{"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":"3.12.4"},"accelerator":"GPU","colab":{"gpuType":"T4","provenance":[]},"kaggle":{"accelerator":"none","dataSources":[{"sourceId":96000,"databundleVersionId":11397519,"sourceType":"competition"}],"isInternetEnabled":true,"language":"python","sourceType":"notebook","isGpuEnabled":false}},"nbformat_minor":4,"nbformat":4,"cells":[{"cell_type":"markdown","source":"# Соревнование Kaggle \"Классификация фотографий бабочек\"","metadata":{"id":"oRZ0_j6j15_y"}},{"cell_type":"markdown","source":"## 1. Описание задачи и данные","metadata":{"id":"AEbHEJZR2eWd"}},{"cell_type":"markdown","source":"![butter_kaggle.png](data:image/png;base64,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)","metadata":{"id":"XxYN6CdNuvu_"}},{"cell_type":"markdown","source":"В рамках соревнования необходимо решить задачу классификации фотографий бабочек на 50 классов. Каждая фотография представляет из себя цветное изображение размера 224×224.\n\nДаны два набора данных:\n\n- Папка train содержит в себе 50 папок, разбитых на классы. Всего порядка 5000 фото.\n- Папка test содержит в себе 250 фотографий, которые необходимо классифицировать.\n- Используется метрика accuracy.\n\nСсылка на соревнование и на данные: https://www.kaggle.com/competitions/butterflies-classification","metadata":{"id":"2jYw5v3M2qhc"}},{"cell_type":"code","source":"# Установим библиотеку для работы с датасетами на kaggle\n!pip install opendatasets --quiet","metadata":{"id":"41M9EPrnPdnm"},"outputs":[],"execution_count":null},{"cell_type":"code","source":"import numpy as np\nimport pandas as pd\nimport matplotlib.pyplot as plt\n\nimport opendatasets as od","metadata":{"id":"xW9KirnC2kuO"},"outputs":[],"execution_count":null},{"cell_type":"code","source":"# Подгружаем датасет\nurl = 'https://www.kaggle.com/competitions/butterflies-classification'\nod.download(url)","metadata":{"id":"a1vqtoe626dY","outputId":"edd15726-d5db-4a18-f748-a4deae837253"},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"## 1. Предварительное исследование данных\n\n- Определяем баланс изображений внутри классов\n- Описываем особенности датасета\n- Оцениваем, нужна ли очистка и предобработка данных","metadata":{"id":"gOLWe4Df3KpZ"}},{"cell_type":"code","source":"import os\nfrom PIL import Image\nfrom collections import Counter\n\nimport torch\nimport torch.nn as nn\nfrom torch.optim import Adam\nfrom torchvision import datasets, transforms, io, utils, models\nfrom torchvision.datasets import ImageFolder\nfrom torch.utils.data import DataLoader, Dataset\n\nfrom sklearn.model_selection import train_test_split\n\nfrom tqdm import tqdm","metadata":{"id":"fjCOYMQARaz2"},"outputs":[],"execution_count":null},{"cell_type":"code","source":"HOME_DIR = './butterflies-classification/'\nTRAIN_DIR = HOME_DIR + 'train_butterflies/train_split'\nTEST_DIR = HOME_DIR + 'test_butterflies/valid/'\n\n# Подгружаем размеченные данные\nbase_ds = ImageFolder(root=TRAIN_DIR)\n\nclasses = base_ds.classes\nclassidx = base_ds.class_to_idx\ncounts = Counter(base_ds.targets)\n\n# 50 видов бабочек\nprint(classes)\n\ncounts_df = pd.Series(counts).sort_index().rename({\n    i: classes[i] for i in range(len(classes))\n})\ncounts_vals = np.array(list(counts.values()))\n\n# картинки распределены неравномерно по категориям,\n# в каждой mean=99, std=15\n#print(counts_df)\nprint(np.mean(counts_vals), np.std(counts_vals))\n\nwidths, heights = set(), set()\nnum_images = 0\nfor root, _, files in os.walk(TRAIN_DIR):\n    for fname in files:\n        path = os.path.join(root, fname)\n        with Image.open(path) as img:\n            w, h = img.size\n            widths.add(w)\n            heights.add(h)\n        num_images += 1\n\n# все картинки в разрешении 224x224\nprint(widths, heights)\n\n# всего 4955 картинок\nprint(num_images)\n\n\"\"\"\nОсобенности датасета:\n\n1. У бабочки могут быть сложены или раскрыты\nкрылья — задача классификации усложняется,\nнейросеть может ошибчно учитывать количество\nкрыльев, которые видны: одно или два.