    "deepnote_block_type": "text-cell-h2",
    "deepnote_block_id": "70d8c10e08e94d9c8eb1a8b534ebdaff",
    "html_anchor": "cell-5"
   },
   "source": [
    "## 2. HOG：固定窗口保留怎样的粗位置？"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c14e2f597eec4bd4a6da9338c4d915e3",
   "metadata": {
    "deepnote_block_type": "markdown",
    "deepnote_block_id": "c14e2f597eec4bd4a6da9338c4d915e3",
    "html_anchor": "cell-6",
    "deepnote": {
     "block_id": "c14e2f597eec4bd4a6da9338c4d915e3"
    }
   },
   "source": [
    "**课件与图解**：[HOG：固定窗口里的方向统计，保留规则网格](https://codingai-lec02.pages.dev/course/reference/hog-local-shape.html)\n",
    "\n",
    "HOG 以固定窗口与规则网格组织方向信息，和每个 SIFT 关键点自己的尺度/方向参照不同。本例窗口宽 64、高 128 像素；每个 cell 为 8×8，所以是 16 行×8 列，不能把宽高和行列倒置。每个 cell 统计 9 个无符号方向桶，相邻 2×2 cells 组成 36 维 block，再归一化、拼接。\n",
    "\n",
    "block 每次移动 1 个 cell，因此有 15 行×7 列=105 个 block，完整描述为 105×36=3780 维。HOG 可以直接交给分类器，不要求先经过视觉词袋。局部汇总保留 cell 地址，却不保存窗口内每个像素的全部细节。\n",
    "\n",
    "可核对的像素来自程序 cell(row=4,col=3)，即第 5 行第 4 列。其无符号方向约 170.6398°、幅值 0.3616783，线性分给 160° 桶 0.1692689 和循环相邻的 0° 桶 0.1924094，两份之和仍等于幅值。本例使用透明教学变体：cell 硬分配、方向插值、block L2 归一化，未声称逐位复现完整 Dalal–Triggs 检测器。输入行人是生成情境图，梯度由实际像素计算，未训练行人分类器。\n",
    "\n",
    "**先预测与核对**：先只给出宽高和 cell 大小，算出行列、block 数与总维数，再核对两份方向票之和。下一问题是：一张图局部数量不固定时，怎样给分类器固定长度的输入？"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "af7ed15c4a6d4954a75e86da24e9c119",
   "metadata": {
    "deepnote_block_type": "text-cell-h2",
    "deepnote_block_id": "af7ed15c4a6d4954a75e86da24e9c119",
    "html_anchor": "cell-7"
   },
   "source": [
    "## 3. 单图六词计数以后，为什么还需要共同词典？"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "308eeda4b8b24cd599ff64368777ada0",
   "metadata": {
    "deepnote_block_type": "markdown",
    "deepnote_block_id": "308eeda4b8b24cd599ff64368777ada0",
    "html_anchor": "cell-8",
    "deepnote": {
     "block_id": "308eeda4b8b24cd599ff64368777ada0"
    }
   },
   "source": [
    "**课件与图解**：[600个局部描述，最后落进这6个计数](https://codingai-lec02.pages.dev/course/reference/visual-codebook-six-patches.html) · [词典学习一次，新图像沿用同一套词编号](https://codingai-lec02.pages.dev/course/reference/codebook-training-versus-assignment.html)\n",
    "\n",
    "街景算例实际提取 600 个 SIFT 描述，做单图聚类演示，用 K=6 个中心分配描述，得到词频 [91,65,78,152,113,101]，合计 600。中心是描述子空间中的数值向量，代表裁片只是靠近中心的观测，不是中心生成的物体。词编号不是“车轮”“建筑”等语义标签。\n",
    "\n",
    "这个单图结果说明数量不定的局部怎样汇总为六维计数；要训练跨图像的分类器，还必须保证每一维含义一致。正式流程从训练图像收集局部描述，学习并保存共同中心和编号；新图仅向已保存中心分配，再按相同规则计数。若每张新图各自重新聚类，“第 1 维”可能代表另一种模式，旧分类器无法直接沿用。词典本身可以由数据学习，但本讲分阶段方案中的分类训练不会继续改变已经冻结的词典。\n",
    "\n",
    "**先预测与核对**：先相加核对 600，再把两张图的“词 1”和“词 2”互换，说明为何向量长度没变但坐标意义变了。下一节固定词典与描述集合，只改变位置，检查全局计数遗漏了什么。"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0cd36916667e4e2395d3ccee5fbc10c6",
   "metadata": {
    "deepnote_block_type": "text-cell-h2",
    "deepnote_block_id": "0cd36916667e4e2395d3ccee5fbc10c6",
    "html_anchor": "cell-9"
