{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": "# Lesson 36: Image Classification in Practice\n\nEvery classifier so far has been evaluated with a single number: overall accuracy. That number hides a lot. This lesson builds a 4-class classifier on a deliberately imbalanced subset of **CIFAR-10** — real photographs, not synthetic shapes — and shows why accuracy alone can be misleading, using the tools that reveal what's actually going wrong: the **confusion matrix**, and **per-class precision and recall**."
  },
  {
   "cell_type": "code",
   "id": "57fdea49",
   "source": "import pickle\nimport tarfile\nimport urllib.request\nfrom pathlib import Path\n\nimport numpy as np\nimport torch\nimport torch.nn as nn\nimport torch.nn.functional as F\nimport matplotlib.pyplot as plt",
   "metadata": {},
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "markdown",
   "id": "2733014c",
   "source": "## A 4-class, imbalanced dataset\n\nFour real photo categories from **CIFAR-10** (<a href=\"../references.html#krizhevsky-2009-cifar\">Krizhevsky, 2009</a>): cat, dog, automobile, and truck. The test set has an even 150 images per class, but the *training* set is deliberately starved of trucks (90 examples, versus 400 each for the other three) — a stand-in for the common real-world situation where some classes are just rarer to collect than others. Cat and dog are included deliberately as a visually similar pair, regardless of how much training data either gets.",
   "metadata": {}
  },
  {
   "cell_type": "code",
   "id": "94248cb5",
   "source": "CIFAR_URL = 'https://www.cs.toronto.edu/~kriz/cifar-10-python.tar.gz'\nCACHE_ROOT = Path.home() / '.cache' / 'cvintro'\nCACHE_DIR = CACHE_ROOT / 'cifar-10-batches-py'\n\ndef ensure_cifar10():\n    if CACHE_DIR.exists():\n        return\n    CACHE_ROOT.mkdir(parents=True, exist_ok=True)\n    archive_path = CACHE_ROOT / 'cifar-10-python.tar.gz'\n    if not archive_path.exists():\n        print('Downloading CIFAR-10 (~163 MB, one-time, cached under ~/.cache/cvintro)...')\n        urllib.request.urlretrieve(CIFAR_URL, archive_path)\n    print('Extracting...')\n    with tarfile.open(archive_path) as tar:\n        tar.extractall(CACHE_ROOT)\n\ndef load_cifar_batch(path):\n    with open(path, 'rb') as f:\n        d = pickle.load(f, encoding='bytes')\n    imgs = d[b'data'].reshape(-1, 3, 32, 32).transpose(0, 2, 3, 1).astype(np.float32) / 255.0\n    labels = np.array(d[b'labels'], dtype=np.int64)\n    return imgs, labels\n\nensure_cifar10()\n\nCIFAR_LABELS = ['airplane', 'automobile', 'bird', 'cat', 'deer', 'dog', 'frog', 'horse', 'ship', 'truck']\nCLASSES = ['cat', 'dog', 'automobile', 'truck']\nCLASS_IDS = {name: CIFAR_LABELS.index(name) for name in CLASSES}\n\ntrain_imgs, train_labels = [], []\nfor i in range(1, 6):\n    imgs, labels = load_cifar_batch(CACHE_DIR / f'data_batch_{i}')\n    train_imgs.append(imgs); train_labels.append(labels)\ntrain_imgs, train_labels = np.concatenate(train_imgs), np.concatenate(train_labels)\ntest_imgs, test_labels = load_cifar_batch(CACHE_DIR / 'test_batch')\n\nrng = np.random.default_rng(7)\n\ndef subset_for_classes(imgs, labels, per_class_counts):\n    out_imgs, out_labels = [], []\n    for i, name in enumerate(CLASSES):\n        idx = np.where(labels == CLASS_IDS[name])[0]\n        idx = rng.permutation(idx)[:per_class_counts[name]]\n        out_imgs.append(imgs[idx])\n        out_labels.append(np.full(len(idx), i, dtype=np.int64))\n    out_imgs, out_labels = np.concatenate(out_imgs), np.concatenate(out_labels)\n    perm = rng.permutation(len(out_imgs))\n    return out_imgs[perm], out_labels[perm]\n\n# training set: cat/dog/automobile well represented, truck deliberately starved\nX_train, y_train = subset_for_classes(train_imgs, train_labels,\n    {'cat': 400, 'dog': 400, 'automobile': 400, 'truck': 90})\n# test set: balanced across all four, so the imbalance is purely a training-time artifact\nX_test, y_test = subset_for_classes(test_imgs, test_labels,\n    {'cat': 150, 'dog': 150, 'automobile': 150, 'truck': 150})\n\nprint('training class counts:', dict(zip(CLASSES, np.bincount(y_train, minlength=4))))\nprint('test class counts:    ', dict(zip(CLASSES, np.bincount(y_test, minlength=4))))\n\nfig, axes = plt.subplots(1, 4, figsize=(8, 2.2))\nfor ax, name in zip(axes, CLASSES):\n    ax.imshow(X_train[y_train == CLASSES.index(name)][0])\n    ax.set_title(name, fontsize=9)\n    ax.axis('off')\nplt.show()",
