Skip to content

Instantly share code, notes, and snippets.

Show Gist options
  • Save AlisonDavey/17c5a5c09c7bbf31a3b95515c7cb70e6 to your computer and use it in GitHub Desktop.
Save AlisonDavey/17c5a5c09c7bbf31a3b95515c7cb70e6 to your computer and use it in GitHub Desktop.
Image_Classifier_Project.ipynb
Display the source blob
Display the rendered blob
Raw
{
"cells": [
{
"metadata": {},
"cell_type": "markdown",
"source": "# Udacity PyTorch Challenge Classification Project\n\nIn this project an image classifier is trained to recognize different species of flowers using [the Oxford dataset](http://www.robots.ox.ac.uk/~vgg/data/flowers/102/index.html) of 102 flower categories.\n\nThe project is broken down into:\n\n* Create a test dataset\n* Load and preprocess the image dataset\n* Train the image classifier on the dataset\n* Use the trained classifier to predict image content\n\n[PEP 8 -- Style Guide for Python Code](https://www.python.org/dev/peps/pep-0008/) says that imports are always put at the top of the file.\n\nSource files https://github.com/udacity/pytorch_challenge\n"
},
{
"metadata": {
"trusted": false
},
"cell_type": "code",
"source": "%reload_ext autoreload\n%autoreload 2\n%matplotlib inline\n\nfrom PIL import Image\nimport matplotlib.pyplot as plt\nimport numpy as np\nimport pandas as pd\nfrom collections import OrderedDict\n\nimport torch\nimport torch.optim as optim\nimport torch.nn as nn\nfrom torchvision import datasets, transforms, models\n\nimport scipy.io as sio\nfrom pathlib import Path\nfrom shutil import copy\nimport time\n#import copy\nimport shutil\nimport os\nimport random",
"execution_count": 1,
"outputs": []
},
{
"metadata": {},
"cell_type": "markdown",
"source": "## Data Downloading and Creation of a Test Dataset\n\nCreate a test folder (with its labels) for the Udacity PyTorch Challenge flower dataset that contains the images in the original Oxford dataset that are not in the Udacity dataset.\n\nThe dataset will then have three parts: training, validation and test. "
},
{
"metadata": {
"trusted": false
},
"cell_type": "code",
"source": "# Download the original dataset of 102 different categories of flowers common to the UK \n# from the Visual Geometry Group, Department of Engineering Science, University of Oxford \n# http://www.robots.ox.ac.uk/~vgg/data/flowers/\n!wget -O ./assets/102flowers.tgz \"http://www.robots.ox.ac.uk/~vgg/data/flowers/102/102flowers.tgz\"\n",
"execution_count": 14,
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": "--2019-01-19 17:06:09-- http://www.robots.ox.ac.uk/~vgg/data/flowers/102/102flowers.tgz\nResolving www.robots.ox.ac.uk (www.robots.ox.ac.uk)... 129.67.94.2\nConnecting to www.robots.ox.ac.uk (www.robots.ox.ac.uk)|129.67.94.2|:80... connected.\nHTTP request sent, awaiting response... 200 OK\nLength: 344862509 (329M) [application/x-gzip]\nSaving to: ‘./assets/102flowers.tgz’\n\n./assets/102flowers 100%[===================>] 328.89M 21.7MB/s in 16s \n\n2019-01-19 17:06:27 (20.4 MB/s) - ‘./assets/102flowers.tgz’ saved [344862509/344862509]\n\n"
}
]
},
{
"metadata": {
"trusted": false
},
"cell_type": "code",
"source": "# Untar the Oxford dataset\n!tar xzf ./assets/102flowers.tgz -C ./assets/\n",
"execution_count": 15,
"outputs": []
},
{
"metadata": {
"trusted": false
},
"cell_type": "code",