\n\n2. Окрас и рисунок на разных сторонах крыльев\nможет быть разный. Поэтому изображения могут\nзначительно различаться даже внутри классов.\nНапример, в class_10 это серый и бурый цвета.\n\n3. Бабочки расположены под разными углами\nи ракурсами, яркость картинок также различается.\nЗначит, уместна соответствующая аугментация\n\n4. У картинок есть фон, в основном это зелень.\nСтоит его исключать, чтобы он не мешал\nклассификации бабочек. То есть это задача\nсегментации, можно использовать простые\nпороговые методы\n\n5. Датасет несбалансирован, как выяснили в\nкоде выше. Картинок одних классов ~70,\nа других — ~130\n\n6. Отдельного упоминания требует class_39.\nОн содержит картинки, на которых бабочки\nнапоминают небольших жуков. Причём бабочки\nэти много меньше цветков, на которых\nони запечатлены. Нейросеть может ошибочно\nклассифицировать фотографии с цветами как\nэкземпляры class_39\n\n7. Цвет важен при классификации бабочек,\nпоэтому лучше не аугментировать по нему\n\nБез аугментации и сегментации нейросеть\nпоказывает accuracy на тесте, близкое \nк 100%\n\"\"\"","metadata":{"id":"RvTy75-y3ljg","outputId":"9d6cf11b-6328-4f5e-d268-b3c62363a58d"},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"## 2. Подготовка данных для обучения","metadata":{"id":"8Y7alZSf3rAC"}},{"cell_type":"markdown","source":"- Создать тренировочную и тестовую подвыборки для обучения и тестирования работы моделей\n- Создать класс на фреймворке `PyTorch` для создания датасета\n- Создать `loader` для передачи данных в модель нейронной сети","metadata":{"id":"MbSx5XVa3w61"}},{"cell_type":"code","source":"class MyDataset(Dataset):\n    def __init__(self, path, labels, transform=None):\n        self.paths = np.array(path)\n        self.labels = np.array(labels)\n        self.transform = transform\n\n    def __len__(self):\n        return len(self.paths)\n\n    def __getitem__(self, idx):\n        img = Image.open(self.paths[idx]).convert('RGB')\n        label = self.labels[idx]\n\n        if self.transform is not None:\n            img = self.transform(img)\n\n        return (img, label)","metadata":{"id":"f3wAd-n43wXr"},"outputs":[],"execution_count":null},{"cell_type":"code","source":"samples = base_ds.samples\npaths, labels = zip(*samples)\n\n# Разделили размеченные данные на тренировочную и тестовую выборки\n# stratify=labels для равномерности количества экземпляров по классам\ntrain_paths, valid_paths, train_labels, valid_labels = train_test_split(\n    paths, labels, test_size=0.2, stratify=labels\n)\n\n# Трансформация для использования нейросетей на основе ImageNet\ntest_transform = transforms.Compose([\n    transforms.Resize((224, 224)),\n    transforms.ToTensor(),\n    transforms.Normalize(\n        mean=[0.485, 0.456, 0.406],\n        std=[0.229, 0.224, 0.225]\n    ),\n])\n\n# Такая же трансформация, что и для теста.