{
 "nbformat": 4,
 "nbformat_minor": 5,
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "name": "python"
  },
  "lecture_migration": {
   "project_id": "c0b185e7-d7e3-48b6-a605-8f1acdfe243c",
   "notebook_id": "8c459221aa7c4c3490731aa3291807ec",
   "destination": "L2",
   "cloud_run": false
  },
  "title": "LAB 07 · 空间格、方向桶与传统表示"
 },
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "deepnote": {
     "block_id": "68dd42c5d6504caabacb71b61a882059",
     "type": "text-cell-h1",
     "source_block_id": null,
     "logical_key": null
    },
    "html_anchor": "cell-1"
   },
   "source": "# LAB 07 · 空间格、方向桶与传统表示",
   "id": "68dd42c5d6504caabacb71b61a882059"
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deepnote": {
     "block_id": "075ea745b1344ad9bc945d4756f6d7ab",
     "type": "markdown",
     "source_block_id": null,
     "logical_key": null
    },
    "html_anchor": "cell-2"
   },
   "source": [
    "[返回第二讲 · 传统表示的构造核对](https://codingai-lec02.pages.dev/course/reference.html#history) · [第二讲导读与学习材料](https://codingai-lec02.pages.dev/course/guide.html) · [问卷作业 ↗](https://tally.so/r/VLroVv)（右键在新标签页打开）。\n",
    "\n",
    "**传统表示按需练习：先问哪个信息被保留，再看哪个信息消失。** 这里的方向、视觉词与布局都是明确构造，不读取照片、不下载数据；需要 NumPy、Matplotlib。\n",
    "\n",
    "| 选一项小问题 | 只改一项 | 最小运行顺序 |\n",
    "| --- | --- | --- |\n",
    "| 方向相同，为什么地址不同 | `neighbor_cell: [2,4]→[3,3]` | 准备 → 第1节投票 → 第2节地址 |\n",
    "| 亮度幅度改变后，归一化保留什么 | `hog_gain: 1→2` | 准备 → 第3节HOG |\n",
    "| 全图计数看不到的布局，分区能看到吗 | `pyramid_grid: [2,2]→[1,2]` | 准备 → 第4节共同词典/布局 → 第5节空间计数 |\n",
    "| 不装额外库，只查词袋与位置 | 保持词编号，修改末尾 `tokens` 的位置 | 文末独立标准库代码块 |\n",
    "\n",
    "**推荐选择空间计数。** 两张图交换上下半部，全局 A/B/C 数量相同。先预测只分左／右能否看到变化，再运行第4–5节，比较向量维数和平方距离；接着改为 `[2,1]` 单独检验上下分区。\n",
    "\n",
    "本册是第二讲参考区材料。无标签词典的实际学习继续读 [第三讲学习信号](https://codingai-lec03.pages.dev/course/c02-signals.html)；本页的预设分组均值不改称为已训练的 K-means。每次改 `P` 先重跑准备与对应依赖块；无需从头跑全部章节。"
   ],
   "id": "075ea745b1344ad9bc945d4756f6d7ab"
  },
  {
   "cell_type": "code",
   "metadata": {
    "deepnote": {
     "block_id": "1409e1704bc0444b9d906796ad5eac87",
     "type": "code",
     "source_block_id": null,
     "logical_key": null
    },
    "html_anchor": "cell-3"
   },
   "source": "import json\nimport numpy as np\nimport matplotlib.pyplot as plt\nnp.set_printoptions(precision=4, suppress=True)\nplt.rcParams.update({'figure.dpi':110,'font.size':10,'axes.spines.top':False,'axes.spines.right':False})\nBLUE, GOLD, GREEN, PURPLE = '#002676','#FDB515','#00856A','#69458D'\nCHECKS={}; RESULTS={}\ndef check(name,condition):\n    CHECKS[name]=bool(condition)\n    assert CHECKS[name], name\ndef table(headers,rows,title):\n    fig,ax=plt.subplots(figsize=(max(7,len(headers)*1.5),1+len(rows)*.36))\n    ax.axis('off'); ax.set_title(title,pad=12,color=BLUE)\n    tab=ax.table(cellText=rows,colLabels=headers,loc='center',cellLoc='center')\n    tab.auto_set_font_size(False);tab.set_fontsize(10);tab.scale(1,1.45)\n    for (r,c),cell in tab.get_celld().items():\n        cell.set_edgecolor('#CBD5DF')\n        if r==0:cell.set_facecolor('#E9F0F7')\n    fig.tight_layout();plt.show()\n\nP=dict(sift_rotation=0.0, dominant_orientation=0.0,\n       sift_magnitudes=[1.0,2.0,1.0,2.0], neighbor_cell=[2,4],\n       hog_angles=[10.0,190.0,90.0],hog_magnitudes=[2.0,2.0,1.0],\n       hog_offset=0.0,hog_gain=1.0,normalization_epsilon=1e-6,\n       pyramid_grid=[2,2])\nP.update(globals().get('USER_OVERRIDES',{}))\nprint('Teaching configuration:',json.dumps(P,ensure_ascii=False))",
   "id": "1409e1704bc0444b9d906796ad5eac87",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deepnote": {
     "block_id": "479d9b7551514d628ba3a263e23efc43",
     "type": "text-cell-h2",
     "source_block_id": null,
     "logical_key": null
    },
    "html_anchor": "cell-4"
   },
   "source": "## 1 · r2c3的一格：箭头如何进入八个方向桶",
   "id": "479d9b7551514d628ba3a263e23efc43"
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deepnote": {
     "block_id": "d419e65ce6bb4d66836c2354ea26080e",
     "type": "markdown",
     "source_block_id": null,
     "logical_key": null
    },
    "html_anchor": "cell-5"
   },
   "source": "**先读箭头，再把票送入桶。** r2c3 中有四个构造梯度；方向为 0°、45°、45°、90°，强度为 1、2、1、2。\n\n![四个有明确箭头的梯度对应八个有向方向桶](data:image/webp;base64,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)\n\n*默认参考图：箭头等长，尖端表示方向，m 表示强度；x 向右、y 向上。右侧按强度投票，45° 桶得到 2+1=3。教学构造，由本页代码本地计算，并非真实照片的 SIFT 输出。*\n\n**再把这一格放回整张描述子。** 同样是 45°，r2c3 的票写入索引 49，r2c4 的票写入 57；方向桶相同，空间格使地址不同。\n\n![空间格、方向桶与展开索引的对应](data:image/webp;base64,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)\n\n*下一节的默认地址构造：行列从 1 编号，方向桶和展开索引从 0 编号。每格 8 个桶，4×4 格共 128 维；其余格置零只为看清地址。下一节会让你亲手移动邻格。*\n\n**只改一次：** 令 `sift_rotation=45`，保持 `dominant_orientation=0`，先预测哪几根柱向哪里移动。第二次再将主方向也改为 45，观察相对方向是否恢复。相对方向=输入方向−主方向。\n\n本格使用 8 个中心（0°、45°…315°）的最近中心硬投票；`sift_magnitudes` 可改强度。完整 SIFT 的插值、加权与归一化在这里暂时省略，先看清一格内部的对应关系。",
   "id": "d419e65ce6bb4d66836c2354ea26080e"
  },
  {
   "cell_type": "code",
   "metadata": {
    "deepnote": {
     "block_id": "408fbc70a1ce4cbabd64b85cd32af345",