"source": "## Download the Oxford labels\n!wget -O ./assets/imagelabels.mat \"http://www.robots.ox.ac.uk/~vgg/data/flowers/102/imagelabels.mat\"\n",
"execution_count": 16,
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": "--2019-01-19 17:07:20-- http://www.robots.ox.ac.uk/~vgg/data/flowers/102/imagelabels.mat\nResolving www.robots.ox.ac.uk (www.robots.ox.ac.uk)... 129.67.94.2\nConnecting to www.robots.ox.ac.uk (www.robots.ox.ac.uk)|129.67.94.2|:80... connected.\nHTTP request sent, awaiting response... 200 OK\nLength: 502\nSaving to: ‘./assets/imagelabels.mat’\n\n./assets/imagelabel 100%[===================>] 502 --.-KB/s in 0s \n\n2019-01-19 17:07:20 (102 MB/s) - ‘./assets/imagelabels.mat’ saved [502/502]\n\n"
}
]
},
{
"metadata": {
"trusted": false
},
"cell_type": "code",
"source": "# Download the Udacity PyTorch Challenge flower dataset \n!wget -O ./assets/flower_data.zip \"https://s3.amazonaws.com/content.udacity-data.com/courses/nd188/flower_data.zip\"\n",
"execution_count": 2,
"outputs": []
},
{
"metadata": {
"trusted": false
},
"cell_type": "code",
"source": "# Unzip the Udacity dataset\n!unzip ./assets/flower_data.zip -d ./assets/\n",
"execution_count": 3,
"outputs": []
},
{
"metadata": {
"trusted": false
},
"cell_type": "code",
"source": "root_dir = Path('./assets')\noriginal_dir = root_dir/'jpg'\nlabels_file = root_dir/'imagelabels.mat'\nudacity_dir = root_dir/'flower_data'\nudacity_train_dir = udacity_dir/'train'\nudacity_valid_dir = udacity_dir/'valid'\nudacity_test_dir = udacity_dir/'test'",
"execution_count": 22,
"outputs": []
},
{
"metadata": {
"trusted": false
},
"cell_type": "code",
"source": "# Script prepared by a student in the Udacity PyTorch Scholarship Challenge, apologies that \n# I didn't note their name\nlabels=sio.loadmat(labels_file)['labels'][0]\n(_, _, original_images) = next(os.walk(original_dir))\noriginal_images = sorted(original_images)\nimage_to_label = {name: labels[i] for i, name in enumerate(original_images)}\nudacity_images = []\nfor root, dirs, files in os.walk(udacity_dir): udacity_images.extend(files)\ndiff = set(original_images) - set(udacity_images)\nudacity_test_dir.mkdir(parents=True, exist_ok=True)\nfor file in diff:\n dest_dir = udacity_test_dir/str(image_to_label[file])\n dest_dir.mkdir(parents=True, exist_ok=True)\n copy(original_dir/file, dest_dir)",
"execution_count": 29,
"outputs": []
},
{
"metadata": {},
"cell_type": "markdown",
"source": "## Data Loading and Exploration\n\nFor the training, transformations such as random scaling, cropping, and flipping are applied. This helps the network generalize leading to better performance. Input data is resized to 224x224 pixels as required by the networks.\n\nThe validation set is used to measure the model's performance on data it hasn't seen yet. For this scaling or rotation transformations are not used, but the images are resized then cropped to the appropriate size.\n\nThe test set was created by a student in the challenge by comparing the data selected by Udacity to the original data.\n\nThe pre-trained networks available from `torchvision` were trained on the ImageNet dataset where each colour channel was normalized separately. The image data here needs to be normalized using the same means and standard deviations: `[0.485, 0.456, 0.406] [0.229, 0.224, 0.225]`, as those calculated from the ImageNet images. These values will shift each colour channel to be centered at 0 and range from -1 to 1."