   },
   "source": [
    "## 4. 固定同一批描述，只补回上下地址"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ee601aaf07f242b2b461b2ca1be2e963",
   "metadata": {
    "deepnote_block_type": "markdown",
    "deepnote_block_id": "ee601aaf07f242b2b461b2ca1be2e963",
    "html_anchor": "cell-10",
    "deepnote": {
     "block_id": "ee601aaf07f242b2b461b2ca1be2e963"
    }
   },
   "source": [
    "**课件与图解**：[传统视觉参考总入口](https://codingai-lec02.pages.dev/course/reference.html#history)\n",
    "\n",
    "**先预测**：下面只交换上下两排，A、B、C 的总数会变吗？能不能由总数恢复原来的布局？\n",
    "\n",
    "![同样16个局部描述交换上下位置，全局计数仍为6,6,4，按上半下半拼接的六维向量发生改变](data:image/webp;base64,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)\n",
    "\n",
    "两图使用同样的 16 个已提取描述、同样的词典归属；这里只改变位置，没有把拼接图重新送入 SIFT。\n",
    "\n",
    "| 布局 | 上半 (A,B,C) | 下半 (A,B,C) | 全局 (A,B,C) |\n",
    "|---|---|---|---|\n",
    "| 甲 | (5,2,1) | (1,4,3) | (6,6,4) |\n",
    "| 乙 | (1,4,3) | (5,2,1) | (6,6,4) |\n",
    "\n",
    "**全局汇总丢掉了哪里，区域汇总就补回哪里。** 按“上半后下半”的固定顺序拼接，甲变成 (5,2,1,1,4,3)，乙变成 (1,4,3,5,2,1)。增加的是粗区域地址，仍没有保存每个描述的精确位置。\n",
    "\n",
    "**先预测与核对**：先用手遮住下方向量，分别计数，再运行末尾代码核对。这个两区域构造只检验汇聚如何丢失与补回位置，不是完整空间金字塔算法，也不保证任意图像重排后重新提取的特征都不变。接下来固定这些坐标含义，看看标签怎样选择分类边界。"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "32b451f7dde84d4ba0d31f5de5f8e318",
   "metadata": {
    "deepnote_block_type": "text-cell-h2",
    "deepnote_block_id": "32b451f7dde84d4ba0d31f5de5f8e318",
    "html_anchor": "cell-11"
   },
   "source": [
    "## 5. 学会边界之后，新点沿用什么？"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "88b25e38c47b4572b45023e7983bb549",
   "metadata": {
    "deepnote_block_type": "markdown",
    "deepnote_block_id": "88b25e38c47b4572b45023e7983bb549",
    "html_anchor": "cell-12",
    "deepnote": {
     "block_id": "88b25e38c47b4572b45023e7983bb549"
    }
   },
   "source": [
    "**课件与图解**：[表示交给分类器之后，边界怎样学出来？](https://codingai-lec02.pages.dev/course/reference/svm-on-designed-features.html) · [传统视觉参考总入口](https://codingai-lec02.pages.dev/course/reference.html#history) · [沿着总图，解释训练与新图预测](https://codingai-lec02.pages.dev/course/reference/traditional-pipeline-recap.html) · [分类错误，能改到哪些测量？](https://codingai-lec02.pages.dev/#/learn-the-local-features/0)\n",
    "\n",
    "为单独看清分类接口，主讲使用八个手工指定的二维训练点与标签，并实际求解最大间隔分类器。这里的坐标不是从相机图、街景或轮子检测器量出来的。训练得到 w=(1,−1)、b=0，即 s(x)=x₁−x₂，边界为 x₁=x₂。正负标签与训练目标共同约束参数；规则一旦固定，新点只做前向计算。\n",
    "\n",
    "| 未参与训练的点 | 坐标 | 分数 | 本例显示 |\n",
    "|---|---|---:|---|\n",
    "| Q1 | (3.5,1.5) | +2 | A |\n",
    "| Q2 | (1.5,3.5) | −2 | B |\n",
    "| Q3 | (2.5,2.5) | 0 | ? |\n",
    "\n",
    "Q3 恰在边界，本例保留问号；分数 0 不是两类各 50% 的概率。三个查询没有真实测试标签，能够计算预测，却不能据此报告测试准确率。\n",
    "\n",
    "回到传统总流程：训练图像提供词典学习的描述，标签训练分类器；应用时固定词典编号、统计方法和分类器权重。区域统计决定哪些粗位置信息被保留。各前端可能曾独立学习，但在这里的分类训练中已经冻结，最终分类损失主要改变读出。\n",
    "\n",