\n# Две переменные заведены, для более\n# удобной аугментации, но в итоге она\n# значительно не повышала точность\ntrain_transform = transforms.Compose([\n    transforms.Resize((224, 224)),\n    transforms.ToTensor(),\n    transforms.Normalize(\n        mean=[0.485, 0.456, 0.406],\n        std=[0.229, 0.224, 0.225]\n    ),\n])\n\ntrain_data = MyDataset(train_paths, train_labels, transform=train_transform)\nvalid_data = MyDataset(valid_paths, valid_labels, transform=train_transform)\n\n# перемешивание shuffle=True, чтобы нейросеть не запоминала\n# порядок индексов в тренировочной выборке\ntrain_loader = DataLoader(\n    train_data, batch_size=32, shuffle=True, num_workers=2\n)\nvalid_loader = DataLoader(\n    valid_data, batch_size=32, shuffle=False, num_workers=2\n)","metadata":{"id":"-omNDt8f4PUs"},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"## 3. Реализация модели\n\n- Создать класс для модели\n- Создать функцию для обучения модели\n- Создать функцию для валидации модели","metadata":{"id":"MkAU6J_L4ZgK"}},{"cell_type":"code","source":"# Используем предобученную свёрточную нейронную сеть типа ImageNet\nclass RSNAModel(nn.Module):\n    def __init__(self, num_classes, backbone_name='resnet18'):\n        super().__init__()\n\n        if backbone_name == 'resnet18':\n            self.backbone = models.resnet18(pretrained=True)\n            in_features = self.backbone.fc.in_features\n            self.backbone.fc = nn.Linear(in_features, num_classes)\n        elif backbone_name == 'resnet50':\n            self.backbone = models.resnet50(pretrained=True)\n            in_features = self.backbone.fc.in_features\n            self.backbone.fc = nn.Linear(in_features, num_classes)\n        elif backbone_name == 'efficientnet_b0':\n            self.backbone = models.efficientnet_b0(pretrained=True)\n            in_features = self.backbone.classifier[1].in_features\n            self.backbone.classifier[1] = nn.Linear(in_features, num_classes)\n        else:\n            raise ValueError(f'Unknown nn: {backbone_name}')\n\n    def forward(self, batch):\n        return self.backbone(batch)","metadata":{"id":"0foopS2P4rLh"},"outputs":[],"execution_count":null},{"cell_type":"code","source":"def train_one_epoch(\n    model,\n    train_dataloader,\n    optimizer,\n    loss_fn,\n    epoch,\n    device,\n    log_wandb=False,\n    verbose=False\n):\n    model.train()\n\n    # Инициализируем счётчики потерь, правильных предсказаний\n    # и числа образцов\n    running_loss = 0\n    running_corrects = 0\n    total_samples = 0\n\n    # Свой прогресс-бар через tqdm для батчей\n    pbar = tqdm(\n        enumerate(train_dataloader),\n        total=len(train_dataloader),\n        desc=f'train epoch: {epoch}'\n    )\n\n    for batch_idx, (inputs, labels) in pbar:\n        inputs = inputs.to(device)\n        labels = labels.to(device)\n\n        outputs = model(inputs)\n        loss = loss_fn(outputs, labels)\n\n        # Обнуляем градиенты, обратное распр. ошибки и оптимизация\n        optimizer.zero_grad()\n        loss.backward()\n        optimizer.step()\n\n        # Считаем метрики\n        running_loss += loss.item() * inputs.size(0)\n        preds = outputs.argmax(dim=1)\n        running_corrects += (preds == labels).sum().item()\n        total_samples += inputs.size(0)\n\n        # Выводим результаты\n        if verbose:\n            pbar.set_postfix({\n                'train_loss': f'{running_loss/total_samples:.4f}',\n                'train_acc': f'{running_corrects/total_samples:.4f}'\n            })\n\n    epoch_loss = running_loss / total_samples\n    epoch_accuracy = running_corrects / total_samples\n\n    return epoch_loss, epoch_accuracy\n\n\n# Аналогично\n@torch.no_grad()\ndef valid_one_epoch(\n    model,\n    valid_dataloader,\n    loss_fn,\n    epoch,\n    device,\n    log_wandb=False,\n    verbose=False\n):\n    model.eval()\n\n    running_loss = 0\n    running_corrects = 0\n    