},
{
"metadata": {
"trusted": false
},
"cell_type": "code",
"source": "#Load and preprocess the image dataset\ndata_dir = './assets/flower_data'\ntrain_dir = data_dir + '/train'\nvalid_dir = data_dir + '/valid'\ntest_dir = data_dir + '/test'",
"execution_count": 2,
"outputs": []
},
{
"metadata": {
"trusted": false
},
"cell_type": "code",
"source": "image = Image.open(train_dir+'/54/image_05459.jpg')\nplt.imshow(image);",
"execution_count": 27,
"outputs": [
{
"data": {
"image/png": "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\n",
"text/plain": "<Figure size 432x288 with 1 Axes>"
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
}
]
},
{
"metadata": {
"trusted": false
},
"cell_type": "code",
"source": "#Exploratory Data Analysis\ndef eda_counts():\n dirs = [train_dir,valid_dir,test_dir]\n for directory in dirs:\n counts = []\n total_images = 0\n min_count = 100\n max_count = 0\n print ('Dataset: ',directory)\n folders = ([name for name in os.listdir(directory)])\n for folder in folders:\n contents = os.listdir(os.path.join(directory,folder))\n total_images += len(contents)\n if len(contents)<min_count:\n min_count = len(contents)\n if len(contents)>max_count:\n max_count = len(contents)\n counts.append((folder,len(contents)))\n print ('Classes: ', len(counts), 'Fewest: ', min_count, 'Most: ', max_count)\n print ('Number of images: ', total_images)\n print ('Counts per class: ', counts)\n print ()\neda_counts()",
"execution_count": 30,
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": "Dataset: ./assets/flower_data/train\nClasses: 102 Fewest: 27 Most: 206\nNumber of images: 6552\nCounts per class: [('87', 51), ('99', 50), ('71', 64), ('60', 85), ('27', 36), ('85', 48), ('100', 35), ('53', 70), ('24', 35), ('48', 57), ('96', 72), ('81', 135), ('13', 38), ('62', 48), ('40', 54), ('64', 42), ('49', 38), ('47', 61), ('32', 36), ('89', 153), ('8', 70), ('44', 73), ('10', 38), ('11', 68), ('98', 68), ('46', 157), ('42', 49), ('61', 36), ('20', 46), ('73', 147), ('56', 92), ('25', 34), ('30', 61), ('22', 47), ('43', 100), ('84', 66), ('18', 65), ('70', 51), ('15', 38), ('69', 46), ('37', 92), ('16', 36), ('33', 31), ('57', 50), ('78', 112), ('101', 49), ('67', 36), ('65', 88), ('34', 28), ('3', 36), ('26', 33), ('36', 62), ('21', 34), ('52', 67), ('92', 53), ('55', 56), ('51', 206), ('28', 55), ('45', 33), ('54', 47), ('41', 97), ('88', 116), ('7', 33), ('77', 205), ('14', 44), ('4', 44), ('83', 104), ('50', 73), ('39', 33), ('63', 42), ('90', 66), ('80', 82), ('97', 54), ('59', 56), ('58', 86), ('76', 83), ('29', 62), ('95', 101), ('72', 77), ('5', 54), ('31', 48), ('79', 34), ('12', 73), ('23', 72), ('91', 59), ('66', 51), ('86', 48), ('74', 142), ('75', 95), ('2', 49), ('17', 60), ('93', 34), ('35', 33), ('9', 41), ('94', 132), ('1', 27), ('102', 36), ('68', 43), ('82', 82), ('6', 35), ('19', 38), ('38', 44)]\n\nDataset: ./assets/flower_data/valid\nClasses: 102 Fewest: 1 Most: 28\nNumber of images: 818\nCounts per class: [('87', 6), ('99', 6), ('71', 5), ('60', 14), ('27', 1), ('85', 5), ('100', 6), ('53', 9), ('24', 5), ('48', 9), ('96', 10), ('81', 18), ('13', 5), ('62', 3), ('40', 