    "**先预测与核对**：手算三个新点，再沿总图各说一句“训练改什么”“新图沿用什么”“上下交换在哪一步可见”。**核对**：最后的问题不是“传统方法有没有学习”，而是“当前类别错误能反馈到哪里”。共同词典已经从无类别标签的描述子中学习，分类器也从标签学习。新的问题是：分类错误能否一路反馈到提取图像信息的前端？端到端 CNN 保留人工选择的局部与共享结构，同时让类别损失调整前面的核权重。"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7f6966ffb8cf45688f69e876ee9907cf",
   "metadata": {
    "deepnote_block_type": "text-cell-h2",
    "deepnote_block_id": "7f6966ffb8cf45688f69e876ee9907cf",
    "html_anchor": "cell-13"
   },
   "source": [
    "## 把本段主线连起来"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6c6385d02de54caebb74343e631c586f",
   "metadata": {
    "deepnote_block_type": "markdown",
    "deepnote_block_id": "6c6385d02de54caebb74343e631c586f",
    "html_anchor": "cell-14",
    "deepnote": {
     "block_id": "6c6385d02de54caebb74343e631c586f"
    }
   },
   "source": [
    "**课件与图解**：[本段起点](https://codingai-lec02.pages.dev/course/reference/feature-pipeline-before-classifier.html)；逐页入口见上方各节。\n",
    "\n",
    "用同一条链复述：局部参照使描述更可比较；共同词典使不同图像的坐标含义一致；区域地址补回粗位置；带标签的训练选择分类边界。然后指出：本讲分阶段方案中，最终分类损失没有继续修改已冻结的前端。\n",
    "\n",
    "相机对应、HOG、单图六词、固定描述重排与二维 SVM 各有自己的证据范围。不要把 349 对几何一致匹配、600 个词分配或三个查询分数写成同一个系统的图像分类性能。更多可复查操作见 [LAB 07 · 空间格、方向桶与传统表示](https://codingai-lec02.pages.dev/course/lab07.html)。\n",
    "\n",
    "来源沿用主讲：[Lowe · SIFT](https://www.cs.ubc.ca/~lowe/papers/ijcv04.pdf)、[Dalal 与 Triggs · HOG](https://lear.inrialpes.fr/people/triggs/pubs/Dalal-cvpr05.pdf)、[Sivic 与 Zisserman · 视觉词](https://www.robots.ox.ac.uk/~vgg/publications/2003/Sivic03/sivic03.pdf)、[Lazebnik 等 · 空间组织](https://slazebni.cs.illinois.edu/publications/cvpr06b.pdf)。"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "318dbdcb8b7b4d58ab1d8e9212dfc09f",
   "metadata": {
    "deepnote_block_type": "text-cell-h2",
    "deepnote_block_id": "318dbdcb8b7b4d58ab1d8e9212dfc09f",
    "html_anchor": "cell-15"
   },
   "source": [
    "## 课堂核对：先预测，再运行"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "83761352409143c889b71d4881f958a9",
   "metadata": {
    "deepnote_block_type": "markdown",
    "deepnote_block_id": "83761352409143c889b71d4881f958a9",
    "html_anchor": "cell-16",
    "deepnote": {
     "block_id": "83761352409143c889b71d4881f958a9"
    }
   },
   "source": [
    "**课件与图解**：[本段起点](https://codingai-lec02.pages.dev/course/reference/sift-local-descriptor.html)；逐页入口见上方各节。\n",
    "\n",
    "下面只核对构造的描述子地址、固定描述集合的汇总与既有二维分类规则。代码不提取新图像特征，不训练新的分类器。先写出索引 49 的来源、区域交换后的两个向量，以及 Q3 的显示。"
   ]
  },
  {
   "cell_type": "code",
   "id": "d5dbcfa5a53f41e3a7af112f97ad334d",
   "metadata": {
    "deepnote_block_type": "code",
    "deepnote_block_id": "d5dbcfa5a53f41e3a7af112f97ad334d",
    "html_anchor": "cell-17"
   },
   "source": [
    "# 教学构造；r/c一基，方向桶b和向量地址零基。\n",
    "bins = [0]*8\n",
    "for direction, magnitude in [(0, 1), (45, 2), (45, 1), (90, 2)]:\n",
    "    bins[direction//45] += magnitude\n",
    "r, c, b = 2, 3, 1\n",
    "j = ((r-1)*4+c-1)*8+b\n",