     "type": "code",
     "source_block_id": null,
     "logical_key": null
    },
    "html_anchor": "cell-6"
   },
   "source": "centers=np.arange(8)*45.\nangles=(np.array([0.,45.,45.,90.])+float(P['sift_rotation'])-float(P['dominant_orientation']))%360\nmags=np.asarray(P['sift_magnitudes'],dtype=float);assert mags.shape==(4,) and np.all(mags>=0)\ndef hard_signed_vote(theta,magnitude):\n    ids=np.floor((np.asarray(theta)%360)/45.+.5).astype(int)%8\n    return np.bincount(ids,weights=magnitude,minlength=8)\nsift_hist=hard_signed_vote(angles,mags)\ncheck('SIFT_vote_mass',np.isclose(sift_hist.sum(),mags.sum()))\nif np.array_equal(angles,[0,45,45,90]) and np.array_equal(mags,[1,2,1,2]):check('SIFT_slide_r2c3',np.array_equal(sift_hist,[1,3,2,0,0,0,0,0]))\nfig,axes=plt.subplots(1,2,figsize=(10,4));origins=np.array([[.3,.35],[1.45,.35],[.3,1.45],[1.45,1.45]])\nfor (x,y),a,m in zip(origins,angles,mags):\n    rad=np.deg2rad(a);axes[0].annotate('',xy=(x+.65*np.cos(rad),y+.65*np.sin(rad)),xytext=(x,y),arrowprops=dict(arrowstyle='-|>',color=BLUE,lw=2))\n    axes[0].text(x-.1,y-.25,f'{a:g} deg; m={m:g}',fontsize=9)\naxes[0].set(xlim=(-.2,2.4),ylim=(-.2,2.4),aspect='equal',title='Constructed gradients in cell r2c3',xlabel='x (right)',ylabel='y (up)');axes[0].grid(alpha=.2)\naxes[1].bar(centers,sift_hist,width=30,color=BLUE);axes[1].set(xticks=centers,xlabel='Relative direction (degrees)',ylabel='Sum of magnitudes',title='Eight signed direction bins')\nfig.tight_layout();plt.show()\ntable(['bin b (zero-based)','direction','vote'],[[i,f'{a:g} deg',f'{v:g}'] for i,(a,v) in enumerate(zip(centers,sift_hist))],'Local direction histogram')\nRESULTS['sift_vote']={'relative_angles':angles.tolist(),'histogram':sift_hist.tolist()}",
   "id": "408fbc70a1ce4cbabd64b85cd32af345",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deepnote": {
     "block_id": "f68a18d6d10840e986a7f475d2d5733e",
     "type": "markdown",
     "source_block_id": null,
     "logical_key": null
    },
    "html_anchor": "cell-7"
   },
   "source": [
    "**运行后看这里。** 先检查8个桶的总票数是否等于输入强度总和，再看旋转后哪几个桶改变。数箭头与按强度投票不是同一个操作。"
   ],
   "id": "f68a18d6d10840e986a7f475d2d5733e"
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deepnote": {
     "block_id": "9520598e75ed4381b89998445ded04ba",
     "type": "markdown",
     "source_block_id": null,
     "logical_key": null
    },
    "html_anchor": "cell-8"
   },
   "source": "**观察与边界：** 默认结果是`[1,3,2,0,0,0,0,0]`。强度参与加权，而不是简单数箭头。这个离散构造省略了真实SIFT的空间/方向插值、Gaussian加权、截断和归一化，也没有从图像检测主方向；不能据此声称复现完整SIFT或自动旋转不变。\n\n**记录：** 预测票____；修改____；观察到票怎样移动____；若同步改主方向____。",
   "id": "9520598e75ed4381b89998445ded04ba"
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deepnote": {
     "block_id": "7714a237a6a449fe8f879b16eb719251",
     "type": "text-cell-h2",
     "source_block_id": null,
     "logical_key": null
    },
    "html_anchor": "cell-9"
   },
   "source": "## 2 · 相邻格：方向一样，描述子地址仍不同",
   "id": "7714a237a6a449fe8f879b16eb719251"
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deepnote": {
     "block_id": "b050462a425346539aa7a13177f5452d",
     "type": "markdown",
     "source_block_id": null,
     "logical_key": null
    },
    "html_anchor": "cell-10"
   },
   "source": "**预测：** r2c3中的45°与r2c4中的45°会写入同一个坐标吗？为什么前者为49、后者为57？\n\n**可改参数：** `neighbor_cell`默认[2,4]，可改[3,3]。行列标签从1开始，方向桶$b$与展开向量索引$j$从0开始：\n\n$$j=((r-1)\\times4+(c-1))\\times8+b.$$\n\nr2c3占48–55，其中45°桶$b=1$为49。邻格另构造45°票1、225°票2，直方图`[0,1,0,0,0,2,0,0]`。其余格设0只为看清地址，不表示真实图像平坦。",
   "id": "b050462a425346539aa7a13177f5452d"
  },
  {
   "cell_type": "code",
   "metadata": {
    "deepnote": {
     "block_id": "f7c3970bae824f14823fbb8abf855be2",
     "type": "code",
     "source_block_id": null,
     "logical_key": null
    },
    "html_anchor": "cell-11"
   },
   "source": "def address(r,c,b):\n    assert 1<=r<=4 and 1<=c<=4 and 0<=b<8\n    return ((r-1)*4+c-1)*8+b\nnr,nc=map(int,P['neighbor_cell']);assert 1<=nr<=4 and 1<=nc<=4 and (nr,nc)!=(2,3)\nneighbor_hist=hard_signed_vote([45,225],[1,2])\ndescriptor=np.zeros((4,4,8));descriptor[1,2]=sift_hist;descriptor[nr-1,nc-1]=neighbor_hist\nflat=descriptor.ravel();focus_j=address(2,3,1);neighbor_j=address(nr,nc,1)\ncheck('SIFT_49_address',focus_j==49)\ncheck('SIFT_neighbor_address_distinct',neighbor_j!=focus_j)\ncheck('SIFT_flatten_roundtrip',np.array_equal(flat.reshape(4,4,8),descriptor))\ncheck('SIFT_128_dimensions',len(flat)==128)\nfig,axes=plt.subplots(1,2,figsize=(11,4))\nfor r in range(4):\n    for c in range(4):\n        fill='#FFF0B7' if (r,c)==(1,2) else '#DDECE7' if (r,c)==(nr-1,nc-1) else '#F3F6FA'\n        axes[0].add_patch(plt.Rectangle((c,3-r),1,1,facecolor=fill,edgecolor='#AABBCD'))\n        axes[0].text(c+.5,3-r+.66,f'r{r+1}c{c+1}',ha='center');axes[0].text(c+.5,3-r+.28,f'{address(r+1,c+1,0)}-{address(r+1,c+1,7)}',ha='center',fontsize=9)\naxes[0].set(xlim=(0,4),ylim=(0,4),aspect='equal',title='4 x 4 cells; each stores eight bins');axes[0].axis('off')\naxes[1].bar(np.arange(128),flat,color=BLUE,width=.8);axes[1].axvline(focus_j,color=GOLD,lw=2,label='r2c3, b1 -> 49');axes[1].axvline(neighbor_j,color=GREEN,lw=2,label=f'neighbor b1 -> {neighbor_j}')\naxes[1].set(xlim=(0,127),xlabel='Descriptor index (zero-based)',ylabel='Vote',title='Same direction, different spatial addresses');axes[1].legend(fontsize=8);fig.tight_layout();plt.show()\ntable(['cell','bin','index','value'],[['r2c3',1,focus_j,float(flat[focus_j])],[f'r{nr}c{nc}',1,neighbor_j,float(flat[neighbor_j])]],'Address and value are different concepts')\nRESULTS['sift_address']={'index49_value':float(flat[49]),'neighbor_index':neighbor_j,'descriptor_size':int(flat.size)}",