5), ('64', 5), ('49', 8), ('47', 3), ('32', 3), ('89', 16), ('8', 5), ('44', 9), ('10', 4), ('11', 10), ('98', 10), ('46', 18), ('42', 6), ('61', 6), ('20', 7), ('73', 19), ('56', 9), ('25', 2), ('30', 10), ('22', 8), ('43', 14), ('84', 10), ('18', 11), ('70', 7), ('15', 7), ('69', 5), ('37', 8), ('16', 2), ('33', 7), ('57', 6), ('78', 11), ('101', 5), ('67', 2), ('65', 7), ('34', 7), ('3', 2), ('26', 3), ('36', 6), ('21', 4), ('52', 10), ('92', 2), ('55', 8), ('51', 28), ('28', 5), ('45', 4), ('54', 10), ('41', 16), ('88', 25), ('7', 1), ('77', 21), ('14', 1), ('4', 6), ('83', 13), ('50', 11), ('39', 3), ('63', 8), ('90', 2), ('80', 12), ('97', 7), ('59', 4), ('58', 14), ('76', 20), ('29', 7), ('95', 13), ('72', 8), ('5', 7), ('31', 2), ('79', 4), ('12', 5), ('23', 12), ('91', 9), ('66', 6), ('86', 5), ('74', 15), ('75', 12), ('2', 6), ('17', 16), ('93', 6), ('35', 4), ('9', 3), ('94', 14), ('1', 8), ('102', 6), ('68', 8), ('82', 13), ('6', 1), ('19', 4), ('38', 4)]\n\nDataset: ./assets/flower_data/test\nClasses: 102 Fewest: 2 Most: 28\nNumber of images: 819\nCounts per class: [('87', 6), ('99', 7), ('71', 9), ('60', 10), ('27', 3), ('85', 10), ('100', 8), ('53', 14), ('24', 2), ('48', 5), ('96', 9), ('81', 13), ('13', 6), ('62', 4), ('40', 8), ('64', 5), ('49', 3), ('47', 3), ('32', 6), ('89', 15), ('8', 10), ('44', 11), ('10', 3), ('11', 9), ('98', 4), ('46', 21), ('42', 4), ('61', 8), ('20', 3), ('73', 28), ('56', 8), ('25', 5), ('30', 14), ('22', 4), ('43', 16), ('84', 10), ('18', 6), ('70', 4), ('15', 4), ('69', 3), ('37', 8), ('16', 3), ('33', 8), ('57', 11), ('78', 14), ('101', 4), ('67', 4), ('65', 7), ('34', 5), ('3', 2), ('26', 5), ('36', 7), ('21', 2), ('52', 8), ('92', 11), ('55', 7), ('51', 24), ('28', 6), ('45', 3), ('54', 4), ('41', 14), ('88', 13), ('7', 6), ('77', 25), ('14', 3), ('4', 6), ('83', 14), ('50', 8), ('39', 5), ('63', 4), ('90', 14), ('80', 11), ('97', 5), ('59', 7), ('58', 14), ('76', 4), ('29', 9), ('95', 14), ('72', 11), ('5', 4), ('31', 2), ('79', 3), ('12', 9), ('23', 7), ('91', 8), ('66', 4), ('86', 5), ('74', 14), ('75', 13), ('2', 5), ('17', 9), ('93', 6), ('35', 6), ('9', 2), ('94', 16), ('1', 5), ('102', 6), ('68', 3), ('82', 17), ('6', 9), ('19', 7), ('38', 8)]\n\n"
}
]
},
{
"metadata": {
"trusted": false
},
"cell_type": "code",
"source": "# For the classes in the training set with few exampls, \n# copy random images in the sub-folder to increase the sample size\nmin_number = 54 #twice the smallest class\nfolders = ([name for name in os.listdir(train_dir)])\nfor folder in folders:\n contents = os.listdir(os.path.join(train_dir,folder))\n if len(contents) < min_number: \n images_to_add = min_number-len(contents)\n idx = np.random.choice(len(contents), images_to_add, replace=False)\n for x in range(images_to_add):\n shutil.copy(train_dir + '/' + str(folder) + '/' + contents[idx[x]], train_dir + '/' + str(folder) + '/cp_' + contents[idx[x]])\n print ('sub_folder ', folder, 'copied ' + str(images_to_add) + ' images')",
"execution_count": 31,