total_samples = 0\n\n    pbar = tqdm(\n        enumerate(valid_dataloader),\n        total=len(valid_dataloader),\n        desc=f'valid epoch: {epoch}'\n    )\n\n    for batch_idx, (inputs, labels) in pbar:\n        inputs = inputs.to(device)\n        labels = labels.to(device)\n\n        outputs = model(inputs)\n        loss = loss_fn(outputs, labels)\n\n        running_loss += loss.item() * inputs.size(0)\n        preds = outputs.argmax(dim=1)\n        running_corrects += (preds == labels).sum().item()\n        total_samples += inputs.size(0)\n\n        if verbose:\n            pbar.set_postfix({\n                'val_loss': f'{running_loss/total_samples:.4f}',\n                'val_acc': f'{running_corrects/total_samples:.4f}'\n            })\n\n    epoch_loss = running_loss / total_samples\n    epoch_accuracy = running_corrects / total_samples\n\n    return epoch_loss, epoch_accuracy","metadata":{"id":"05KUACqv4zlx"},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"## 4. Обучение модели\n*   Определяем гиперпараметры модели\n*   Обучаем модель\n*   Оцениваем эффективность её работы \n\n","metadata":{"id":"kohYYP3N5GkT"}},{"cell_type":"code","source":"# Гиперпараметры\nNUM_CLASSES = 50\nLR = 1e-4\nEPOCHS = 5\nDEVICE  = torch.device('cuda' if torch.cuda.is_available() else 'cpu')","metadata":{"id":"af73F7U89aDY"},"outputs":[],"execution_count":null},{"cell_type":"code","source":"# Модель\nmodel = RSNAModel(num_classes=NUM_CLASSES).to(DEVICE)\nloss_fn = nn.CrossEntropyLoss() # задача классификации\noptimizer = Adam(model.parameters(), lr=LR)\n\n# Цикл обучения\nfor epoch in range(1, EPOCHS + 1):\n    train_loss, train_acc = train_one_epoch(\n        model, train_loader, optimizer, loss_fn, epoch, DEVICE\n    )\n    valid_loss, valid_acc = valid_one_epoch(\n        model, valid_loader, loss_fn, epoch, DEVICE\n    )\n    print(\n        f'epoch: {epoch:02d}, train_loss={train_loss:.4f}, '\n        f'train_acc={train_acc:.4f}, valid_loss={valid_loss:.4f}, '\n        f'valid_acc={valid_acc:.4f}'\n    )\n\n\"\"\"\nResNet18 достаточно быстро переобучается: train_acc=0.9990,\nloss на тренировочной выборке в 10 раз меньше, чем на\nвалидации. С валидацией справляется достаточно хорошо,\nточность 0.9435\n\"\"\"","metadata":{"id":"SMN4I1wB5jtP","outputId":"70eec2d1-3b63-458c-d893-e68587093b6e"},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"## 5. Выбор наилучшей модели \n\n- Тестируем 3 разных модели для решения задачи\n- Сравниваем их эффективность (по метрикам и потерям)\n- Анализируем результат","metadata":{"id":"Jo1g3CnF5Xyp"}},{"cell_type":"code","source":"model_names = ['resnet18', 'resnet50', 'efficientnet_b0']\nresults = []\n\nfor name in model_names:\n    print(f'--- Training model: {name} ---')\n    model = RSNAModel(num_classes=NUM_CLASSES, backbone_name=name).to(DEVICE)\n    optimizer = Adam(model.parameters(), lr=LR)\n\n\n    for epoch in range(1, EPOCHS + 1):\n        train_loss, train_acc = train_one_epoch(\n            model, train_loader, optimizer, loss_fn, epoch, DEVICE\n        )\n        valid_loss, valid_acc = valid_one_epoch(\n            model, valid_loader, loss_fn, epoch, DEVICE\n        )\n        print(\n            f'{name} | epoch {epoch:02d} | '\n            f'train_acc: {train_acc:.4f} | val_acc: {valid_acc:.4f}'\n        )\n\n    results.append((name, train_acc, valid_acc))","metadata":{"id":"EXV2-qaG5k6G","outputId":"ade324b3-e4cd-49a9-98c7-ac00c98b69da"},"outputs":[],"execution_count":null},{"cell_type":"code","source":"df_results = pd.DataFrame(results, columns=['model', 'train_acc', 'valid_acc'])\ndf_sorted = df_results.sort_values('valid_acc', ascending=False).reset_index(drop=True)\ndf_sorted","metadata":{"id":"zciXDoayaXQH","outputId":"5deabf14-2c7b-459d-b33e-0c62cfece46e"},"outputs":[],"execution_count":null},{"cell_type":"code","source":"\"\"\"\nСравним нейросети:\n\nResNet50 самая медленная, зато с высокой валидационной точностью.