    "assert bins == [1, 3, 2, 0, 0, 0, 0, 0] and j == 49\n",
    "assert ((2-1)*4+4-1)*8+1 == 57\n",
    "print('r2c3方向计票:', bins, '45度地址:', j, '值:', bins[b])\n",
    "\n",
    "word_counts = [91, 65, 78, 152, 113, 101]\n",
    "assert sum(word_counts) == 600\n",
    "upper, lower = (5, 2, 1), (1, 4, 3)\n",
    "global_a = tuple(a+b for a, b in zip(upper, lower))\n",
    "global_b = tuple(a+b for a, b in zip(lower, upper))\n",
    "regional_a, regional_b = upper+lower, lower+upper\n",
    "assert global_a == global_b == (6, 6, 4)\n",
    "assert regional_a != regional_b\n",
    "print('全局:', global_a, global_b)\n",
    "print('区域:', regional_a, regional_b)\n",
    "\n",
    "queries = [(3.5, 1.5), (1.5, 3.5), (2.5, 2.5)]\n",
    "scores = [x1-x2 for x1, x2 in queries]\n",
    "labels = ['?' if abs(s)<1e-8 else ('A' if s>0 else 'B') for s in scores]\n",
    "assert scores == [2, -2, 0] and labels == ['A', 'B', '?']\n",
    "print('新点分数:', scores, '本例显示:', labels)\n",
    "print('没有查询真标签，因此这里不计算准确率。')"
   ],
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "markdown",
   "id": "f11218b24f65489383bd11a71d2323ef",
   "metadata": {
    "deepnote_block_type": "markdown",
    "deepnote_block_id": "f11218b24f65489383bd11a71d2323ef",
    "html_anchor": "cell-18",
    "deepnote": {
     "block_id": "f11218b24f65489383bd11a71d2323ef"
    }
   },
   "source": [
    "**课件与图解**：[本段起点](https://codingai-lec02.pages.dev/course/reference/codebook-training-versus-assignment.html)；逐页入口见上方各节。\n",
    "\n",
    "**运行后修改一项**：保持词编号与描述集合，只交换上下区域；再对照“全局不变、区域改变”的预测。若想讨论重新提取图像后的结果，必须到 [LAB 07 · 空间格、方向桶与传统表示](https://codingai-lec02.pages.dev/course/lab07.html) 明确另设实验，不能沿用当前固定描述的前提。离开本册时，用一句话说明 CNN 要把哪条反馈路径继续向前延伸。"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9a4dc3e86e504d7fbc72e22226b30698",
   "metadata": {
    "deepnote_block_type": "text-cell-h2",
    "deepnote_block_id": "9a4dc3e86e504d7fbc72e22226b30698",
    "html_anchor": "cell-19"
   },
   "source": [
    "## 课程导航"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a503d86b4af142cd82cfc5cba34fdbc9",
   "metadata": {
    "deepnote_block_type": "markdown",
    "deepnote_block_id": "a503d86b4af142cd82cfc5cba34fdbc9",
    "html_anchor": "cell-20",
    "deepnote": {
     "block_id": "a503d86b4af142cd82cfc5cba34fdbc9"
    }
   },
   "source": [
    "[第二讲导读与完整索引](https://codingai-lec02.pages.dev/course/guide.html) · [回到 CLS 03 · 端到端 CNN](https://codingai-lec02.pages.dev/course/c03.html)\n",
    "\n",
    "[在 Deepnote 返回 CLS 00 · 第二讲导读](https://deepnote.com/workspace/ise3309jingao-5ba66b0a-4283-4f5a-ae06-3f68d3201e99/project/c0b185e7-d7e3-48b6-a605-8f1acdfe243c/notebook/008c71041d2d43569af77c9f25509be8?utm_source=openai&utm_medium=mcp&utm_campaign=openaimcp&utm_content=008c71041d2d43569af77c9f25509be8&utm_term=get_notebook)。静态课件链接返回对应主题；需保留原 Slides 停点时，请在新标签页打开材料。"
   ]
  }
 ]
}