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": "sub_folder 87 copied 3 images\nsub_folder 99 copied 4 images\nsub_folder 27 copied 18 images\nsub_folder 85 copied 6 images\nsub_folder 100 copied 19 images\nsub_folder 24 copied 19 images\nsub_folder 13 copied 16 images\nsub_folder 62 copied 6 images\nsub_folder 64 copied 12 images\nsub_folder 49 copied 16 images\nsub_folder 32 copied 18 images\nsub_folder 10 copied 16 images\nsub_folder 42 copied 5 images\nsub_folder 61 copied 18 images\nsub_folder 20 copied 8 images\nsub_folder 25 copied 20 images\nsub_folder 22 copied 7 images\nsub_folder 70 copied 3 images\nsub_folder 15 copied 16 images\nsub_folder 69 copied 8 images\nsub_folder 16 copied 18 images\nsub_folder 33 copied 23 images\nsub_folder 57 copied 4 images\nsub_folder 101 copied 5 images\nsub_folder 67 copied 18 images\nsub_folder 34 copied 26 images\nsub_folder 3 copied 18 images\nsub_folder 26 copied 21 images\nsub_folder 21 copied 20 images\nsub_folder 92 copied 1 images\nsub_folder 45 copied 21 images\nsub_folder 54 copied 7 images\nsub_folder 7 copied 21 images\nsub_folder 14 copied 10 images\nsub_folder 4 copied 10 images\nsub_folder 39 copied 21 images\nsub_folder 63 copied 12 images\nsub_folder 31 copied 6 images\nsub_folder 79 copied 20 images\nsub_folder 66 copied 3 images\nsub_folder 86 copied 6 images\nsub_folder 2 copied 5 images\nsub_folder 93 copied 20 images\nsub_folder 35 copied 21 images\nsub_folder 9 copied 13 images\nsub_folder 1 copied 27 images\nsub_folder 102 copied 18 images\nsub_folder 68 copied 11 images\nsub_folder 6 copied 19 images\nsub_folder 19 copied 16 images\nsub_folder 38 copied 10 images\n"
}
]
},
{
"metadata": {},
"cell_type": "markdown",
"source": "### Label mapping\n\nThe file `cat_to_name.json` is provided for the mapping from category label to category name. It's a JSON object which can be read in with the [`json` module](https://docs.python.org/2/library/json.html). This gives a dictionary mapping the integer encoded categories to the actual names of the flowers."
},
{
"metadata": {
"trusted": false
},
"cell_type": "code",
"source": "!wget -O ./assets/cat_to_name.json \"https://github.com/udacity/pytorch_challenge/blob/master/cat_to_name.json\"",
"execution_count": 7,
"outputs": []
},
{
"metadata": {
"trusted": false
},
"cell_type": "code",
"source": "import json\nwith open('cat_to_name.json', 'r') as f: cat_to_name = json.load(f) \ncat_to_name ",
"execution_count": 2,
"outputs": [
{
"data": {
"text/plain": "{'1': 'pink primrose',\n '10': 'globe thistle',\n '100': 'blanket flower',\n '101': 'trumpet creeper',\n '102': 'blackberry lily',\n '11': 'snapdragon',\n '12': \"colt's foot\",\n '13': 'king protea',\n '14': 'spear thistle',\n '15': 'yellow iris',\n '16': 'globe-flower',\n '17': 'purple coneflower',\n '18': 'peruvian lily',\n '19': 'balloon flower',\n '2': 'hard-leaved pocket orchid',\n '20': 'giant white arum lily',\n '21': 'fire lily',\n '22': 'pincushion flower',\n '23': 'fritillary',\n '24': 'red ginger',\n '25': 'grape hyacinth',\n '26': 'corn poppy',\n '27': 'prince of wales feathers',\n '28': 'stemless gentian',\n '29': 'artichoke',\n '3': 'canterbury bells',\n '30': 'sweet william',\n '31': 'carnation',\n '32': 'garden phlox',\n '33': 