   "id": "f7c3970bae824f14823fbb8abf855be2",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deepnote": {
     "block_id": "5926038adf6a450cac35ef57e281e0c3",
     "type": "markdown",
     "source_block_id": null,
     "logical_key": null
    },
    "html_anchor": "cell-12"
   },
   "source": [
    "**运行后看这里。** 并排读 `cell / bin / index / value`：移动空间格改变的是地址；方向桶编号相同，不意味着128维向量中使用同一个坐标。"
   ],
   "id": "5926038adf6a450cac35ef57e281e0c3"
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deepnote": {
     "block_id": "2be1559e1c19471da665cafdeeed571a",
     "type": "markdown",
     "source_block_id": null,
     "logical_key": null
    },
    "html_anchor": "cell-13"
   },
   "source": "**观察与边界：** 同方向的票在邻格拥有另一组地址；空间格顺序固定后才有可比较的128维坐标。改方向会改**值**，移动邻格会改**地址**。128维不等于128个物体类别。真实SIFT的空间插值会让一个梯度对邻格同时贡献，本构造刻意先把这种耦合拆开。\n\n**记录：** 改前/后邻格地址____；49处的值____；哪个变化来自位置，哪个来自方向____。",
   "id": "2be1559e1c19471da665cafdeeed571a"
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deepnote": {
     "block_id": "861c1b3323dc485393a746568a9f080f",
     "type": "text-cell-h2",
     "source_block_id": null,
     "logical_key": null
    },
    "html_anchor": "cell-14"
   },
   "source": "## 3 · HOG：无向梯度，两个相邻桶分票",
   "id": "861c1b3323dc485393a746568a9f080f"
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deepnote": {
     "block_id": "98c3c698001e486d91fd519f75e3ca33",
     "type": "markdown",
     "source_block_id": null,
     "logical_key": null
    },
    "html_anchor": "cell-15"
   },
   "source": "**预测：** 10°与190°在无向HOG中是否相同？如果9个桶中心间隔20°，10°、强度2应怎样分给0°与20°？\n\n**可改参数：** `hog_angles`、`hog_magnitudes`、`hog_offset`、`hog_gain`。默认[10,190,90]、强度[2,2,1]。角度先mod180，再向相邻桶中心线性分票，末尾与0°循环相接。\n\n这里另把同一方向直方图乘亮度幅度`hog_gain`，比较原票与L2归一化票。它演示归一化机制，没有构建完整图像HOG。",
   "id": "98c3c698001e486d91fd519f75e3ca33"
  },
  {
   "cell_type": "code",
   "metadata": {
    "deepnote": {
     "block_id": "bc9751b5f3cc4b439093ed70886d2811",
     "type": "code",
     "source_block_id": null,
     "logical_key": null
    },
    "html_anchor": "cell-16"
   },
   "source": "def unsigned_vote(theta,magnitude,bins=9):\n    theta=np.asarray(theta,dtype=float)%180.;magnitude=np.asarray(magnitude,dtype=float)\n    assert theta.shape==magnitude.shape and np.all(magnitude>=0)\n    q=theta/(180./bins);lo=np.floor(q).astype(int)%bins;fraction=q-np.floor(q)\n    hist=np.zeros(bins);np.add.at(hist,lo,magnitude*(1-fraction));np.add.at(hist,(lo+1)%bins,magnitude*fraction)\n    return hist\nha=np.asarray(P['hog_angles'],dtype=float)+float(P['hog_offset']);hm=np.asarray(P['hog_magnitudes'],dtype=float)\nhog=unsigned_vote(ha,hm);rotated=unsigned_vote(ha+180,hm)\ngain=float(P['hog_gain']);assert gain>0;eps=float(P['normalization_epsilon']);assert eps>0\nscaled=hog*gain\nnormalize=lambda a:a/np.sqrt(np.dot(a,a)+eps**2)\ncheck('HOG_unsigned_180_invariance',np.allclose(hog,rotated))\ncheck('HOG_vote_mass',np.isclose(hog.sum(),hm.sum()))\ncheck('HOG_10deg_two_votes',np.allclose(unsigned_vote([10],[2]),[1,1,0,0,0,0,0,0,0]))\nfig,axes=plt.subplots(1,2,figsize=(11,4));bins=np.arange(9)*20\naxes[0].bar(bins-3,hog,6,color=BLUE,label='base');axes[0].bar(bins+3,scaled,6,color=GOLD,label='gain-scaled')\naxes[1].bar(bins-3,normalize(hog),6,color=BLUE,label='base normalized');axes[1].bar(bins+3,normalize(scaled),6,color=GREEN,label='scaled normalized')\nfor ax in axes:ax.set(xticks=bins,xlabel='Unsigned bin center (degrees)',ylabel='Vote');ax.legend(fontsize=8)\naxes[0].set_title('Before normalization');axes[1].set_title('One-vector L2 mechanism');fig.tight_layout();plt.show()\ntable(['input angle','angle mod 180','magnitude'],[[f'{a:g}',f'{a%180:g}',f'{m:g}'] for a,m in zip(ha,hm)],'Direction polarity is discarded')\nprint('Raw votes:',hog,' normalized:',normalize(hog))\nprint('Classic window count: 64 x 128 px / 8 px cells -> 8 x 16 cells; 2 x 2-cell blocks -> 7 x 15 x 36 =',7*15*36)\nRESULTS['hog']={'histogram':hog.tolist(),'normalized':normalize(hog).tolist(),'unsigned_180_error':float(np.max(np.abs(hog-rotated))),'illustrative_window_dimensions':3780}",
   "id": "bc9751b5f3cc4b439093ed70886d2811",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deepnote": {
     "block_id": "fa1595d3dcd0457f9afd47f609afd8e6",
     "type": "markdown",
     "source_block_id": null,
     "logical_key": null
    },
    "html_anchor": "cell-17"
   },
   "source": [
    "**运行后看这里。** 左图看缩放后的原票，右图看归一化后的票；10°与190°先取模180后才进入同一组桶。"
   ],
   "id": "fa1595d3dcd0457f9afd47f609afd8e6"
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deepnote": {
     "block_id": "16fd8c00a4384560b1e0740e81da7f66",
     "type": "markdown",
     "source_block_id": null,
     "logical_key": null
    },
    "html_anchor": "cell-18"
   },
   "source": "**观察与边界：** 无向方向合并了相反梯度；9桶与插值规则必须一起说明。真实HOG通常对包含多个cell的block归一化，并可能使用截断；这里单向量L2只隔离幅度缩放。除以带epsilon的范数近似抵消增益，epsilon并非严格齐次。3780维依赖所写窗口、cell、block及步距配置，不是所有HOG的固定长度。\n\n**记录：** 预测分票____；角度/增益修改____；票量与归一化形状____；仍缺的完整算法步骤____。",
   "id": "16fd8c00a4384560b1e0740e81da7f66"
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deepnote": {
     "block_id": "8ce2116c587f4c9487a6437469cfa8ce",
     "type": "text-cell-h2",
     "source_block_id": null,
     "logical_key": null
    },
    "html_anchor": "cell-19"
   },
   "source": "## 4 · 共同词典：局部模式组成相同，布局仍可不同",
   "id": "8ce2116c587f4c9487a6437469cfa8ce"
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deepnote": {
     "block_id": "14afb0b9776447e28c043198f3429207",
     "type": "markdown",
     "source_block_id": null,
     "logical_key": null
    },
    "html_anchor": "cell-20"
   },
   "source": "**两张布局明显不同的图，为什么汇总成同一个向量？** 先找出上半部怎样移动，再看最右侧的柱。\n\n![交换上下半图后不同的视觉词布局仍有相同词袋](data:image/webp;base64,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)\n\n*默认参考：A/B/C 是固定共同词典中的视觉词编号。两图都计数为 (6,6,4)，只改变了布局。这里是 Slides 的 16 格构造，由本页代码本地计算，不是对象类别或实拍分类结果。*\n\n代码先用二维描述检查“最近共同中心 → 编号 → 计数”，中心固定为 (1,1)、(5,4)，应用阶段不为每张新图重新聚类。\n\n**只改一次：** 先交换 `WORDS_A` 中两个不同词的位置，预测词袋是否变化；另一轮再把一个 A 改成 C，预测哪个柱减少、哪个增加。`new_descriptors` 可单独用于检查新描述如何分配到同一词典。\n\n**这就留下了下一个问题：怎样找回粗布局？** 下节沿用同样的 A/B/C 编号，给计数加上区域地址；全图柱仍重合，2×2 区域柱开始不同。\n\n![同一词布局加入2乘2空间区域后的计数](data:image/webp;base64,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)\n\n*下一节默认构造：全图 3 维 + 四个区域各 3 维 = 15 维，按行优先拼接，各区权重为 1；两向量平方距离为 56。先理解本节为何丢失位置，再进入下一节只分左/右的对照。图由当前代码本地计算。*\n",
   "id": "14afb0b9776447e28c043198f3429207"
  },
  {
   "cell_type": "code",
   "metadata": {
    "deepnote": {
     "block_id": "c9b6eed7dd364b78934270ee3d4b15da",
     "type": "code",
     "source_block_id": null,
     "logical_key": null
    },
    "html_anchor": "cell-21"
   },
   "source": "training_groups=[np.array([[0,0],[0,2],[2,0],[2,2]]),np.array([[4,3],[4,5],[6,3],[6,5]])]\ncodebook=np.stack([g.mean(0) for g in training_groups])\nnew_descriptors=np.array([[2.,1.],[4.,4.]]) # 学生可改这两个新局部描述；不要重建词典。\ndistance=((new_descriptors[:,None,:]-codebook[None,:,:])**2).sum(2)\nassignment=distance.argmin(1);counts=np.bincount(assignment,minlength=len(codebook))\ntable(['new descriptor','nearest center ID','squared distances'],[[str(d.tolist()),int(a)+1,str(ds.tolist())] for d,a,ds in zip(new_descriptors,assignment,distance)],'Training centers retained: (1,1), (5,4)')\nSLIDE_WORDS_A=np.array(['A','A','A','B','A','A','B','C','B','B','C','C','B','C','A','B']).reshape(4,4)\nWORDS_A=SLIDE_WORDS_A.copy() # 可在这里改位置或组成；例如 WORDS_A[0,0]='C'。\nassert WORDS_A.shape==(4,4) and np.isin(WORDS_A,['A','B','C']).all()\nWORDS_B=np.vstack([WORDS_A[2:],WORDS_A[:2]])\nvocabulary=['A','B','C']\ncount_words=lambda a:np.array([(a==k).sum() for k in vocabulary])\nbag_a,bag_b=count_words(WORDS_A),count_words(WORDS_B)\ncheck('row_permutation_preserves_bag',np.array_equal(bag_a,bag_b))\nprint('Two layouts actually differ:',not np.array_equal(WORDS_A,WORDS_B))\nif np.array_equal(WORDS_A,SLIDE_WORDS_A):check('slide_bag_counts',np.array_equal(bag_a,[6,6,4]))\nfig,axes=plt.subplots(1,3,figsize=(11,3.5));colors={'A':BLUE,'B':GOLD,'C':GREEN}\nfor ax,words,title in zip(axes[:2],[WORDS_A,WORDS_B],['Layout A','Layout B: swap row halves']):\n    for r in range(4):\n        for c in range(4):\n            word=words[r,c];ax.add_patch(plt.Rectangle((c,3-r),1,1,facecolor=colors[word],edgecolor='white'));ax.text(c+.5,3-r+.5,word,ha='center',va='center',color='black' if word=='B' else 'white',fontsize=14)\n    ax.set(xlim=(0,4),ylim=(0,4),aspect='equal',title=title);ax.axis('off')\naxes[2].bar(np.arange(3)-.17,bag_a,.34,color=BLUE,label='layout A');axes[2].bar(np.arange(3)+.17,bag_b,.34,color=GOLD,label='layout B');axes[2].set(xticks=np.arange(3),xticklabels=vocabulary,ylabel='Count',title='Same global bag');axes[2].legend();fig.tight_layout();plt.show()\nRESULTS['bag']={'centers':codebook.tolist(),'assignments':assignment.tolist(),'new_counts':counts.tolist(),'layout_A_counts':bag_a.tolist(),'layout_B_counts':bag_b.tolist()}\n# 真正的 Lloyd fit 已独立迁入第三讲 CLS02 前史选读；本节只保留预设均值与布局计数。\n",
   "id": "c9b6eed7dd364b78934270ee3d4b15da",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deepnote": {
     "block_id": "11534b0f6e464081bcb79d3b3629943c",
     "type": "markdown",
     "source_block_id": null,
     "logical_key": null
    },
    "html_anchor": "cell-22"
   },
   "source": [
    "**运行后看这里。** 先确认两张布局图确实不同，再核对两条全局计数是否相同。最近中心分配与空间布局是两项分开的构造，不能把词编号当物体类别。"
   ],
   "id": "11534b0f6e464081bcb79d3b3629943c"
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deepnote": {