   "metadata": {},
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "code",
   "id": "b2a478d6",
   "source": "class CNN(nn.Module):\n    def __init__(self):\n        super().__init__()\n        self.conv = nn.Sequential(\n            nn.Conv2d(3, 16, 5, padding=2), nn.ReLU(),\n            nn.MaxPool2d(2),\n            nn.Conv2d(16, 32, 5, padding=2), nn.ReLU(),\n            nn.MaxPool2d(2),\n            nn.Conv2d(32, 64, 3, padding=1), nn.ReLU(),\n            nn.AdaptiveMaxPool2d(1),\n        )\n        self.fc = nn.Linear(64, 4)\n\n    def forward(self, x):\n        return self.fc(self.conv(x).flatten(1))\n\ntorch.manual_seed(0)\nmodel = CNN()\nopt = torch.optim.Adam(model.parameters(), lr=0.0007)\nXt = torch.tensor(X_train).permute(0, 3, 1, 2); yt = torch.tensor(y_train)\nfor _ in range(150):\n    opt.zero_grad()\n    loss = F.cross_entropy(model(Xt), yt)\n    loss.backward()\n    opt.step()\n\nXte = torch.tensor(X_test).permute(0, 3, 1, 2)\nwith torch.no_grad():\n    preds = model(Xte).argmax(1).numpy()\n\nacc = (preds == y_test).mean()\nprint(f'overall test accuracy: {acc:.1%}')",
   "metadata": {},
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "markdown",
   "id": "26cf352b",
   "source": "Just over 50% overall accuracy, on a 4-class task where chance is 25%, sounds like a reasonable result for a small from-scratch CNN on real photos. It hides something important: the model is not equally good at all four classes.\n\n## Confusion matrix\n\nRow `i`, column `j` counts test examples of true class `i` predicted as class `j`. A perfect classifier is diagonal; everything off the diagonal is a specific, nameable mistake.",
   "metadata": {}
  },
  {
   "cell_type": "code",
   "id": "38bf8aa6",
   "source": "cm = np.zeros((4, 4), dtype=int)\nfor t, p in zip(y_test, preds):\n    cm[t, p] += 1\n\nprint('confusion matrix (rows=true, cols=predicted):')\nprint(f'{\"\":>10}' + ''.join(f'{s:>10}' for s in CLASSES))\nfor i, s in enumerate(CLASSES):\n    print(f'{s:>10}' + ''.join(f'{cm[i, j]:>10}' for j in range(4)))\n\nfig, ax = plt.subplots(figsize=(4.5, 4))\nim = ax.imshow(cm, cmap='Blues')\nax.set_xticks(range(4)); ax.set_xticklabels(CLASSES, rotation=45)\nax.set_yticks(range(4)); ax.set_yticklabels(CLASSES)\nax.set_xlabel('predicted'); ax.set_ylabel('true')\nfor i in range(4):\n    for j in range(4):\n        ax.text(j, i, cm[i, j], ha='center', va='center',\n                 color='white' if cm[i, j] > cm.max() / 2 else 'black')\nplt.title('Confusion matrix')\nplt.tight_layout()\nplt.show()",
   "metadata": {},
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "markdown",
   "id": "f44a0b1d",
   "source": "## True positives, false positives, true negatives, false negatives\n\nEvery cell of the confusion matrix above is a specific kind of correctness or mistake, but the standard vocabulary for talking about them is binary: pick one class and ask only \"is it this, or not?\" Collapsing the 4-class matrix down to \"truck vs. everything else\" gives exactly four outcomes:\n\n- **True positive (TP)**: actually a truck, predicted truck. A hit.\n- **False negative (FN)**: actually a truck, predicted something else. A miss — the model let it slip past.\n- **False positive (FP)**: actually *not* a truck, predicted truck anyway. A false alarm.\n- **True negative (TN)**: actually not a truck, correctly predicted not-truck.\n\nThis is the same 2x2 table underlying every binary classifier's evaluation (a medical test's \"positive/negative\" result, a spam filter's \"spam/not spam\" decision) — a multi-class confusion matrix is just this table computed once per class, with everything off that class's row/column collapsed into \"not this class.\"",
   "metadata": {}
  },
  {
   "cell_type": "code",
   "id": "06300321",