'love in the mist',\n '34': 'mexican aster',\n '35': 'alpine sea holly',\n '36': 'ruby-lipped cattleya',\n '37': 'cape flower',\n '38': 'great masterwort',\n '39': 'siam tulip',\n '4': 'sweet pea',\n '40': 'lenten rose',\n '41': 'barbeton daisy',\n '42': 'daffodil',\n '43': 'sword lily',\n '44': 'poinsettia',\n '45': 'bolero deep blue',\n '46': 'wallflower',\n '47': 'marigold',\n '48': 'buttercup',\n '49': 'oxeye daisy',\n '5': 'english marigold',\n '50': 'common dandelion',\n '51': 'petunia',\n '52': 'wild pansy',\n '53': 'primula',\n '54': 'sunflower',\n '55': 'pelargonium',\n '56': 'bishop of llandaff',\n '57': 'gaura',\n '58': 'geranium',\n '59': 'orange dahlia',\n '6': 'tiger lily',\n '60': 'pink-yellow dahlia',\n '61': 'cautleya spicata',\n '62': 'japanese anemone',\n '63': 'black-eyed susan',\n '64': 'silverbush',\n '65': 'californian poppy',\n '66': 'osteospermum',\n '67': 'spring crocus',\n '68': 'bearded iris',\n '69': 'windflower',\n '7': 'moon orchid',\n '70': 'tree poppy',\n '71': 'gazania',\n '72': 'azalea',\n '73': 'water lily',\n '74': 'rose',\n '75': 'thorn apple',\n '76': 'morning glory',\n '77': 'passion flower',\n '78': 'lotus lotus',\n '79': 'toad lily',\n '8': 'bird of paradise',\n '80': 'anthurium',\n '81': 'frangipani',\n '82': 'clematis',\n '83': 'hibiscus',\n '84': 'columbine',\n '85': 'desert-rose',\n '86': 'tree mallow',\n '87': 'magnolia',\n '88': 'cyclamen',\n '89': 'watercress',\n '9': 'monkshood',\n '90': 'canna lily',\n '91': 'hippeastrum',\n '92': 'bee balm',\n '93': 'ball moss',\n '94': 'foxglove',\n '95': 'bougainvillea',\n '96': 'camellia',\n '97': 'mallow',\n '98': 'mexican petunia',\n '99': 'bromelia'}"
},
"execution_count": 2,
"metadata": {},
"output_type": "execute_result"
}
]
},
{
"metadata": {
"trusted": false
},
"cell_type": "code",
"source": "# ImageFolder does not take the folder name as the label; it labels the folders in alphabetical order with numbers \n# (0,101)\nnames = pd.DataFrame.from_dict(cat_to_name, orient='index', columns=['class'])\nnames=names.sort_index(axis=0)\nnames['labels']=range(102)\nnames.head()",
"execution_count": 3,
"outputs": [
{
"data": {
"text/html": "<div>\n<style scoped>\n .dataframe tbody tr th:only-of-type {\n vertical-align: middle;\n }\n\n .dataframe tbody tr th {\n vertical-align: top;\n }\n\n .dataframe thead th {\n text-align: right;\n }\n</style>\n<table border=\"1\" class=\"dataframe\">\n <thead>\n <tr style=\"text-align: right;\">\n <th></th>\n <th>class</th>\n <th>labels</th>\n </tr>\n </thead>\n <tbody>\n <tr>\n <th>1</th>\n <td>pink primrose</td>\n <td>0</td>\n </tr>\n <tr>\n <th>10</th>\n <td>globe thistle</td>\n <td>1</td>\n </tr>\n <tr>\n <th>100</th>\n <td>blanket flower</td>\n <td>2</td>\n </tr>\n <tr>\n <th>101</th>\n <td>trumpet creeper</td>\n <td>3</td>\n </tr>\n <tr>\n <th>102</th>\n <td>blackberry lily</td>\n <td>4</td>\n </tr>\n </tbody>\n</table>\n</div>",
"text/plain": " class labels\n1 pink primrose 0\n10 globe thistle 1\n100 blanket flower 2\n101 trumpet creeper 3\n102 blackberry lily 4"
},
"execution_count": 3,
"metadata": {},
"output_type": "execute_result"
}
]
},
{
"metadata": {