     "block_id": "59497c663a9e44cbb22e9c2dc560513c",
     "type": "markdown",
     "source_block_id": null,
     "logical_key": null
    },
    "html_anchor": "cell-23"
   },
   "source": "**观察与边界：** 默认词袋都是(6,6,4)，因此仅依靠该向量的确定性分类器不能区分这两张构造图。中心来自训练描述的约定分组均值，此小构造没有运行完整K-means；真正词典学习应使用训练数据。单张图各自重建编号会破坏已有分类器的坐标含义。\n\n**记录：** 我只改位置/组成中的____；词袋是否变____；共同词典固定了____；丢失的是____。\n\n**另一个实际 fit，而不是改名。** 后半代码只把 8 个构造训练描述交给 Lloyd 交替分配/更新中心，不提供类别标签，也不提供原预设分组。迭代目标是训练描述到最近中心的平方距离之和；训练结束后冻结中心与编号，再分配 4 个未参与 fit 的保留描述。图像词典与现代 SSL 都可从无类别标签数据构造学习信号，但目标、参数化和不变性假设不同。这个小实验只证明本代码完成了无标签聚类学习，不证明学到了语义。\n\n**只改一个条件**：固定训练描述与 seed，只移动一个保留描述，检查训练中心不变、保留分配可能改变。再另起一轮只改训练 seed，比较目标与中心；选择 seed 不能偷看保留标签。本例没有类别标签，也不报告分类准确率。\n\n![2026-09-18 本地执行 Lloyd fit：训练目标与保留分配](data:image/png;base64,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)\n\n*本次本地 CPU 保存输出：seed=20260918；训练平方距离和 36→23.7333→16→16；最终中心 (1,1)、(5,4)，保留分配 [0,1,0,1]。描述子是明确构造的数据，fit 是实际执行。Deepnote 云端未运行。*",
   "id": "59497c663a9e44cbb22e9c2dc560513c"
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deepnote": {
     "block_id": "919e64e67b5e431295cb1a7ee3b7ea92",
     "type": "text-cell-h2",
     "source_block_id": null,
     "logical_key": null
    },
    "html_anchor": "cell-24"
   },
   "source": "## 5 · 计数加地址：上/下与二维空间金字塔",
   "id": "919e64e67b5e431295cb1a7ee3b7ea92"
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deepnote": {
     "block_id": "fd6add884339493bacb979cb17d953f5",
     "type": "markdown",
     "source_block_id": null,
     "logical_key": null
    },
    "html_anchor": "cell-25"
   },
   "source": "**预测：** 对刚才两张图分别数上半和下半，会得到什么？如果只保留左/右区域，能分开“交换上下半图”吗？\n\n**可改参数：** `pyramid_grid`默认[2,2]（2行×2列），可试[1,2]（左/右）或[2,1]（上/下）。网格行列须整除4。\n\n先核对Slides的上(5,2,1)、下(1,4,3)，再拼接“全图计数+按行优先的区域计数”。每一级区域权重均为1；这是明确的两级计数构造，不宣称复现论文加权kernel。",
   "id": "fd6add884339493bacb979cb17d953f5"
  },
  {
   "cell_type": "code",
   "metadata": {
    "deepnote": {
     "block_id": "0952ed8eeccd497a805b4ef8553f3e0b",
     "type": "code",
     "source_block_id": null,
     "logical_key": null
    },
    "html_anchor": "cell-26"
   },
   "source": "upper_a,lower_a=count_words(WORDS_A[:2]),count_words(WORDS_A[2:])\nupper_b,lower_b=count_words(WORDS_B[:2]),count_words(WORDS_B[2:])\nif np.array_equal(WORDS_A,SLIDE_WORDS_A):check('slide_upper_lower_counts',np.array_equal(upper_a,[5,2,1]) and np.array_equal(lower_a,[1,4,3]))\ncheck('swapped_regions',np.array_equal(upper_a,lower_b) and np.array_equal(lower_a,upper_b))\ntable(['layout','global A/B/C','upper A/B/C','lower A/B/C'],[['A',str(bag_a.tolist()),str(upper_a.tolist()),str(lower_a.tolist())],['B',str(bag_b.tolist()),str(upper_b.tolist()),str(lower_b.tolist())]],'Coarse spatial addresses restore a distinction')\ngr,gc=map(int,P['pyramid_grid']);assert gr>0 and gc>0 and 4%gr==0 and 4%gc==0\ndef pyramid(words,rows,cols):\n    result=[count_words(words)];names=['global']\n    for r in range(rows):\n        for c in range(cols):\n            patch=words[r*4//rows:(r+1)*4//rows,c*4//cols:(c+1)*4//cols]\n            result.append(count_words(patch));names.append(f'r{r+1}c{c+1}')\n    return np.concatenate(result),names\npa,regions=pyramid(WORDS_A,gr,gc);pb,_=pyramid(WORDS_B,gr,gc)\ncheck('pyramid_dimensions',len(pa)==3*(1+gr*gc))\ncheck('pyramid_count_mass',pa.sum()==pb.sum()==32) # 每个词在全图和区域级各计一次。\nfig,ax=plt.subplots(figsize=(11,3.7));ids=np.arange(len(pa));ax.bar(ids-.17,pa,.34,label='layout A',color=BLUE);ax.bar(ids+.17,pb,.34,label='layout B',color=GOLD)\nax.set(xticks=np.arange(len(regions))*3+1,xticklabels=regions,xlabel='Region address (three A/B/C counts each)',ylabel='Count',title=f'Global + {gr} x {gc} regional counts: {len(pa)} dimensions')\nfor edge in np.arange(len(regions)+1)*3-.5:ax.axvline(edge,color='#ADBFD2',lw=.7)\nax.legend();fig.tight_layout();plt.show()\ndistance2=int(np.square(pa-pb).sum())\nprint('Squared distance: bag =',int(np.square(bag_a-bag_b).sum()),'; pyramid =',distance2)\nRESULTS['pyramid']={'grid':[gr,gc],'dimensions':len(pa),'squared_distance':distance2,'vector_A':pa.tolist(),'vector_B':pb.tolist()}\nprint('Checks passed:',sum(CHECKS.values()),'/',len(CHECKS))",
   "id": "0952ed8eeccd497a805b4ef8553f3e0b",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deepnote": {
     "block_id": "755aba57a9af4e1ab8f05d7a46569d76",
     "type": "markdown",
     "source_block_id": null,
     "logical_key": null
    },
    "html_anchor": "cell-27"
   },
   "source": [
    "**运行后看这里。** 核对 `grid / dimensions / squared_distance`：全图3维加每个分区3维，改变网格同时改变表示长度与可区分的布局。"
   ],
   "id": "755aba57a9af4e1ab8f05d7a46569d76"
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deepnote": {
     "block_id": "2a486b5a55ac42bcb1694e76e25bce30",
     "type": "markdown",
     "source_block_id": null,
     "logical_key": null
    },
    "html_anchor": "cell-28"
   },
   "source": "**观察与边界：** 默认2×2区域能区分布局；只分左/右时，这次上下交换仍可能不可见。保留多少位置不是越多越好：细网格维度更大，也可能对小位移更敏感。区域内部重排仍可能得到相同计数；金字塔没有恢复原图全部位置。\n\n**记录模板（自愿练习）：** 我认为应保留的空间差异____；只改网格____；预测距离____；实际距离____；仍然不可见的重排____。",
   "id": "2a486b5a55ac42bcb1694e76e25bce30"
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deepnote": {
     "block_id": "67f331c73d4d478d99accea99a76a5c1",
     "type": "text-cell-h3",
     "source_block_id": null,
     "logical_key": null
    },
    "html_anchor": "cell-29"
   },
   "source": "### 来源与身份",
   "id": "67f331c73d4d478d99accea99a76a5c1"
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deepnote": {
     "block_id": "68f08e87cbb24cf1a3be1776d2ae735f",
     "type": "markdown",
     "source_block_id": null,
     "logical_key": null
    },
    "html_anchor": "cell-30"
   },
   "source": "- 对齐当前Slides `sift-local-descriptor` 的构造r2c3、邻格、索引49；`hog-local-shape` 的无向方向组织；`codebook-training-versus-assignment` 与 `spatial-pyramid-restores-position` 的共同坐标和布局取舍。\n- 算法背景：[Lowe, SIFT (2004)](https://www.cs.ubc.ca/~lowe/papers/ijcv04.pdf)、[Dalal & Triggs, HOG (2005)](https://lear.inrialpes.fr/people/triggs/pubs/Dalal-cvpr05.pdf)、[Lazebnik et al., Spatial Pyramid (2006)](https://slazebni.cs.illinois.edu/publications/cvpr06b.pdf)。本页为新写的小机制构造，不是这些论文的完整实现或实验复现。\n- 既有`public/content/features-v5.json`中保存了真实计算的SIFT/HOG/词频及其来源说明；本Notebook默认不依赖该文件，也不把本页构造值冒充其中实测值。\n- 2025Fall/2026Spring卷积与LeNet历史快照属于算子/模型主题；本页不把这些历史notebook误列为SIFT/HOG实现来源。没有读取或复用历史评论答案。",
   "id": "68f08e87cbb24cf1a3be1776d2ae735f"
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deepnote": {
     "block_id": "70d0a16ba59c4e6690d7956b2a0cfa90",
     "type": "markdown",
     "source_block_id": null,
     "logical_key": null
    },
    "html_anchor": "cell-31"
   },
   "source": "**返回课堂导读** · [返回幻灯片](https://codingai-lec02.pages.dev/)",
   "id": "70d0a16ba59c4e6690d7956b2a0cfa90"
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deepnote": {
     "block_id": "f61821f1ca654566a0a228d595b0db49",
     "type": "text-cell-h2",
     "source_block_id": null,
     "logical_key": null
    },
    "html_anchor": "cell-32"
   },
   "source": "## 按问题查阅：补充推导与机制核对",
   "id": "f61821f1ca654566a0a228d595b0db49"
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deepnote": {
     "block_id": "fa582b8c78c6426fabcee364f2027756",
     "type": "markdown",
     "source_block_id": null,
     "logical_key": null
    },
    "html_anchor": "cell-33"
   },
   "source": [
    "**按问题查阅。** 下方第1–5节分别补充局部描述、HOG、共同词典与空间地址、分类器读出和部件结构。它们解释前面短练省略了什么，不是需要全部复现的训练流水线。\n",
    "\n",
    "若只想做一个无需额外库的核对，可直接去文末“独立核对：先预测，再运行”。"
   ],
   "id": "fa582b8c78c6426fabcee364f2027756"
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deepnote": {
     "block_id": "bb141ca7097f47a5a1b203c02b4dd063",
     "type": "text-cell-h2",
     "source_block_id": null,
     "logical_key": null
    },
    "html_anchor": "cell-34"
   },
   "source": "## 1. 从滤波器组到可比较的局部描述",
   "id": "bb141ca7097f47a5a1b203c02b4dd063"
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deepnote": {
     "block_id": "ce982ce6302e49c780c28fd41f22e185",
     "type": "markdown",
     "source_block_id": null,
     "logical_key": null
    },
    "html_anchor": "cell-35"
   },
   "source": "滤波器组用不同方向、尺度的局部测量描述纹理。亮度值相同的像素，周围结构可能不同；多种响应组成的向量能保留这些差别。选择哪几种测量仍是建模决定。\n\nSIFT 的关键点检测与描述是两个阶段。尺度空间中，相邻模糊图相减形成 DoG；一个候选与本层 8 点、相邻尺度各 9 点比较，共 26 个邻居。极值只是候选，后续还需定位并检查对比度和边缘响应稳定性。\n\n以关键点尺度和主方向建立局部坐标后，描述子在 4×4 个空间格中分别统计 8 个方向桶，形成 128 个分量。每个数有“格子位置＋方向”的地址。贡献在相邻空间格与方向桶间插值，之后归一化、截断、再次归一化，减少量化与局部幅值变化的影响；128 不是 128 个对象类别。\n\n匹配还要检查歧义。最近/次近距离 0.20/0.80 的比为 0.25，而 0.20/0.22 约为 0.91。最近距离一样，独特性却不同。距离比筛选不能替代多个匹配的几何一致性检查。主讲的旋转 24°、缩放 0.78 对应实验检验局部匹配，不是整图分类准确率。\n\n**先判断**：为什么已经找到最近描述，还要看次近描述？\n\n**核对提示**：重复纹理可能产生多个同样相近的候选。最近/次近比值衡量相对歧义；即使通过筛选，仍可能出现不符合整体几何关系的匹配。\n\n**来源与继续阅读**\n\n- [Leung & Malik · Representing and Recognizing Texture with 3D Textons (2001)](https://people.eecs.berkeley.edu/~malik/papers/LM-3dtexton.pdf)\n- [CS231n 2026 · CNNs](https://cs231n.stanford.edu/slides/2026/lecture_5.pdf)\n- [Lowe · Distinctive Image Features from Scale-Invariant Keypoints (2004)](https://www.cs.ubc.ca/~lowe/papers/ijcv04.pdf)",
   "id": "ce982ce6302e49c780c28fd41f22e185"
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deepnote": {
     "block_id": "4c934f8a90bb49b1bb35c1100b6f16a7",
     "type": "text-cell-h2",
     "source_block_id": null,
     "logical_key": null
    },
    "html_anchor": "cell-36"
   },
   "source": "## 2. HOG：一个像素怎样进入一个窗口向量",
   "id": "4c934f8a90bb49b1bb35c1100b6f16a7"
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deepnote": {
     "block_id": "0a3f8ff65bcb4db4b9c21371fe7789f9",
     "type": "markdown",
     "source_block_id": null,
     "logical_key": null
    },
    "html_anchor": "cell-37"
   },
   "source": "HOG 围绕固定分析窗口和规则网格组织方向统计，和每个 SIFT 关键点的局部坐标体系不同。它可以直接把窗口描述交给分类器，不必绕经视觉词袋。\n\n课程保存的实际像素测量位于窗口第 5 行第 4 列 cell（程序索引 row=4,col=3）：方向约 170.64°，幅值 0.36168。无符号方向中 180° 与 0° 相接，贡献按角度分到 160° 桶约 0.16927、0° 桶约 0.19241，两份相加仍为原幅值。这里给的是先前计算并保留的数值，不是本次重新运行测得的结果。\n\n在 64×128 窗口、8×8 像素 cell、9 个方向桶的具体配置下，2×2 cells 构成 36 维 block；按一个 cell 步长滑动，共 (8−1)×(16−1)=7×15 个 block，拼接为 3780 维。相邻 block 重叠，使同一 cell 参加不同的局部归一化。改变配置，维度也会变。\n\n**先判断**：为什么不能直接用 8×16×9 算作这个 HOG 配置的最终长度？\n\n**核对提示**：那是逐 cell 直方图的长度，尚未按重叠的 2×2 block 做归一化与拼接。最终向量包含重叠 block 的多份局部表示。\n\n**来源与继续阅读**\n\n- [Dalal & Triggs · Histograms of Oriented Gradients for Human Detection (2005)](https://lear.inrialpes.fr/people/triggs/pubs/Dalal-cvpr05.pdf)\n- [Felzenszwalb et al. · Object Detection with Discriminatively Trained Part-Based Models (2010)](https://people.cs.uchicago.edu/~pff/papers/lsvm-pami.pdf)",
   "id": "0a3f8ff65bcb4db4b9c21371fe7789f9"
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deepnote": {
     "block_id": "201b0bfaeea04802ac3c171516f3af5e",
     "type": "text-cell-h2",
     "source_block_id": null,
     "logical_key": null
    },
    "html_anchor": "cell-38"
   },
   "source": "## 3. 共同词典、层级地址与位置交换",
   "id": "201b0bfaeea04802ac3c171516f3af5e"
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deepnote": {
     "block_id": "e087c487a07c4954901c8b4a6926328a",
     "type": "markdown",
     "source_block_id": null,
     "logical_key": null
    },
    "html_anchor": "cell-39"
   },
   "source": "课程的街景展示先前实际提取的 600 个 SIFT 描述，经归一化后聚为 K=6 个簇。词中心是 128 维数值向量；代表裁片来自靠近中心的真实观测，不是“中心生成的物体”。编号 1–6 不预先代表车轮、窗户等语义类别。\n\n用于分类时，词典应由训练数据建立，新图沿用同一套中心与编号，再统计数量。如果每张图独立聚类，“词 1”的含义会改变，分类器收到的坐标便失去共同定义。街景六词计数只演示描述与量化，没有连到一个已经训练好的街景分类器。\n\n纯词袋只保留各词总数。交换同一批已提取描述的位置，计数不变，布局差别被合并。空间金字塔让计数附带区域地址：全图 1 格、2×2 的 4 格、4×4 的 16 格，共 21 个区域；K=3 时有 63 个分量。同一描述可进入不同层级，越细的格子越能区别布局，也越容易受跨格移动影响。这里的 16 个描述位置交换是独立的手工构造，不能说成街景分类实验。\n\n**先判断**：位置交换后，什么时候全局词袋不变而空间统计改变？\n\n**核对提示**：固定词典和已提取描述，只改变它们的坐标时，全局计数不变；只要跨过所选区域边界，区域计数就可能改变。若重新从修改后的像素提取描述，还会引入另一种变化，必须另行说明。\n\n**来源与继续阅读**\n\n- [Sivic & Zisserman · Video Google (2003)](https://www.robots.ox.ac.uk/~vgg/publications/2003/Sivic03/sivic03.pdf)\n- [Csurka et al. · Visual Categorization with Bags of Keypoints (2004)](https://www.cs.cmu.edu/~efros/courses/LBMV07/Papers/csurka-eccv-04.pdf)\n- [Lazebnik et al. · Beyond Bags of Features (2006)](https://slazebni.cs.illinois.edu/publications/cvpr06b.pdf)\n- [CS231n 2026 · CNNs](https://cs231n.stanford.edu/slides/2026/lecture_5.pdf)\n- [Lazebnik, Schmid & Ponce · Beyond Bags of Features: Spatial Pyramid Matching (2006)](https://inria.hal.science/inria-00548585/file/cvpr06_lana.pdf)",
   "id": "e087c487a07c4954901c8b4a6926328a"
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deepnote": {
     "block_id": "3113bed86d6b409a8e163b75c72de442",
     "type": "text-cell-h2",
     "source_block_id": null,
     "logical_key": null
    },
    "html_anchor": "cell-40"
   },
   "source": "## 4. 分类器读出：分数、间隔和表示各做什么",
   "id": "3113bed86d6b409a8e163b75c72de442"
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deepnote": {
     "block_id": "cb35ef50f0af438b90d16c37782380a5",
     "type": "markdown",
     "source_block_id": null,
     "logical_key": null
    },
    "html_anchor": "cell-41"
   },
   "source": "标签训练决定特征空间中的边界，新图沿同一表示与权重计算分数。以 w=(1,0),b=−1 为例，边界是 x₁=1；点 (3,1) 的分数为 2。同时将 w,b 乘 10，边界不变，分数变 20，但分数除以 ‖w‖ 得到的有符号几何距离仍为 2。裸分数不能直接当距离。\n\n二分类 hinge=max(0,1−ys) 鼓励功能间隔，配合权重惩罚才约束参数尺度。主讲只需训练与应用接口，本节才展开间隔的含义。\n\n“线性”描述的是当前特征空间中的读出。手工构造的内外环例子中，外点半径平方为 4，内点为 0.25；映射 φ(x)=x₁²+x₂² 后可以用一维阈值区分。这个新表示改变了系统，不是仅把同一批原坐标的线性参数调好。课程的二维 SVM 训练点、内外环和街景词袋是不同算例。\n\n**先判断**：若新表示提高了分类准确率，可以把全部收益归给 SVM 吗？\n\n**核对提示**：不能。前面的表示改变了哪些信息可被简单边界读取。应固定数据协议，并明确比较的是完整流程还是相同特征上的分类器。\n\n**来源与继续阅读**\n\n- [Stanford CS231n · Linear Classification and SVM Loss](https://cs231n.github.io/linear-classify/)\n- [CS231n 2026 · Image Classification with Linear Classifiers](https://cs231n.stanford.edu/slides/2026/lecture_2.pdf)\n- [Dalal & Triggs · Histograms of Oriented Gradients (2005)](https://lear.inrialpes.fr/people/triggs/pubs/Dalal-cvpr05.pdf)\n- [ILSVRC 2012 · Official Results](https://image-net.org/challenges/LSVRC/2012/results.html)",
   "id": "cb35ef50f0af438b90d16c37782380a5"
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deepnote": {
     "block_id": "02fa92668c2840fcba4a1ad91496703e",
     "type": "text-cell-h2",
     "source_block_id": null,
     "logical_key": null
    },
    "html_anchor": "cell-42"
   },
   "source": "## 5. 从统计到部件：结构先验并非 CNN 才出现",
   "id": "02fa92668c2840fcba4a1ad91496703e"
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deepnote": {
     "block_id": "4e055c1357bc4730b7514f26992735f2",
     "type": "markdown",
     "source_block_id": null,
     "logical_key": null
    },
    "html_anchor": "cell-43"
   },
   "source": "DPM 同时使用对象整体的根模型、局部部件响应和形变代价，允许部件偏离参考位置，承担检测中的类别与位置判断。以同一个根框为参照：候选 A 外观分 4、偏移 0，总分 4；候选 B 外观分 6、偏移 3、代价 0.5Δx²=4.5，总分 1.5。因此局部最像的部件，不一定组成总分最高的布局。\n\n传统流程持续发展：2010 年 NEC–UIUC 系统使用 SIFT/LBP、非线性编码与随机 SVM，2011 年 XRCE 使用高维 Fisher 表示等机制。这里是既有历史系统资料，不能从两个名称推导某个单一组件的因果效果。\n\n传统路线已经包含学习词典、训练分类器和结构先验。端到端 CNN 的转折是让最终任务误差可以继续影响前端测量；它没有取消人工选择局部性、共享和网络结构。前端已经丢掉的区别，后端无法凭空恢复，因而需要判断哪些接口应当由任务共同调整。\n\n**先判断**：DPM 的形变代价和词袋的位置忽略，分别表达什么偏好？\n\n**核对提示**：DPM 保留相对布局并允许有代价的移动；纯词袋主动合并位置差别。它们针对不同任务与计算组织，不应排成简单的“有位置就更好”。\n\n**来源与继续阅读**\n\n- [Felzenszwalb et al. · Object Detection with Discriminatively Trained Part-Based Models (2010)](https://people.cs.uchicago.edu/~pff/papers/lsvm-pami.pdf)\n- [ILSVRC 2012 · Official Results](https://image-net.org/challenges/LSVRC/2012/results.html)\n- [Felzenszwalb et al. · Object Detection with Discriminatively Trained Part-Based Models (2010)](https://cs.brown.edu/people/pfelzens/papers/lsvm-pami.pdf)\n- [CS231n 2026 · CNNs](https://cs231n.stanford.edu/slides/2026/lecture_5.pdf)\n- [Krizhevsky et al. · ImageNet Classification with Deep CNNs (2012)](https://papers.nips.cc/paper_files/paper/2012/file/c399862d3b9d6b76c8436e924a68c45b-Paper.pdf)\n- [ILSVRC 2010 · Official Results and Descriptions](https://image-net.org/challenges/LSVRC/2010/results)\n- [ILSVRC 2011 · Official Results and Descriptions](https://image-net.org/challenges/LSVRC/2011/results)",
   "id": "4e055c1357bc4730b7514f26992735f2"
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deepnote": {
     "block_id": "661d6e4bf1004050bbfa04c8eb5bb6b8",
     "type": "text-cell-h2",
     "source_block_id": null,
     "logical_key": null
    },
    "html_anchor": "cell-44"
   },
   "source": "## 从主讲继续：推导边界与研究来源",
   "id": "661d6e4bf1004050bbfa04c8eb5bb6b8"
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deepnote": {
     "block_id": "d2a976cf407d48e0b6a359709f9dee3f",
     "type": "markdown",
     "source_block_id": null,
     "logical_key": null
    },
    "html_anchor": "cell-45"
   },
   "source": "把传统流程的训练与应用分别写一遍：训练图建立共同词典，标签训练读出；新图沿用词典编号、统计方式与分类器权重。SIFT 对应、HOG 投票、街景六词计数、位置交换与二维 SVM 的算例身份必须分别保留。用流程图连接接口，不把它们拼成一次已经完成的端到端街景分类实验。\n\n**主讲研究来源**\n\n- [CS231n 2026 · CNNs](https://cs231n.stanford.edu/slides/2026/lecture_5.pdf)\n- [Sivic & Zisserman · Video Google (2003)](https://www.robots.ox.ac.uk/~vgg/publications/2003/Sivic03/sivic03.pdf)\n- [Lazebnik et al. · Beyond Bags of Features (2006)](https://slazebni.cs.illinois.edu/publications/cvpr06b.pdf)\n- [Lowe · Distinctive Image Features from Scale-Invariant Keypoints (2004)](https://www.cs.ubc.ca/~lowe/papers/ijcv04.pdf)\n- [OpenCV 4.13 · Introduction to SIFT](https://docs.opencv.org/4.13.0/da/df5/tutorial_py_sift_intro.html)\n- [Dalal & Triggs · Histograms of Oriented Gradients (2005)](https://lear.inrialpes.fr/people/triggs/pubs/Dalal-cvpr05.pdf)\n- [Felzenszwalb et al. · Object Detection with Discriminatively Trained Part-Based Models (2010)](https://people.cs.uchicago.edu/~pff/papers/lsvm-pami.pdf)\n- [Csurka et al. · Visual Categorization with Bags of Keypoints (2004)](https://www.cs.cmu.edu/~efros/courses/LBMV07/Papers/csurka-eccv-04.pdf)\n- [CS231n 2026 · Image Classification with Linear Classifiers](https://cs231n.stanford.edu/slides/2026/lecture_2.pdf)\n- [ILSVRC 2012 · Official Results](https://image-net.org/challenges/LSVRC/2012/results.html)\n- [D2L · Convolutions for Images](https://d2l.ai/chapter_convolutional-neural-networks/conv-layer.html)\n- [D2L · LeNet](https://d2l.ai/chapter_convolutional-neural-networks/lenet.html)",
   "id": "d2a976cf407d48e0b6a359709f9dee3f"
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deepnote": {
     "block_id": "ca009e0465a744eb984298cacaa8c1c7",
     "type": "text-cell-h2",
     "source_block_id": null,
     "logical_key": null
    },
    "html_anchor": "cell-46"
   },
   "source": "## 独立核对：先预测，再运行",
   "id": "ca009e0465a744eb984298cacaa8c1c7"
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deepnote": {
     "block_id": "afaa57d17ea34cb09ad0081ec18ad4b4",
     "type": "markdown",
     "source_block_id": null,
     "logical_key": null
    },
    "html_anchor": "cell-47"
   },
   "source": "下面是完整、独立的小规模 Python 核对，只使用标准库；无需下载、安装、训练或访问外部服务。它保留“教学构造”的身份。先写预测，运行后解释哪一项支持判断，再只改一个条件复查。下方没有预填远端执行输出。",
   "id": "afaa57d17ea34cb09ad0081ec18ad4b4"
  },
  {
   "cell_type": "code",
   "metadata": {
    "deepnote": {
     "block_id": "06256d442549408687c38f2a20784a60",
     "type": "code",
     "source_block_id": null,
     "logical_key": null
    },
    "html_anchor": "cell-48"
   },
   "source": "# 独立构造：只交换已提取描述的位置，词编号固定。\nfrom collections import Counter\ntokens=[('A',0.1),('A',0.2),('B',0.7),('B',0.8)]\nswapped=[(word,1-y) for word,y in tokens]\ndef describe(rows):\n    return dict(global_counts=dict(Counter(w for w,y in rows)),\n                top=dict(Counter(w for w,y in rows if y<.5)),\n                bottom=dict(Counter(w for w,y in rows if y>=.5)))\nbefore,after=describe(tokens),describe(swapped)\nprint('原布局',before,'\\n交换位置',after)\nassert before['global_counts']==after['global_counts']\nassert before['top']!=after['top']\nprint('HOG 配置维度：',(64//8-1)*(128//8-1)*2*2*9)\nprint('三层金字塔 K=3：',(1+4+16)*3)\nassert abs(.16927+.19241-.36168)<1e-10\n",
   "id": "06256d442549408687c38f2a20784a60",
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deepnote": {
     "block_id": "c718ba660a274141bb8fbf95489bb869",
     "type": "markdown",
     "source_block_id": null,
     "logical_key": null
    },
    "html_anchor": "cell-49"
   },
   "source": "**继续一步**：修改一个数值或条件，先写变化方向，再运行核对。不要删除断言来让代码“通过”；若主动改变了断言的前提，应同时写出新预期。机制小算例通过，只说明这一小段计算与预期一致，不等于完整图像实验或新架构训练成功。",
   "id": "c718ba660a274141bb8fbf95489bb869"
  }
 ]
}