   "source": "cls = CLASSES.index('truck')\ntp = cm[cls, cls]\nfn = cm[cls, :].sum() - tp          # true truck, predicted something else\nfp = cm[:, cls].sum() - tp          # predicted truck, actually something else\ntn = cm.sum() - tp - fn - fp        # everything else, correctly not called truck\n\nprint(f'{\"\":>18}{\"predicted truck\":>20}{\"predicted NOT truck\":>24}')\nprint(f'{\"actually truck\":>18}{tp:>20}{fn:>24}')\nprint(f'{\"actually NOT truck\":>18}{fp:>20}{tn:>24}')\nprint()\nprint(f'TP={tp}, FP={fp}, FN={fn}, TN={tn}, total={tp+fp+fn+tn} (test set size={len(y_test)})')",
   "metadata": {},
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "markdown",
   "id": "f50b41fd",
   "source": "## Precision and recall\n\nTwo numbers per class, computed directly from TP/FP/FN:\n- **Recall** = TP / (TP + FN) — of everything that really *was* class `i`, what fraction did the model catch? Low recall means the model misses that class often.\n- **Precision** = TP / (TP + FP) — of everything the model *called* class `i`, what fraction actually was? Low precision means the model cries wolf on that class often.\n\n(A less commonly needed but related pair, built from the other two quadrants: **specificity** = TN / (TN + FP), how well the model avoids false alarms on the negative class, and its complement the **false positive rate** = FP / (FP + TN) = 1 − specificity.)",
   "metadata": {}
  },
  {
   "cell_type": "code",
   "id": "fccace41",
   "source": "print(f'{\"class\":>10} {\"precision\":>10} {\"recall\":>8} {\"support\":>8}')\nfor i, name in enumerate(CLASSES):\n    tp = cm[i, i]\n    fn = cm[i, :].sum() - tp\n    fp = cm[:, i].sum() - tp\n    precision = tp / (tp + fp) if (tp + fp) > 0 else float('nan')\n    recall = tp / (tp + fn) if (tp + fn) > 0 else float('nan')\n    support = cm[i, :].sum()\n    print(f'{name:>10} {precision:>10.2f} {recall:>8.2f} {support:>8}')",
   "metadata": {},
   "execution_count": null,
   "outputs": []
  },
  {
   "cell_type": "markdown",
   "id": "171afc8c",
   "source": "Truck — the class starved to 90 training images, versus 400 for each other class — has high precision but very low recall: when the model does say \"truck,\" it's usually right, but it fails to recognize the overwhelming majority of actual trucks, defaulting instead to whichever classes it saw plenty of during training (mostly \"automobile,\" the visually closest well-represented class). That is the standard signature of class imbalance, and it is completely invisible in the single overall-accuracy number from before. Cat and dog, by contrast, are both well-represented in training but get confused *with each other* far more than with automobile or truck — a different failure mode entirely, caused by genuine visual similarity between the two animals rather than by a lack of data. Automobile ends up with unusually high recall but only middling precision, for the same reason truck's recall suffered: it's absorbing the starved truck class's misclassifications, since \"wheeled vehicle\" is an easy fallback guess once the model can't tell trucks apart from automobiles.\n\nThe practical lesson: always inspect the confusion matrix and per-class metrics before trusting a single accuracy figure, especially on any dataset where classes aren't naturally balanced.\n\n### Exercises\n\n1. Increase the truck training count from 90 to 400 (matching the other three classes) and rerun. Does truck recall recover? Does the cat/dog confusion also improve, or does it persist — and why would more data not fix a problem caused by genuine visual similarity rather than data scarcity?\n2. Try weighting the loss by inverse class frequency (`F.cross_entropy(logits, yt, weight=class_weights)`, where `class_weights[i] = 1 / count(class i)`) instead of collecting more truck images. Does it recover truck recall, and at what cost to the other classes' precision?\n3. Compute the **F1 score** (the harmonic mean of precision and recall, `2 * p * r / (p + r)`) for each class. Why might F1 be a better single number to track per-class than accuracy, when accuracy is only meaningful in aggregate?",
   "metadata": {}
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "name": "python",
   "version": "3.10.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}