"trusted": false
},
"cell_type": "code",
"source": "# Transformations\ntrain_data_transforms = transforms.Compose([\n transforms.RandomHorizontalFlip(), # randomly flip and rotate\n transforms.RandomVerticalFlip(),\n transforms.RandomRotation(10),\n transforms.RandomResizedCrop(224),\n transforms.ToTensor(),\n transforms.Normalize((0.485, 0.456, 0.406), \n (0.229, 0.224, 0.225))])\nvalid_data_transforms = transforms.Compose([\n transforms.Resize(256),\n transforms.CenterCrop(224),\n transforms.ToTensor(),\n transforms.Normalize((0.485, 0.456, 0.406), \n (0.229, 0.224, 0.225))])",
"execution_count": 5,
"outputs": []
},
{
"metadata": {
"trusted": false
},
"cell_type": "code",
"source": "# Number of subprocesses to use for data loading\nnum_workers = 0\n\n# Number of samples per batch to load\nbatch_size = 256\n\n# Load the datasets with ImageFolder\ntrain_data = datasets.ImageFolder(train_dir, transform=train_data_transforms)\nvalid_data = datasets.ImageFolder(valid_dir, transform=valid_data_transforms)\n\nvalid_batch_size = len(valid_data) #can be bigger since not optimizing grads\n\n# Using the image datasets and the transforms, define the dataloaders\ntrain_loader = torch.utils.data.DataLoader(train_data, batch_size=batch_size,\n num_workers=num_workers, shuffle=True)\nvalid_loader = torch.utils.data.DataLoader(valid_data, batch_size=valid_batch_size,\n num_workers=num_workers, shuffle=True)\n# Statistics\nprint('Num training images: ', len(train_data))\nprint('Num valid images: ', len(valid_data))",
"execution_count": 6,
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": "Num training images: 7241\nNum valid images: 818\n"
}
]
},
{
"metadata": {
"scrolled": true,
"trusted": false
},
"cell_type": "code",
"source": "class_to_idx = {sorted(train_data.classes)[i]: i for i in range(len(train_data.classes))}\n{k: class_to_idx[k] for k in list(class_to_idx)[:5]}",
"execution_count": 7,
"outputs": [
{
"data": {
"text/plain": "{'1': 0, '10': 1, '100': 2, '101': 3, '102': 4}"
},
"execution_count": 7,
"metadata": {},
"output_type": "execute_result"
}
]
},
{
"metadata": {
"trusted": false
},
"cell_type": "code",
"source": "# Reverse dictionary, used in sanity check\nidx_to_class = {val: key for key, val in class_to_idx.items()}",
"execution_count": 8,
"outputs": []
},
{
"metadata": {
"trusted": false
},
"cell_type": "code",
"source": "# Plot an original image from each training folder, and corresponding labels\n# Choose 60 sub-folders at random\nsub_folders = np.random.choice(len(class_to_idx), 60, replace=False)+1\nimages=[]\nfor folder in sub_folders:\n contents = os.listdir(os.path.join(train_dir,str(folder)))\n images.append(Image.open(train_dir + '/' + str(folder) + '/' + contents[0]))\nfig = plt.figure(figsize=(25, 12))\nfor idx in np.arange(60):\n ax = fig.add_subplot(6, 60/6, idx+1, xticks=[], yticks=[])\n plt.imshow(images[idx])\n title = names['class'][class_to_idx[str(sub_folders[idx])]], names['labels'][class_to_idx[str(sub_folders[idx])]]\n ax.set_title(title[0] + ' ' + str(title[1]))",
"execution_count": 38,
"outputs": [
{
"data": {
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment