{ "cells": [ { "cell_type": "markdown", "id": "f3719b66", "metadata": {}, "source": [ "# Handwriting Rating" ] }, { "cell_type": "code", "execution_count": 1, "id": "433c83fb", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "2.0.1\n", "0.15.2\n" ] } ], "source": [ "import torch\n", "\n", "import torchvision\n", "from torchvision import datasets\n", "from torchvision import transforms\n", "from torchvision.transforms import ToTensor\n", "\n", "import matplotlib.pyplot as plt\n", "\n", "print(torch.__version__)\n", "print(torchvision.__version__)" ] }, { "cell_type": "code", "execution_count": 2, "id": "b95dffec", "metadata": {}, "outputs": [ { "data": { "text/plain": [ "'cuda'" ] }, "execution_count": 2, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# Set device to GPU if available\n", "device = \"cuda\" if torch.cuda.is_available() else \"cpu\"\n", "device" ] }, { "cell_type": "markdown", "id": "c8c3d734", "metadata": {}, "source": [ "## 1. Getting the dataset" ] }, { "cell_type": "code", "execution_count": 3, "id": "7f5df511", "metadata": {}, "outputs": [], "source": [ "# Download training data locally\n", "train_data = datasets.MNIST(\n", " root=\"data\",\n", " train=True,\n", " download=True,\n", " transform=torchvision.transforms.ToTensor(),\n", " target_transform=None)\n", "\n", "# Download testing data locally\n", "test_data = datasets.MNIST(\n", " root=\"data\",\n", " train=False,\n", " download=True,\n", " transform=torchvision.transforms.ToTensor(),\n", " target_transform=None)" ] }, { "cell_type": "code", "execution_count": 4, "id": "1f0b6925", "metadata": {}, "outputs": [ { "data": { "text/plain": [ "(60000, 10000)" ] }, "execution_count": 4, "metadata": {}, "output_type": "execute_result" } ], "source": [ "len(train_data), len(test_data)" ] }, { "cell_type": "code", "execution_count": 5, "id": "23fe862a", "metadata": {}, "outputs": [ { "data": { "text/plain": [ "(tensor([[[0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000],\n", " [0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000],\n", " [0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000],\n", " [0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000],\n", " [0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000],\n", " [0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000, 0.0118, 0.0706, 0.0706, 0.0706,\n", " 0.4941, 0.5333, 0.6863, 0.1020, 0.6510, 1.0000, 0.9686, 0.4980,\n", " 0.0000, 0.0000, 0.0000, 0.0000],\n", " [0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.1176, 0.1412, 0.3686, 0.6039, 0.6667, 0.9922, 0.9922, 0.9922,\n", " 0.9922, 0.9922, 0.8824, 0.6745, 0.9922, 0.9490, 0.7647, 0.2510,\n", " 0.0000, 0.0000, 0.0000, 0.0000],\n", " [0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.1922,\n", " 0.9333, 0.9922, 0.9922, 0.9922, 0.9922, 0.9922, 0.9922, 0.9922,\n", " 0.9922, 0.9843, 0.3647, 0.3216, 0.3216, 0.2196, 0.1529, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000],\n", " [0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0706,\n", " 0.8588, 0.9922, 0.9922, 0.9922, 0.9922, 0.9922, 0.7765, 0.7137,\n", " 0.9686, 0.9451, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000],\n", " [0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.3137, 0.6118, 0.4196, 0.9922, 0.9922, 0.8039, 0.0431, 0.0000,\n", " 0.1686, 0.6039, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000],\n", " [0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0549, 0.0039, 0.6039, 0.9922, 0.3529, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000],\n", " [0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.5451, 0.9922, 0.7451, 0.0078, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000],\n", " [0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0431, 0.7451, 0.9922, 0.2745, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000],\n", " [0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000, 0.1373, 0.9451, 0.8824, 0.6275,\n", " 0.4235, 0.0039, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000],\n", " [0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.3176, 0.9412, 0.9922,\n", " 0.9922, 0.4667, 0.0980, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000],\n", " [0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.1765, 0.7294,\n", " 0.9922, 0.9922, 0.5882, 0.1059, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000],\n", " [0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0627,\n", " 0.3647, 0.9882, 0.9922, 0.7333, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000],\n", " [0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.9765, 0.9922, 0.9765, 0.2510, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000],\n", " [0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.1804, 0.5098,\n", " 0.7176, 0.9922, 0.9922, 0.8118, 0.0078, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000],\n", " [0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000, 0.1529, 0.5804, 0.8980, 0.9922,\n", " 0.9922, 0.9922, 0.9804, 0.7137, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000],\n", " [0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0941, 0.4471, 0.8667, 0.9922, 0.9922, 0.9922,\n", " 0.9922, 0.7882, 0.3059, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000],\n", " [0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0902, 0.2588, 0.8353, 0.9922, 0.9922, 0.9922, 0.9922, 0.7765,\n", " 0.3176, 0.0078, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000],\n", " [0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0706, 0.6706,\n", " 0.8588, 0.9922, 0.9922, 0.9922, 0.9922, 0.7647, 0.3137, 0.0353,\n", " 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000],\n", " [0.0000, 0.0000, 0.0000, 0.0000, 0.2157, 0.6745, 0.8863, 0.9922,\n", " 0.9922, 0.9922, 0.9922, 0.9569, 0.5216, 0.0431, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000],\n", " [0.0000, 0.0000, 0.0000, 0.0000, 0.5333, 0.9922, 0.9922, 0.9922,\n", " 0.8314, 0.5294, 0.5176, 0.0627, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000],\n", " [0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000],\n", " [0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000],\n", " [0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,\n", " 0.0000, 0.0000, 0.0000, 0.0000]]]),\n", " 5)" ] }, "execution_count": 5, "metadata": {}, "output_type": "execute_result" } ], "source": [ "image, label = train_data[0]\n", "image, label" ] }, { "cell_type": "code", "execution_count": 6, "id": "ca50965b", "metadata": {}, "outputs": [ { "data": { "text/plain": [ "['0 - zero',\n", " '1 - one',\n", " '2 - two',\n", " '3 - three',\n", " '4 - four',\n", " '5 - five',\n", " '6 - six',\n", " '7 - seven',\n", " '8 - eight',\n", " '9 - nine']" ] }, "execution_count": 6, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# Create a list of class names\n", "class_names = train_data.classes\n", "class_names" ] }, { "cell_type": "code", "execution_count": 7, "id": "6b09856c", "metadata": {}, "outputs": [ { "data": { "text/plain": [ "tensor([5, 0, 4, ..., 5, 6, 8])" ] }, "execution_count": 7, "metadata": {}, "output_type": "execute_result" } ], "source": [ "train_data.targets" ] }, { "cell_type": "code", "execution_count": 8, "id": "75006520", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Image shape: torch.Size([1, 28, 28])\n", "Image label: 5 - five\n" ] } ], "source": [ "# See information from random sample from dataset\n", "print(f\"Image shape: {image.shape}\")\n", "print(f\"Image label: {class_names[label]}\")" ] }, { "cell_type": "code", "execution_count": 9, "id": "8651cc07", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Image shape: torch.Size([1, 28, 28])\n" ] }, { "data": { "text/plain": [ "Text(0.5, 1.0, '5')" ] }, "execution_count": 9, "metadata": {}, "output_type": "execute_result" }, { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "# Visualize color sample from dataset\n", "print(f\"Image shape: {image.shape}\")\n", "plt.imshow(image.squeeze())\n", "plt.title(label)" ] }, { "cell_type": "code", "execution_count": 10, "id": "7bf0cae4", "metadata": {}, "outputs": [ { "data": { "text/plain": [ "(-0.5, 27.5, 27.5, -0.5)" ] }, "execution_count": 10, "metadata": {}, "output_type": "execute_result" }, { "data": { "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAGbCAYAAAAr/4yjAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjcuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8pXeV/AAAACXBIWXMAAA9hAAAPYQGoP6dpAAAN/0lEQVR4nO3cWYjVdR/H8e8ZtUYtFcsWwpRR0hbLizSYbDGTKNI0RQsql2iBIm8qWi7CwhJyAS3IpASlQgvTisoCtcJIFMsbK4IIKoRWc8nGdM5z8zxfHqlofqcZ52ivF3gx4/n4+ys4b/6O51+pVqvVAICIaOjsCwCgfogCAEkUAEiiAEASBQCSKACQRAGAJAoAJFEAIIkC/Nfbb78dw4cPj8bGxqhUKrFr166YPn16DBw4sLMvDY4YUaBubdy4MSqVyp/++Oijj9r1rB9//DGmTJkS3bt3j6effjpWrFgRPXv2bNcz4GjQtbMvAP7OPffcEyNGjDjsc4MHD27XM7Zs2RJ79uyJxx57LK688sr8/NKlS6O1tbVdz4J6JgrUvUsuuSQmT57coWd89913ERHRp0+fwz7frVu3Dj0X6o1/PuKosGfPnjh48GCH/NqXX355TJs2LSIiRowYEZVKJaZPnx4Rcdj3FH7//ffo27dvzJgx4w+/xu7du6OxsTHuvffe/FxLS0s88sgjMXjw4Dj++OOjf//+cf/990dLS0uH/D6gPYgCdW/GjBnRq1evaGxsjNGjR8fWrVvb9dd/+OGH4/bbb4+IiEcffTRWrFgRd9xxxx9e161bt5g4cWKsWbMmDhw4cNjPrVmzJlpaWuKGG26IiIjW1tYYP358zJs3L8aNGxeLFy+OCRMmxMKFC2Pq1Kntev3QrqpQpzZt2lSdNGlS9bnnnquuXbu2+sQTT1RPOumkamNjY3Xbtm3tetayZcuqEVHdsmXLYZ+fNm1adcCAAfnxunXrqhFRff311w973TXXXFNtamrKj1esWFFtaGiofvDBB4e97plnnqlGRHXTpk3tev3QXtwpULeam5vjlVdeiZkzZ8b48ePjgQceiI8++igqlUo8+OCDnXJNV1xxRZx88smxcuXK/NzPP/8c77777mF3AC+//HKcffbZMXTo0Pjhhx/yxxVXXBERERs2bDji1w5t4RvNHFUGDx4c1113XaxevToOHToUXbp0+dPX7d27N/bu3Zsfd+nSJfr16/ePz+/atWtMmjQpXnzxxWhpaYnjjz8+Vq9eHb///vthUfjiiy/i008//csz//eNbag3osBRp3///nHgwIHYt29f9OrV609fM2/evJg9e3Z+PGDAgPjqq6/a5fwbbrghlixZEm+99VZMmDAhVq1aFUOHDo0LLrggX9Pa2hrDhg2LBQsW/OXvAeqRKHDU+fLLL6OxsTFOOOGEv3zNLbfcEqNGjcqPu3fv3m7nX3rppXH66afHypUrY9SoUbF+/fp4+OGHD3vNoEGDYvv27TFmzJioVCrtdjZ0NFGgbn3//fd/+OeX7du3x2uvvRZXX311NDT89bfEmpqaoqmpqUOuq6GhISZPnhzPP/98jBw5Mg4ePPiH/1E0ZcqUePPNN2Pp0qX5P5v+Z//+/dHa2uod09QlUaBuTZ06Nbp37x7Nzc1xyimnxI4dO+LZZ5+NHj16xNy5czv92hYvXhyPPPJIDBs2LM4+++zDfv7mm2+OVatWxZ133hkbNmyIiy++OA4dOhSfffZZrFq1KtatWxcXXnhhJ109/DVRoG5NmDAhXnjhhViwYEHs3r07+vXrF9dff32+IawzNTc3R//+/ePrr7/+0/cdNDQ0xJo1a2LhwoWxfPnyePXVV6NHjx7R1NQUs2bNirPOOqsTrhr+XqVarVY7+yIAqA/epwBAEgUAkigAkEQBgCQKACRRACC1+X0K3qoPcHRryzsQ3CkAkEQBgCQKACRRACCJAgBJFABIogBAEgUAkigAkEQBgCQKACRRACCJAgBJFABIogBAEgUAkigAkEQBgCQKACRRACCJAgBJFABIogBAEgUAkigAkEQBgCQKACRRACCJAgBJFABIogBAEgUAkigAkEQBgCQKACRRACCJAgBJFABIogBAEgUAkigAkEQBgCQKACRRACCJAgBJFABIogBAEgUAkigAkEQBgCQKACRRACCJAgBJFABIogBAEgUAkigAkEQBgCQKACRRACCJAgBJFABIogBAEgUAkigAkLp29gXA3+nSpUvxpnfv3h1wJe3j7rvvrmnXo0eP4s2QIUOKN3fddVfxZt68ecWbG2+8sXgTEfHbb78Vb+bOnVu8mT17dvHmWOBOAYAkCgAkUQAgiQIASRQASKIAQBIFAJIoAJBEAYAkCgAkUQAgiQIAyQPxjjFnnnlm8ea4444r3jQ3NxdvRo0aVbyJiOjTp0/xZtKkSTWddaz55ptvijeLFi0q3kycOLF4s2fPnuJNRMT27duLN++9915NZ/0buVMAIIkCAEkUAEiiAEASBQCSKACQRAGAJAoAJFEAIIkCAEkUAEiiAECqVKvVapteWKl09LXwf4YPH17Tbv369cWb3r1713QWR1Zra2vxZubMmcWbvXv3Fm9qsXPnzpp2P//8c/Hm888/r+msY01bvty7UwAgiQIASRQASKIAQBIFAJIoAJBEAYAkCgAkUQAgiQIASRQASKIAQBIFAJKnpNapvn371rTbvHlz8aapqamms441tfzZ7dq1q3gzevTo4k1ExIEDB4o3noDL//OUVACKiAIASRQASKIAQBIFAJIoAJBEAYAkCgAkUQAgiQIASRQASKIAQOra2RfAn/vpp59q2t13333Fm2uvvbZ48/HHHxdvFi1aVLyp1SeffFK8GTt2bPFm3759xZtzzz23eBMRMWvWrJp2UMKdAgBJFABIogBAEgUAkigAkEQBgCQKACRRACCJAgBJFABIogBAEgUAUqVarVbb9MJKpaOvhU7Sq1ev4s2ePXuKN0uWLCneRETceuutxZubbrqpePPSSy8Vb+Bo0pYv9+4UAEiiAEASBQCSKACQRAGAJAoAJFEAIIkCAEkUAEiiAEASBQCSKACQunb2BdD5du/efUTO+eWXX47IORERt912W/Fm5cqVxZvW1tbiDdQzdwoAJFEAIIkCAEkUAEiiAEASBQCSKACQRAGAJAoAJFEAIIkCAEkUAEiiAECqVKvVapteWKl09LVwjOvZs2dNu9dff714c9lllxVvrr766uLNO++8U7yBztKWL/fuFABIogBAEgUAkigAkEQBgCQKACRRACCJAgBJFABIogBAEgUAkigAkDwQj7o3aNCg4s22bduKN7t27SrebNiwoXizdevW4k1ExNNPP128aeNfb/4lPBAPgCKiAEASBQCSKACQRAGAJAoAJFEAIIkCAEkUAEiiAEASBQCSKACQPBCPY9LEiROLN8uWLSvenHjiicWbWj300EPFm+XLlxdvdu7cWbzh6OCBeAAUEQUAkigAkEQBgCQKACRRACCJAgBJFABIogBAEgUAkigAkEQBgOSBePBf5513XvFmwYIFxZsxY8YUb2q1ZMmS4s2cOXOKN99++23xhiPPA/EAKCIKACRRACCJAgBJFABIogBAEgUAkigAkEQBgCQKACRRACCJAgDJA/HgH+jTp0/xZty4cTWdtWzZsuJNLX9v169fX7wZO3Zs8YYjzwPxACgiCgAkUQAgiQIASRQASKIAQBIFAJIoAJBEAYAkCgAkUQAgiQIASRQASJ6SCkeJlpaW4k3Xrl2LNwcPHizeXHXVVcWbjRs3Fm/4ZzwlFYAiogBAEgUAkigAkEQBgCQKACRRACCJAgBJFABIogBAEgUAkigAkMqflgXHqPPPP794M3ny5OLNiBEjijcRtT3crhY7duwo3rz//vsdcCV0BncKACRRACCJAgBJFABIogBAEgUAkigAkEQBgCQKACRRACCJAgBJFABIHohH3RsyZEjx5u677y7eXH/99cWb0047rXhzJB06dKh4s3PnzuJNa2tr8Yb65E4BgCQKACRRACCJAgBJFABIogBAEgUAkigAkEQBgCQKACRRACCJAgDJA/GoSS0PgrvxxhtrOquWh9sNHDiwprPq2datW4s3c+bMKd689tprxRuOHe4UAEiiAEASBQCSKACQRAGAJAoAJFEAIIkCAEkUAEiiAEASBQCSKACQPBDvGHPqqacWb84555zizVNPPVW8GTp0aPGm3m3evLl48+STT9Z01tq1a4s3ra2tNZ3Fv5c7BQCSKACQRAGAJAoAJFEAIIkCAEkUAEiiAEASBQCSKACQRAGAJAoAJFEAIHlK6hHQt2/f4s2SJUtqOmv48OHFm6ampprOqmcffvhh8Wb+/PnFm3Xr1hVv9u/fX7yBI8WdAgBJFABIogBAEgUAkigAkEQBgCQKACRRACCJAgBJFABIogBAEgUA0r/6gXgXXXRR8ea+++4r3owcObJ4c8YZZxRv6t2vv/5a027RokXFm8cff7x4s2/fvuINHGvcKQCQRAGAJAoAJFEAIIkCAEkUAEiiAEASBQCSKACQRAGAJAoAJFEAIP2rH4g3ceLEI7I5knbs2FG8eeONN4o3Bw8eLN7Mnz+/eBMRsWvXrpp2QDl3CgAkUQAgiQIASRQASKIAQBIFAJIoAJBEAYAkCgAkUQAgiQIASRQASJVqtVpt0wsrlY6+FgA6UFu+3LtTACCJAgBJFABIogBAEgUAkigAkEQBgCQKACRRACCJAgBJFABIogBAEgUAkigAkEQBgCQKACRRACCJAgBJFABIogBAEgUAkigAkEQBgCQKACRRACCJAgBJFABIogBAEgUAkigAkEQBgCQKACRRACCJAgBJFABIogBAEgUAkigAkEQBgNS1rS+sVqsdeR0A1AF3CgAkUQAgiQIASRQASKIAQBIFAJIoAJBEAYAkCgCk/wC69UjAsc8powAAAABJRU5ErkJggg==", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "# Visualize a grayscale sample from dataset\n", "plt.imshow(image.squeeze(), cmap=\"gray\")\n", "plt.title(class_names[label])\n", "plt.axis(\"off\")" ] }, { "cell_type": "code", "execution_count": 11, "id": "6b2f1ac8", "metadata": {}, "outputs": [ { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "# Visualize a random sample from dataset\n", "fig = plt.figure(figsize=(9,9))\n", "rows, cols = 4, 4\n", "for i in range(1, rows*cols+1):\n", " random_idx = torch.randint(0, len(train_data), size=[1]).item()\n", " img, label = train_data[random_idx]\n", " fig.add_subplot(rows, cols, i)\n", " plt.imshow(img.squeeze(), cmap=\"gray\")\n", " plt.title(class_names[label])\n", " plt.axis(\"off\")" ] }, { "cell_type": "markdown", "id": "2f075ea4", "metadata": {}, "source": [ "## 2. DataLoader" ] }, { "cell_type": "code", "execution_count": 12, "id": "8b5d0667", "metadata": {}, "outputs": [], "source": [ "from torch.utils.data import DataLoader" ] }, { "cell_type": "code", "execution_count": 13, "id": "f65e026d", "metadata": {}, "outputs": [], "source": [ "# Set hyperparameters\n", "BATCH_SIZE = 32\n", "NUM_WORKERS = 0\n", "\n", "# Train dataloader\n", "train_dataloader = DataLoader(\n", " dataset=train_data,\n", " batch_size=BATCH_SIZE,\n", " shuffle=True,\n", " num_workers=NUM_WORKERS)\n", "\n", "# Test dataloder\n", "test_dataloader = DataLoader(\n", " dataset=test_data,\n", " batch_size=BATCH_SIZE,\n", " shuffle=False,\n", " num_workers=NUM_WORKERS)" ] }, { "cell_type": "code", "execution_count": 14, "id": "eb83508c", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Length of train_dataloader: 1875 batches of 32\n", "Length of test_dataloader: 313 batches of 32\n" ] } ], "source": [ "# Dataloader information\n", "print(f\"Length of train_dataloader: {len(train_dataloader)} batches of {BATCH_SIZE}\")\n", "print(f\"Length of test_dataloader: {len(test_dataloader)} batches of {BATCH_SIZE}\")" ] }, { "cell_type": "code", "execution_count": 15, "id": "d0c7922c", "metadata": {}, "outputs": [ { "data": { "text/plain": [ "(torch.Size([32, 1, 28, 28]), torch.Size([32]))" ] }, "execution_count": 15, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# Sample the dataloader\n", "train_features_batch, train_labels_batch = next(iter(train_dataloader))\n", "train_features_batch.shape, train_labels_batch.shape" ] }, { "cell_type": "code", "execution_count": 16, "id": "6f2d5799", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Image size: torch.Size([1, 28, 28])\n", "Label: 1\n" ] }, { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "# Print random sample from batch\n", "random_idx = torch.randint(0, len(train_features_batch), size=[1]).item()\n", "img, label = train_features_batch[random_idx], train_labels_batch[random_idx]\n", "plt.imshow(img.squeeze(), cmap=\"gray\")\n", "plt.title(class_names[label])\n", "plt.axis(\"off\")\n", "print(f\"Image size: {img.shape}\")\n", "print(f\"Label: {label}\")" ] }, { "cell_type": "markdown", "id": "39b02aff", "metadata": {}, "source": [ "# 3. Models" ] }, { "cell_type": "markdown", "id": "6296f954", "metadata": {}, "source": [ "### 3.1 Model 1\n", "\n", "Simple model with only linear layers" ] }, { "cell_type": "code", "execution_count": 17, "id": "78f67457", "metadata": {}, "outputs": [], "source": [ "from torch import nn\n", "from torchinfo import summary" ] }, { "cell_type": "code", "execution_count": 18, "id": "34463fd6", "metadata": {}, "outputs": [], "source": [ "class MNISTModelv1(nn.Module):\n", " def __init__(self,\n", " input_shape: int,\n", " hidden_units: int,\n", " output_shape: int):\n", " super().__init__()\n", " self.layer_stack = nn.Sequential(\n", " nn.Flatten(),\n", " nn.Linear(in_features=input_shape,\n", " out_features=hidden_units),\n", " nn.Linear(in_features=hidden_units,\n", " out_features=output_shape)\n", " )\n", " \n", " def forward(self, x):\n", " return self.layer_stack(x)" ] }, { "cell_type": "code", "execution_count": 19, "id": "ec92bf9a", "metadata": {}, "outputs": [], "source": [ "# Initiate model_!\n", "model_1 = MNISTModelv1(\n", " input_shape=784,\n", " hidden_units=10,\n", " output_shape=len(class_names)\n", ").to(device)" ] }, { "cell_type": "code", "execution_count": 20, "id": "4fc51cd1", "metadata": {}, "outputs": [ { "data": { "text/plain": [ "==========================================================================================\n", "Layer (type:depth-idx) Output Shape Param #\n", "==========================================================================================\n", "MNISTModelv1 [1, 10] --\n", "├─Sequential: 1-1 [1, 10] --\n", "│ └─Flatten: 2-1 [1, 784] --\n", "│ └─Linear: 2-2 [1, 10] 7,850\n", "│ └─Linear: 2-3 [1, 10] 110\n", "==========================================================================================\n", "Total params: 7,960\n", "Trainable params: 7,960\n", "Non-trainable params: 0\n", "Total mult-adds (M): 0.01\n", "==========================================================================================\n", "Input size (MB): 0.00\n", "Forward/backward pass size (MB): 0.00\n", "Params size (MB): 0.03\n", "Estimated Total Size (MB): 0.04\n", "==========================================================================================" ] }, "execution_count": 20, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# Visualize model_1\n", "summary(model_1, input_size=[1, 1, 28, 28])" ] }, { "cell_type": "markdown", "id": "f4ea42d4", "metadata": {}, "source": [ "### 3.2 Model 2\n", "\n", "More complex model based on the TinyVGG architecture" ] }, { "cell_type": "code", "execution_count": 21, "id": "acfd64e4", "metadata": {}, "outputs": [], "source": [ "class MNISTModelv2(nn.Module):\n", " def __init__(self,\n", " input_shape: int,\n", " hidden_units: int,\n", " output_shape: int) -> None:\n", " super().__init__()\n", " self.conv_block_1 = nn.Sequential(\n", " nn.Conv2d(in_channels=input_shape,\n", " out_channels=hidden_units,\n", " kernel_size=3,\n", " stride=1,\n", " padding=0),\n", " nn.ReLU(),\n", " nn.Conv2d(in_channels=hidden_units,\n", " out_channels=hidden_units,\n", " kernel_size=3,\n", " stride=1,\n", " padding=0),\n", " nn.ReLU(),\n", " nn.MaxPool2d(kernel_size=2,\n", " stride=2)\n", " )\n", " self.conv_block_2 = nn.Sequential(\n", " nn.Conv2d(in_channels=hidden_units,\n", " out_channels=hidden_units,\n", " kernel_size=3,\n", " stride=1,\n", " padding=0),\n", " nn.ReLU(),\n", " nn.Conv2d(in_channels=hidden_units,\n", " out_channels=hidden_units,\n", " kernel_size=3,\n", " stride=1,\n", " padding=0),\n", " nn.ReLU(),\n", " nn.MaxPool2d(kernel_size=2,\n", " stride=2)\n", " )\n", " self.classifier = nn.Sequential(\n", " nn.Flatten(),\n", " nn.Linear(in_features=hidden_units*4*4,\n", " out_features=output_shape)\n", " )\n", " \n", " def forward(self, x):\n", " return self.classifier(self.conv_block_2(self.conv_block_1(x)))" ] }, { "cell_type": "code", "execution_count": 22, "id": "a31f7ae7", "metadata": {}, "outputs": [], "source": [ "# Initiate model_2\n", "model_2 = MNISTModelv2(\n", " input_shape=1,\n", " hidden_units=10,\n", " output_shape=len(class_names)\n", ").to(device)" ] }, { "cell_type": "code", "execution_count": 23, "id": "d50a42d8", "metadata": {}, "outputs": [ { "data": { "text/plain": [ "==========================================================================================\n", "Layer (type:depth-idx) Output Shape Param #\n", "==========================================================================================\n", "MNISTModelv2 [1, 10] --\n", "├─Sequential: 1-1 [1, 10, 12, 12] --\n", "│ └─Conv2d: 2-1 [1, 10, 26, 26] 100\n", "│ └─ReLU: 2-2 [1, 10, 26, 26] --\n", "│ └─Conv2d: 2-3 [1, 10, 24, 24] 910\n", "│ └─ReLU: 2-4 [1, 10, 24, 24] --\n", "│ └─MaxPool2d: 2-5 [1, 10, 12, 12] --\n", "├─Sequential: 1-2 [1, 10, 4, 4] --\n", "│ └─Conv2d: 2-6 [1, 10, 10, 10] 910\n", "│ └─ReLU: 2-7 [1, 10, 10, 10] --\n", "│ └─Conv2d: 2-8 [1, 10, 8, 8] 910\n", "│ └─ReLU: 2-9 [1, 10, 8, 8] --\n", "│ └─MaxPool2d: 2-10 [1, 10, 4, 4] --\n", "├─Sequential: 1-3 [1, 10] --\n", "│ └─Flatten: 2-11 [1, 160] --\n", "│ └─Linear: 2-12 [1, 10] 1,610\n", "==========================================================================================\n", "Total params: 4,440\n", "Trainable params: 4,440\n", "Non-trainable params: 0\n", "Total mult-adds (M): 0.74\n", "==========================================================================================\n", "Input size (MB): 0.00\n", "Forward/backward pass size (MB): 0.11\n", "Params size (MB): 0.02\n", "Estimated Total Size (MB): 0.13\n", "==========================================================================================" ] }, "execution_count": 23, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# Visualize model_2\n", "summary(model_2, input_size=[1, 1, 28, 28])" ] }, { "cell_type": "markdown", "id": "5334fd03", "metadata": {}, "source": [ "### 3.3 Model 3\n", "\n", "Even more complex model based on the VGG16 architecture" ] }, { "cell_type": "code", "execution_count": 24, "id": "02f6d149", "metadata": {}, "outputs": [], "source": [ "class MNISTModelv3(nn.Module):\n", " def __init__(self,\n", " input_shape: int,\n", " output_shape: int) -> None:\n", " super().__init__()\n", "\n", " self.conv_block_1 = nn.Sequential(\n", " nn.Conv2d(in_channels=input_shape, \n", " out_channels=64, \n", " kernel_size=3, \n", " padding=1),\n", " nn.ReLU(),\n", " nn.Conv2d(in_channels=64, \n", " out_channels=64, \n", " kernel_size=3, \n", " padding=1),\n", " nn.ReLU(),\n", " nn.MaxPool2d(kernel_size=2, \n", " stride=2),\n", " )\n", " self.conv_block_2 = nn.Sequential(\n", " nn.Conv2d(in_channels=64, \n", " out_channels=128, \n", " kernel_size=3, \n", " padding=1),\n", " nn.ReLU(),\n", " nn.Conv2d(in_channels=128, \n", " out_channels=128, \n", " kernel_size=3, \n", " padding=1),\n", " nn.ReLU(),\n", " nn.MaxPool2d(kernel_size=2, \n", " stride=2),\n", " )\n", " self.conv_block_3 = nn.Sequential(\n", " nn.Conv2d(in_channels=128, \n", " out_channels=256, \n", " kernel_size=3, \n", " padding=1),\n", " nn.ReLU(),\n", " nn.Conv2d(in_channels=256, \n", " out_channels=256, \n", " kernel_size=3, \n", " padding=1),\n", " nn.ReLU(),\n", " nn.MaxPool2d(kernel_size=2, \n", " stride=2)\n", " )\n", " self.classifier = nn.Sequential(\n", " nn.Flatten(),\n", " nn.Linear(in_features=256*3*3, \n", " out_features=4096),\n", " nn.ReLU(),\n", " nn.Dropout(),\n", " nn.Linear(in_features=4096, \n", " out_features=4096),\n", " nn.ReLU(),\n", " nn.Dropout(),\n", " nn.Linear(in_features=4096, \n", " out_features=output_shape)\n", " )\n", " \n", " def forward(self, x):\n", " x = self.classifier(self.conv_block_3(self.conv_block_2(self.conv_block_1(x))))\n", " return x" ] }, { "cell_type": "code", "execution_count": 25, "id": "89e7af8b", "metadata": {}, "outputs": [], "source": [ "# Initiate model_3\n", "model_3 = MNISTModelv3(\n", " input_shape=1,\n", " output_shape=len(class_names)\n", ").to(device)" ] }, { "cell_type": "code", "execution_count": 26, "id": "2a044d68", "metadata": {}, "outputs": [ { "data": { "text/plain": [ "==========================================================================================\n", "Layer (type:depth-idx) Output Shape Param #\n", "==========================================================================================\n", "MNISTModelv3 [1, 10] --\n", "├─Sequential: 1-1 [1, 64, 14, 14] --\n", "│ └─Conv2d: 2-1 [1, 64, 28, 28] 640\n", "│ └─ReLU: 2-2 [1, 64, 28, 28] --\n", "│ └─Conv2d: 2-3 [1, 64, 28, 28] 36,928\n", "│ └─ReLU: 2-4 [1, 64, 28, 28] --\n", "│ └─MaxPool2d: 2-5 [1, 64, 14, 14] --\n", "├─Sequential: 1-2 [1, 128, 7, 7] --\n", "│ └─Conv2d: 2-6 [1, 128, 14, 14] 73,856\n", "│ └─ReLU: 2-7 [1, 128, 14, 14] --\n", "│ └─Conv2d: 2-8 [1, 128, 14, 14] 147,584\n", "│ └─ReLU: 2-9 [1, 128, 14, 14] --\n", "│ └─MaxPool2d: 2-10 [1, 128, 7, 7] --\n", "├─Sequential: 1-3 [1, 256, 3, 3] --\n", "│ └─Conv2d: 2-11 [1, 256, 7, 7] 295,168\n", "│ └─ReLU: 2-12 [1, 256, 7, 7] --\n", "│ └─Conv2d: 2-13 [1, 256, 7, 7] 590,080\n", "│ └─ReLU: 2-14 [1, 256, 7, 7] --\n", "│ └─MaxPool2d: 2-15 [1, 256, 3, 3] --\n", "├─Sequential: 1-4 [1, 10] --\n", "│ └─Flatten: 2-16 [1, 2304] --\n", "│ └─Linear: 2-17 [1, 4096] 9,441,280\n", "│ └─ReLU: 2-18 [1, 4096] --\n", "│ └─Dropout: 2-19 [1, 4096] --\n", "│ └─Linear: 2-20 [1, 4096] 16,781,312\n", "│ └─ReLU: 2-21 [1, 4096] --\n", "│ └─Dropout: 2-22 [1, 4096] --\n", "│ └─Linear: 2-23 [1, 10] 40,970\n", "==========================================================================================\n", "Total params: 27,407,818\n", "Trainable params: 27,407,818\n", "Non-trainable params: 0\n", "Total mult-adds (M): 142.50\n", "==========================================================================================\n", "Input size (MB): 0.00\n", "Forward/backward pass size (MB): 1.47\n", "Params size (MB): 109.63\n", "Estimated Total Size (MB): 111.11\n", "==========================================================================================" ] }, "execution_count": 26, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# Visualize model_3\n", "summary(model_3, input_size=[1, 1, 28, 28])" ] }, { "cell_type": "markdown", "id": "d30583ad", "metadata": {}, "source": [ "## 4. Model evaluation functions and setup" ] }, { "cell_type": "markdown", "id": "94f29b94", "metadata": {}, "source": [ "### 4.1 Loss, Optimizer, and Evaluation metrics" ] }, { "cell_type": "code", "execution_count": 27, "id": "c0a66a69", "metadata": {}, "outputs": [], "source": [ "from typing import Tuple, Dict, List\n", "from timeit import default_timer as timer\n", "from tqdm.auto import tqdm" ] }, { "cell_type": "code", "execution_count": 28, "id": "e76f93ab", "metadata": {}, "outputs": [], "source": [ "# Create accuracy function\n", "def accuracy_fn(y_true, y_pred):\n", " correct = torch.eq(y_true, y_pred).sum().item()\n", " accuracy = (correct/len(y_pred)) * 100\n", " return accuracy" ] }, { "cell_type": "code", "execution_count": 29, "id": "626043cf", "metadata": {}, "outputs": [], "source": [ "# Setup loss function\n", "loss_fn = nn.CrossEntropyLoss()\n", "\n", "# Setup optimizers\n", "optimizer_1 = torch.optim.SGD(params = model_1.parameters(),\n", " lr=0.01)\n", "optimizer_2 = torch.optim.SGD(params = model_2.parameters(),\n", " lr=0.01)\n", "optimizer_3 = torch.optim.SGD(params = model_3.parameters(),\n", " lr=0.01)" ] }, { "cell_type": "markdown", "id": "27eb2eec", "metadata": {}, "source": [ "### 4.2 Create functions" ] }, { "cell_type": "code", "execution_count": 30, "id": "96e71e68", "metadata": {}, "outputs": [], "source": [ "# Create train_step() function\n", "def train_step(model: torch.nn.Module,\n", " dataloader: torch.utils.data.DataLoader,\n", " loss_fn: torch.nn.Module,\n", " optimizer: torch.optim.Optimizer,\n", " device=device):\n", " # Switches the model to train mode\n", " model.train()\n", " \n", " # Initialize variables to store total training loss and accuracy\n", " train_loss, train_acc = 0, 0\n", " \n", " # Loop over batches from the DataLoader\n", " for batch, (X, y) in enumerate(dataloader):\n", " \n", " # Send batch of images and labels to the computation device (CPU/GPU)\n", " X, y = X.to(device), y.to(device)\n", " \n", " # Perform a forward pass through the model to get the predictions\n", " y_pred = model(X)\n", " \n", " # Calculate the loss between the predictions and actual values\n", " loss = loss_fn(y_pred, y)\n", " # Add up the loss values\n", " train_loss += loss.item()\n", " \n", " # Reset the gradients from the previous iteration\n", " optimizer.zero_grad()\n", " \n", " # Perform backward propagation to calculate gradients\n", " loss.backward()\n", " \n", " # Perform a step of optimization\n", " optimizer.step()\n", " \n", " # Get the predicted class by taking the maximum probability from the softmax output\n", " y_pred_class = torch.argmax(torch.softmax(y_pred, dim=1), dim=1)\n", " \n", " # Calculate accuracy by comparing predicted class to actual class, and add up for all instances\n", " train_acc += (y_pred_class==y).sum().item()/len(y_pred)\n", " \n", " train_loss = train_loss / len(dataloader) # Calculate average training loss\n", " train_acc = train_acc / len(dataloader) # Calculate average training accuracy\n", " return train_loss, train_acc # Return average training loss and accuracy" ] }, { "cell_type": "code", "execution_count": 31, "id": "c3c1d7d6", "metadata": {}, "outputs": [], "source": [ "# Create test_step() function\n", "def test_step(model: torch.nn.Module,\n", " dataloader: torch.utils.data.DataLoader,\n", " loss_fn: torch.nn.Module,\n", " device=device):\n", " \n", " # Switches the model to evaluation mode\n", " model.eval()\n", " \n", " # Initialize variables to store total test loss and accuracy\n", " test_loss, test_acc = 0, 0\n", " \n", " # Disable calculation of gradients for performance boost during inference\n", " with torch.inference_mode():\n", " \n", " # Loop over batches from the DataLoader\n", " for batch, (X, y) in enumerate(dataloader):\n", " \n", " # Send batch of images and labels to the computation device (CPU/GPU)\n", " X, y = X.to(device), y.to(device)\n", " \n", " # Perform a forward pass through the model to get the predictions\n", " test_pred_logits = model(X)\n", " \n", " # Calculate the loss between the predictions and actual values\n", " loss = loss_fn(test_pred_logits, y)\n", " # Add up the loss values\n", " test_loss += loss.item()\n", " \n", " # Get the predicted class by taking the index of the maximum logit\n", " test_pred_labels = test_pred_logits.argmax(dim=1)\n", " # Calculate accuracy by comparing predicted class to actual class, and add up for all instances\n", " test_acc += ((test_pred_labels==y).sum().item()/len(test_pred_labels))\n", " \n", " test_loss = test_loss / len(dataloader) # Calculate average test loss\n", " test_acc = test_acc / len(dataloader) # Calculate average test accuracy\n", " return test_loss, test_acc # Return average test loss and accuracy" ] }, { "cell_type": "code", "execution_count": 32, "id": "921ea9e4", "metadata": {}, "outputs": [], "source": [ "# Create train() function\n", "def train(model: torch.nn.Module,\n", " train_dataloader: torch.utils.data.DataLoader,\n", " test_dataloader: torch.utils.data.DataLoader,\n", " optimizer: torch.optim.Optimizer,\n", " loss_fn: torch.nn.Module = nn.CrossEntropyLoss(),\n", " epochs: int = 5,\n", " device = device):\n", " \n", " # Initialize a dictionary to store training and validation losses and accuracies for each epoch\n", " results = {\"train_loss\": [],\n", " \"train_acc\": [],\n", " \"test_loss\": [],\n", " \"test_acc\": []}\n", " \n", " # Establish the start time for training\n", " start_time = timer()\n", " \n", " # Loop over epochs\n", " for epoch in tqdm(range(epochs)):\n", " # Execute a training step and get training loss and accuracy\n", " train_loss, train_acc = train_step(model=model,\n", " dataloader=train_dataloader,\n", " loss_fn=loss_fn,\n", " optimizer=optimizer,\n", " device=device)\n", " # Execute a testing step and get testing loss and accuracy\n", " test_loss, test_acc = test_step(model=model,\n", " dataloader=test_dataloader,\n", " loss_fn=loss_fn,\n", " device=device)\n", " \n", " # Print losses and accuracies for this epoch\n", " print(f\"Epoch: {epoch} | Train loss: {train_loss:.4f} | Train acc: {train_acc:.4f} | Test loss: {test_loss:.4f} | Test acc: {test_acc:.4f}\")\n", " \n", " # Append losses and accuracies to results dictionary\n", " results[\"train_loss\"].append(train_loss)\n", " results[\"train_acc\"].append(train_acc)\n", " results[\"test_loss\"].append(test_loss)\n", " results[\"test_acc\"].append(test_acc)\n", " \n", " # Establish the end time for training\n", " end_time = timer()\n", " \n", " return results, (end_time-start_time)" ] }, { "cell_type": "code", "execution_count": 33, "id": "be1d66a6", "metadata": {}, "outputs": [], "source": [ "# Create plot_loss_curves() function\n", "def plot_loss_curves(results: Dict[str, List[float]]):\n", " # Extract training and validation losses and accuracies from results dictionary\n", " loss = results[\"train_loss\"]\n", " test_loss = results[\"test_loss\"] \n", " accuracy = results[\"train_acc\"]\n", " test_accuracy = results[\"test_acc\"]\n", " \n", " # Number of epochs is the length of any list in results\n", " epochs = range(len(results[\"train_loss\"]))\n", " \n", " # Set figure size\n", " plt.figure(figsize=(15, 7))\n", " \n", " # Subplot for loss\n", " plt.subplot(1, 2, 1)\n", " plt.plot(epochs, loss, label=\"train_loss\")\n", " plt.plot(epochs, test_loss, label=\"test_loss\")\n", " plt.title(\"Loss\")\n", " plt.xlabel(\"Epochs\")\n", " plt.legend()\n", " \n", " # Subplot for accuracy\n", " plt.subplot(1, 2, 2)\n", " plt.plot(epochs, accuracy, label=\"train_accuracy\")\n", " plt.plot(epochs, test_accuracy, label=\"test_accuracy\")\n", " plt.title(\"Accuracy\")\n", " plt.xlabel(\"Epochs\")\n", " plt.legend()" ] }, { "cell_type": "code", "execution_count": 34, "id": "85381619", "metadata": {}, "outputs": [], "source": [ "# Create eval_model() function\n", "def eval_model(model: torch.nn.Module,\n", " data_loader: torch.utils.data.DataLoader,\n", " loss_fn: torch.nn.Module,\n", " accuracy_fn,\n", " device=device):\n", " \"\"\"Returns a dictionary containing the results of model predicting on data_loader.\"\"\"\n", " loss, acc= 0, 0\n", " with torch.inference_mode():\n", " for X, y in data_loader:\n", " # Make our data device agnostic\n", " X, y = X.to(device), y.to(device)\n", " # Make predictions\n", " y_pred = model(X)\n", " \n", " # Accumulate the loss and acc values per batch\n", " loss += loss_fn(y_pred, y)\n", " acc += accuracy_fn(y_true=y,\n", " y_pred=y_pred.argmax(dim=1))\n", " \n", " # Scale loss and acc to find the average loss/acc per batch\n", " loss /= len(data_loader)\n", " acc /= len(data_loader)\n", " \n", " return {\"model_name\": model.__class__.__name__,\n", " \"model_loss\": loss.item(),\n", " \"model_acc\": acc}" ] }, { "cell_type": "markdown", "id": "f3936b85", "metadata": {}, "source": [ "## 5 Model evaluation" ] }, { "cell_type": "code", "execution_count": 35, "id": "0db095a0", "metadata": {}, "outputs": [], "source": [ "# Set hyperparameters\n", "NUM_EPOCHS = 20" ] }, { "cell_type": "markdown", "id": "f62ce4be", "metadata": {}, "source": [ "### 5.1 Model 1 evaluation" ] }, { "cell_type": "code", "execution_count": 36, "id": "ad04f43a", "metadata": {}, "outputs": [ { "data": { "application/vnd.jupyter.widget-view+json": { "model_id": "6a1ff1811f9e4cd2a7c34c5d8f3a48be", "version_major": 2, "version_minor": 0 }, "text/plain": [ " 0%| | 0/20 [00:00" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "# Plot loss and accuracy curves\n", "plot_loss_curves(model_1_results)" ] }, { "cell_type": "markdown", "id": "8c3298ae", "metadata": {}, "source": [ "### 5.2 Model 2 evaluation" ] }, { "cell_type": "code", "execution_count": 39, "id": "45c492c6", "metadata": {}, "outputs": [ { "data": { "application/vnd.jupyter.widget-view+json": { "model_id": "a00810f9536b4f5d9348aaa951383e80", "version_major": 2, "version_minor": 0 }, "text/plain": [ " 0%| | 0/20 [00:00" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "# Plot loss and accuracy curves\n", "plot_loss_curves(model_2_results)" ] }, { "cell_type": "markdown", "id": "ad41810f", "metadata": {}, "source": [ "### 5.3 Model 3 evaluation" ] }, { "cell_type": "code", "execution_count": 42, "id": "ecc34e0f", "metadata": {}, "outputs": [ { "data": { "application/vnd.jupyter.widget-view+json": { "model_id": "100c739129b94953a4c160c6a983da6b", "version_major": 2, "version_minor": 0 }, "text/plain": [ " 0%| | 0/20 [00:00" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "# Plot loss and accuracy curves\n", "plot_loss_curves(model_3_results)" ] }, { "cell_type": "markdown", "id": "6cb25d7b", "metadata": {}, "source": [ "### 5.4 Compare results" ] }, { "cell_type": "code", "execution_count": 45, "id": "884a3b21", "metadata": {}, "outputs": [], "source": [ "import pandas as pd" ] }, { "cell_type": "code", "execution_count": 46, "id": "8c9c3896", "metadata": {}, "outputs": [ { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
model_namemodel_lossmodel_acctraining_time
0MNISTModelv10.27148192.302316163.590406
1MNISTModelv20.05204798.302716197.810955
2MNISTModelv30.03725498.951677288.939552
\n", "
" ], "text/plain": [ " model_name model_loss model_acc training_time\n", "0 MNISTModelv1 0.271481 92.302316 163.590406\n", "1 MNISTModelv2 0.052047 98.302716 197.810955\n", "2 MNISTModelv3 0.037254 98.951677 288.939552" ] }, "execution_count": 46, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# Combine results into a single dataframe\n", "compare_results = pd.DataFrame([model_1_final_results,\n", " model_2_final_results,\n", " model_3_final_results])\n", "\n", "# Add training time to dataframe\n", "compare_results[\"training_time\"] = [model_1_time,\n", " model_2_time,\n", " model_3_time]\n", "\n", "# Print results dataframe\n", "compare_results" ] }, { "cell_type": "code", "execution_count": 47, "id": "f2a19979", "metadata": {}, "outputs": [ { "data": { "text/plain": [ "Text(0.5, 1.0, 'Model Training Time')" ] }, "execution_count": 47, "metadata": {}, "output_type": "execute_result" }, { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "# Visualize results\n", "plt.figure(figsize=(10,5))\n", "\n", "# Create accuracy subplot\n", "plt.subplot(1, 2, 1)\n", "compare_results.set_index(\"model_name\")[\"model_acc\"].plot(kind='bar')\n", "plt.xlabel(\"Model\")\n", "plt.ylabel(\"Accuracy (%)\")\n", "plt.title(\"Model Accuracy\")\n", "\n", "# Create training time subplot\n", "plt.subplot(1, 2, 2)\n", "compare_results.set_index(\"model_name\")[\"training_time\"].plot(kind='bar')\n", "plt.xlabel(\"Model\")\n", "plt.ylabel(\"Training Time (s)\")\n", "plt.title(\"Model Training Time\")" ] }, { "cell_type": "code", "execution_count": 48, "id": "fd3fe499", "metadata": {}, "outputs": [ { "data": { "text/plain": [ "" ] }, "execution_count": 48, "metadata": {}, "output_type": "execute_result" }, { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "# Setup a plot\n", "plt.figure(figsize=(15,10))\n", "\n", "# Get number of epochs\n", "epochs = range(len(model_1_results[\"train_loss\"]))\n", "\n", "# Plot train loss\n", "plt.subplot(2, 2, 1)\n", "plt.plot(epochs, model_1_results[\"train_loss\"], label=\"Model 1\")\n", "plt.plot(epochs, model_2_results[\"train_loss\"], label=\"Model 2\")\n", "plt.plot(epochs, model_3_results[\"train_loss\"], label=\"Model 3\")\n", "plt.title(\"Train Loss\")\n", "plt.xlabel(\"Epochs\")\n", "plt.legend()\n", "\n", "# Plot test loss\n", "plt.subplot(2, 2, 2)\n", "plt.plot(epochs, model_1_results[\"test_loss\"], label=\"Model 1\")\n", "plt.plot(epochs, model_2_results[\"test_loss\"], label=\"Model 2\")\n", "plt.plot(epochs, model_3_results[\"test_loss\"], label=\"Model 3\")\n", "plt.title(\"Test Loss\")\n", "plt.xlabel(\"Epochs\")\n", "plt.legend()\n", "\n", "# Plot train acc\n", "plt.subplot(2, 2, 3)\n", "plt.plot(epochs, model_1_results[\"train_acc\"], label=\"Model 1\")\n", "plt.plot(epochs, model_2_results[\"train_acc\"], label=\"Model 2\")\n", "plt.plot(epochs, model_3_results[\"train_acc\"], label=\"Model 3\")\n", "plt.title(\"Train Acc\")\n", "plt.xlabel(\"Epochs\")\n", "plt.legend()\n", "\n", "# Plot test acc\n", "plt.subplot(2, 2, 4)\n", "plt.plot(epochs, model_1_results[\"test_acc\"], label=\"Model 1\")\n", "plt.plot(epochs, model_2_results[\"test_acc\"], label=\"Model 2\")\n", "plt.plot(epochs, model_3_results[\"test_acc\"], label=\"Model 3\")\n", "plt.title(\"Test Acc\")\n", "plt.xlabel(\"Epochs\")\n", "plt.legend()" ] }, { "cell_type": "markdown", "id": "ddb569b6", "metadata": {}, "source": [ "# 6. Making predictions using the best model" ] }, { "cell_type": "markdown", "id": "8fd3b972", "metadata": {}, "source": [ "## 6.1 Create custom functions" ] }, { "cell_type": "code", "execution_count": 49, "id": "edf79b61", "metadata": {}, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ "C:\\Users\\clopt\\AppData\\Local\\anaconda3\\envs\\pytorch\\lib\\site-packages\\torchaudio\\backend\\utils.py:74: UserWarning: No audio backend is available.\n", " warnings.warn(\"No audio backend is available.\")\n" ] } ], "source": [ "import random\n", "from torchmetrics import ConfusionMatrix\n", "from mlxtend.plotting import plot_confusion_matrix\n", "import numpy as np" ] }, { "cell_type": "code", "execution_count": 50, "id": "020ebd64", "metadata": {}, "outputs": [], "source": [ "def predict_label(model, img, device):\n", " \"\"\"predicts the label based on a model, image, and device\"\"\"\n", " img = img.unsqueeze(dim=1).to(device)\n", " with torch.inference_mode():\n", " pred_label = torch.softmax(model(img), dim=1).argmax().item()\n", " return pred_label" ] }, { "cell_type": "code", "execution_count": 51, "id": "aa1caad2", "metadata": {}, "outputs": [], "source": [ "def plot_img(img, pred_label, truth_label):\n", " \"\"\"plots the image, its predicted label, and actual label\"\"\"\n", " plt.imshow(img.squeeze(), cmap=\"gray\")\n", " title_text = f\"Pred: {pred_label} | Truth: {truth_label}\"\n", " title_color = \"g\" if pred_label == truth_label else \"r\"\n", " plt.title(title_text, fontsize=10, color=title_color)\n", " plt.axis(False)" ] }, { "cell_type": "code", "execution_count": 52, "id": "7994a0da", "metadata": {}, "outputs": [], "source": [ "def plot_images_in_grid(images, pred_labels, true_labels, nrows=3, ncols=3):\n", " \"\"\"Plot images in a grid pattern based on row/column input. Does not allow 1x1 grid.\"\"\"\n", " fig, axs = plt.subplots(nrows, ncols, figsize=(9,9))\n", " # If we have only 1 row or column, make sure axs is a 2D array for consistent handling\n", " axs = np.array(axs).reshape(nrows, ncols)\n", " for i, ax in enumerate(axs.flatten()):\n", " if i < len(images): # Make sure we don't go out of bounds\n", " img = images[i].squeeze()\n", " pred_label = pred_labels[i]\n", " truth_label = true_labels[i]\n", " ax.imshow(img, cmap=\"gray\")\n", " title_text = f\"Pred: {pred_label} | Truth: {truth_label}\"\n", " title_color = \"g\" if pred_label == truth_label else \"r\"\n", " ax.set_title(title_text, fontsize=10, color=title_color)\n", " ax.axis(False)\n", " plt.tight_layout()" ] }, { "cell_type": "code", "execution_count": 53, "id": "e2738c43", "metadata": {}, "outputs": [], "source": [ "def create_confusion_matrix(y_true_tensor, y_pred_tensor, class_names):\n", " \"\"\"Create a confusion matrix\"\"\"\n", " confmat = ConfusionMatrix(num_classes=len(class_names), task='multiclass')\n", " confmat_tensor = confmat(preds=y_pred_tensor, target=y_true_tensor)\n", " \n", " # Plot the confusion matrix\n", " fig, ax = plot_confusion_matrix(\n", " conf_mat=confmat_tensor.numpy(),\n", " class_names=class_names,\n", " figsize=(10,7)\n", " )\n", " return confmat_tensor.numpy()" ] }, { "cell_type": "code", "execution_count": 54, "id": "99e1d537", "metadata": {}, "outputs": [], "source": [ "def make_predictions(model, dataloader, device):\n", " \"\"\"Make predictions using models, data, and device on a dataset using a dataloader\"\"\"\n", " y_true = []\n", " y_preds = []\n", " image_list = []\n", " model.eval()\n", " with torch.inference_mode():\n", " for X, y in tqdm(dataloader, desc=\"Making predictions...\"):\n", " X, y = X.to(device), y.to(device)\n", " image_list.append(X.cpu())\n", " y_logit = model(X)\n", " y_true.append(y.cpu())\n", " y_pred = torch.softmax(y_logit.squeeze(), dim=0).argmax(dim=1)\n", " y_preds.append(y_pred.cpu())\n", " image_list_tensor = torch.cat(image_list)\n", " y_true_tensor = torch.cat(y_true)\n", " y_pred_tensor = torch.cat(y_preds)\n", " \n", " return y_true_tensor, y_pred_tensor, image_list_tensor" ] }, { "cell_type": "code", "execution_count": 55, "id": "a2e96fcb", "metadata": {}, "outputs": [], "source": [ "def accuracy_chart(confmat):\n", " # calculate accuracy for each label\n", " label_acc = confmat.diagonal()/confmat.sum(axis=1)\n", " label_acc = np.nan_to_num(label_acc) # in case of NaN or Inf, replace them by 0\n", "\n", " # Sort labels by accuracy\n", " sorted_indices = np.argsort(label_acc)\n", " sorted_label_acc = label_acc[sorted_indices]\n", "\n", " # Get label names (replace with actual label names)\n", " labels = [f\"{i}\" for i in range(len(label_acc))]\n", "\n", " # Sort label names according to accuracy\n", " sorted_labels = [labels[i] for i in sorted_indices]\n", "\n", " # Create the bar chart\n", " plt.figure(figsize=(12,6))\n", " plt.barh(sorted_labels, sorted_label_acc * 100) # multiply by 100 to get percentage\n", " plt.xlabel('Accuracy (%)')\n", " plt.ylabel('Labels')\n", " plt.title('Accuracy of Each Label')\n", " plt.xlim([0, 100])\n", " plt.show()" ] }, { "cell_type": "markdown", "id": "39d2a4b5", "metadata": {}, "source": [ "### 6.2 Predict on custom dataset" ] }, { "cell_type": "code", "execution_count": 56, "id": "a7e7ded2", "metadata": {}, "outputs": [ { "data": { "application/vnd.jupyter.widget-view+json": { "model_id": "068442b5d6014a34b4ba4e898ad72c7e", "version_major": 2, "version_minor": 0 }, "text/plain": [ "Making predictions...: 0%| | 0/313 [00:00" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "# Establish parameters for image grids\n", "NROWS = 4\n", "NCOLS = 4\n", "\n", "# Plot images from MNIST dataset\n", "plot_images_in_grid(image_list_tensor_mnist,\n", " y_pred_tensor_mnist,\n", " y_true_tensor_mnist, \n", " NROWS,\n", " NCOLS)" ] }, { "cell_type": "code", "execution_count": 58, "id": "4b759e15", "metadata": {}, "outputs": [ { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "# Create confusion matrix from model_3 on MNIST data\n", "confmat_mnist = create_confusion_matrix(y_true_tensor_mnist, \n", " y_pred_tensor_mnist, \n", " class_names)" ] }, { "cell_type": "code", "execution_count": 59, "id": "9368a8a9", "metadata": {}, "outputs": [ { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "# Create a bar chart showing accuracy rates\n", "accuracy_chart(confmat_mnist)" ] }, { "cell_type": "markdown", "id": "e90c04a0", "metadata": {}, "source": [ "# 7. Save and load model" ] }, { "cell_type": "markdown", "id": "faf36731", "metadata": {}, "source": [ "### 7.1 Save the best model" ] }, { "cell_type": "code", "execution_count": 60, "id": "0a45c0d0", "metadata": {}, "outputs": [], "source": [ "from pathlib import Path" ] }, { "cell_type": "code", "execution_count": 61, "id": "fa7ada00", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Saving model to: models\\model_3.pth\n" ] } ], "source": [ "# Create model directory path\n", "MODEL_PATH = Path(\"models\")\n", "MODEL_PATH.mkdir(parents=True,\n", " exist_ok=True)\n", "\n", "# Create model save path\n", "MODEL_NAME = \"model_3.pth\"\n", "MODEL_SAVE_PATH = MODEL_PATH / MODEL_NAME\n", "\n", "# Save the model\n", "print(f\"Saving model to: {MODEL_SAVE_PATH}\")\n", "torch.save(obj=model_3.state_dict(),\n", " f=MODEL_SAVE_PATH)" ] }, { "cell_type": "markdown", "id": "86ddb0a3", "metadata": {}, "source": [ "### 7.2 Load the best model" ] }, { "cell_type": "code", "execution_count": 62, "id": "ba8eeacd", "metadata": {}, "outputs": [ { "data": { "text/plain": [ "MNISTModelv3(\n", " (conv_block_1): Sequential(\n", " (0): Conv2d(1, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n", " (1): ReLU()\n", " (2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n", " (3): ReLU()\n", " (4): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n", " )\n", " (conv_block_2): Sequential(\n", " (0): Conv2d(64, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n", " (1): ReLU()\n", " (2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n", " (3): ReLU()\n", " (4): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n", " )\n", " (conv_block_3): Sequential(\n", " (0): Conv2d(128, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n", " (1): ReLU()\n", " (2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n", " (3): ReLU()\n", " (4): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)\n", " )\n", " (classifier): Sequential(\n", " (0): Flatten(start_dim=1, end_dim=-1)\n", " (1): Linear(in_features=2304, out_features=4096, bias=True)\n", " (2): ReLU()\n", " (3): Dropout(p=0.5, inplace=False)\n", " (4): Linear(in_features=4096, out_features=4096, bias=True)\n", " (5): ReLU()\n", " (6): Dropout(p=0.5, inplace=False)\n", " (7): Linear(in_features=4096, out_features=10, bias=True)\n", " )\n", ")" ] }, "execution_count": 62, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# Create new instance\n", "loaded_model_3 = MNISTModelv3(input_shape=1,\n", " output_shape=len(class_names))\n", "\n", "# Load in the saved state_dict()\n", "loaded_model_3.load_state_dict(torch.load(f=MODEL_SAVE_PATH))\n", "\n", "# Send model to the target device\n", "loaded_model_3.to(device)" ] }, { "cell_type": "markdown", "id": "beeafa0a", "metadata": {}, "source": [ "## 8. Predict on custom dataset" ] }, { "cell_type": "code", "execution_count": 63, "id": "bbc8cd57", "metadata": {}, "outputs": [], "source": [ "from torch.utils.data import Dataset\n", "import pathlib\n", "from PIL import Image\n", "import os" ] }, { "cell_type": "markdown", "id": "b3015917", "metadata": {}, "source": [ "### 8.1 Create custom functions" ] }, { "cell_type": "code", "execution_count": 64, "id": "7703e453", "metadata": {}, "outputs": [], "source": [ "class ImageFolderCustom(Dataset):\n", " # Initialize custom dataset\n", " def __init__(self, target_dir: str, transform=None):\n", " # Get all of the image paths\n", " self.paths = list(pathlib.Path(target_dir).glob(\"*/*.jpg\"))\n", " # Setup transforms\n", " self.transform = transform\n", " # Create classes and class_to_idx attributes\n", " self.classes, self.class_to_idx = find_classes(target_dir)\n", " \n", " # Create a function to load image\n", " def load_image(self, index: int) -> Image.Image:\n", " \"Opens an image via a path and returns it.\"\n", " image_path = self.paths[index]\n", " return Image.open(image_path)\n", " \n", " # Overwrite __len()__\n", " def __len__(self) -> int:\n", " \"Returns the total number of samples\"\n", " return len(self.paths)\n", " \n", " # Overwrite __getitem__() method to return a particular sample\n", " def __getitem__(self, index: int) -> Tuple[torch.Tensor, int]:\n", " \"Returns one sample of data, data and label (X, y).\"\n", " img = self.load_image(index)\n", " class_name = self.paths[index].parent.name # expects path in format: data_folder/class_name/image.jpg\n", " class_idx = self.class_to_idx[class_name]\n", " \n", " # Transform if necessary\n", " if self.transform:\n", " return self.transform(img), class_idx # return data, label (X, y)\n", " else:\n", " return img, class_idx # return untransformed image and label" ] }, { "cell_type": "code", "execution_count": 65, "id": "c311991d", "metadata": {}, "outputs": [], "source": [ "def find_classes(directory: str) -> Tuple[List[str], Dict[str, int]]:\n", " \"\"\"Finds the class folder names in a target directory.\"\"\"\n", " # Get the class names by scanning the target directory\n", " classes = sorted(entry.name for entry in os.scandir(directory) if entry.is_dir())\n", " \n", " # Raise an error if class names could not be found\n", " if not classes:\n", " raise FileNotFoundError(f\"Couldn't find any classes in {directory}... please check file structure.\")\n", " \n", " # Create a dictionary of index labels\n", " class_to_idx = {class_name: i for i, class_name in enumerate(classes)}\n", " \n", " return classes, class_to_idx" ] }, { "cell_type": "code", "execution_count": 66, "id": "654f09e4", "metadata": {}, "outputs": [], "source": [ "# Create method for inverting tensors as they are transformed\n", "class Inversion(object):\n", " def __call__(self, tensor):\n", " return 1 - tensor" ] }, { "cell_type": "markdown", "id": "a6674c78", "metadata": {}, "source": [ "### 8.2 Import custom dataset" ] }, { "cell_type": "code", "execution_count": 67, "id": "c7236201", "metadata": {}, "outputs": [], "source": [ "# Set transforms for incoming data\n", "test_transforms = transforms.Compose([transforms.Resize(size=(28,28)),\n", " transforms.Grayscale(),\n", " transforms.ToTensor(),\n", " Inversion()]) # Inversion() is my custom method" ] }, { "cell_type": "code", "execution_count": 68, "id": "7fe4c07a", "metadata": {}, "outputs": [], "source": [ "# Set directory for custom testing data\n", "custom_dir = \"custom_data\"\n", "\n", "test_data_custom = ImageFolderCustom(target_dir=custom_dir,\n", " transform=test_transforms)" ] }, { "cell_type": "code", "execution_count": 69, "id": "8dd788e5", "metadata": {}, "outputs": [], "source": [ "# This code errors out when over 0 NUM_WORKERS are used - investigate later\n", "NUM_WORKERS = 0\n", "\n", "# Create dataloader for custom data\n", "test_dataloader_custom = DataLoader(dataset=test_data_custom,\n", " batch_size=BATCH_SIZE,\n", " num_workers=NUM_WORKERS,\n", " shuffle=True)" ] }, { "cell_type": "markdown", "id": "a59862cc", "metadata": {}, "source": [ "### 8.3 Predict on custom dataset" ] }, { "cell_type": "code", "execution_count": 70, "id": "3b8e0c19", "metadata": {}, "outputs": [ { "data": { "application/vnd.jupyter.widget-view+json": { "model_id": "2c8293144b4f4d54a08c91397ec5b1a9", "version_major": 2, "version_minor": 0 }, "text/plain": [ "Making predictions...: 0%| | 0/4 [00:00" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "# Plot images with prediction/truth values\n", "plot_images_in_grid(image_list_tensor_custom, \n", " y_pred_tensor_custom, \n", " y_true_tensor_custom, \n", " NROWS, \n", " NCOLS)" ] }, { "cell_type": "code", "execution_count": 73, "id": "ab6b1d69", "metadata": {}, "outputs": [ { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "# Create confusion matrix \n", "confmat_custom = create_confusion_matrix(y_true_tensor_custom, \n", " y_pred_tensor_custom, \n", " class_names)" ] }, { "cell_type": "code", "execution_count": 74, "id": "865b31bf", "metadata": {}, "outputs": [ { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "# Create a bar chart showing accuracy rates\n", "accuracy_chart(confmat_custom)" ] }, { "cell_type": "code", "execution_count": null, "id": "ae8fc580", "metadata": {}, "outputs": [], "source": [] } ], "metadata": { "kernelspec": { "display_name": "Python 3 (ipykernel)", "language": "python", "name": "python3" }, "language_info": { "codemirror_mode": { "name": "ipython", "version": 3 }, "file_extension": ".py", "mimetype": "text/x-python", "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", "version": "3.10.13" } }, "nbformat": 4, "nbformat_minor": 5 }