\nИмеет много параметров. valid_loss = 0.02, train_loss=0.17,\nто есть хорошо сходится\n\nEfficientNet-B0 более простая, но по валидации как ResNet50.\ntrain_loss=0.05, valid_loss как у ResNet50\n\nResNet18 самая быстрая, но слишком быстро запоминает train:\nточность на тренировочной выборке 0.9995, низкий train_loss,\nто есть явное переобучение. Валидационная точность чуть хуже,\nчем у других моделей\n\n*значения для случая, если в train_test_split задали seed=42.\nИногда EfficientNet-B0 показывает наиболее стабильный результат:\nне переобучается и показывает наибольшую точность на valid\n\"\"\"","metadata":{"id":"JY6TW7bfdYov","outputId":"32f9233f-433a-480b-c37f-93fa6bab1f2c"},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"# 6. Тест на Kaggle\n\n- Реализуем класс датасета с тестовыми данными для соревнования\n- Применяем лучшую из обученных моделей для классификации\n- Отправляем результаты в систему проверок Kaggle","metadata":{"id":"WNerHrfwgq09"}},{"cell_type":"code","source":"class TestDataset(Dataset):\n    def __init__(self, path, transform=None):\n        self.paths = sorted(\n            [\n                os.path.join(path, fname)\n                for fname in os.listdir(path)\n            ],\n            key=lambda x: int(os.path.splitext(os.path.basename(x))[0])\n        )\n        self.transform = transform\n\n    def __len__(self):\n        return len(self.paths)\n\n    def __getitem__(self, idx):\n        img = Image.open(self.paths[idx]).convert('RGB')\n\n        if self.transform is not None:\n            img = self.transform(img)\n\n        return img","metadata":{"id":"OeZH4RsKhBlv"},"outputs":[],"execution_count":null},{"cell_type":"code","source":"test_data = TestDataset(TEST_DIR, transform=test_transform)\ntest_loader = DataLoader(test_data, batch_size=32, shuffle=False)","metadata":{"id":"03IH1QRbi-iR"},"outputs":[],"execution_count":null},{"cell_type":"code","source":"@torch.no_grad()\ndef predict(model, dataloader, device):\n    model.eval()\n    preds = []\n\n    for batch in tqdm(dataloader):\n        batch = batch.to(device)\n        outputs = model(batch)\n        batch_preds = outputs.argmax(dim=1).cpu().numpy()\n        preds.extend(batch_preds)\n\n    return preds","metadata":{"id":"gKpEzt6Gjgme"},"outputs":[],"execution_count":null},{"cell_type":"code","source":"# Используем ResNet50 как лучшую на valid среди протестированных нейросетей.\n# Обучаем 10 эпох, что долго, зато добиваемся наилучших весов и сохраняем их\nmodel = RSNAModel(num_classes=NUM_CLASSES, backbone_name='resnet50').to(DEVICE)\nloss_fn = nn.CrossEntropyLoss()\noptimizer = Adam(model.parameters(), lr=LR)\n\nbest_acc = 0.0\n\nfor epoch in range(1, 10 + 1):\n    train_loss, train_acc = train_one_epoch(\n        model, train_loader, optimizer, loss_fn, epoch, DEVICE\n    )\n    valid_loss, valid_acc = valid_one_epoch(\n        model, valid_loader, loss_fn, epoch, DEVICE\n    )\n    print(\n        f'epoch: {epoch:02d}, train_loss={train_loss:.4f}, '\n        f'train_acc={train_acc:.4f}, valid_loss={valid_loss:.4f}, '\n        f'valid_acc={valid_acc:.4f}'\n    )\n\n    if valid_acc > best_acc:\n        best_acc = valid_acc\n        torch.save(model.state_dict(), 'best_model.pth')\n\n# Наилучшие веса сохранены в файл best_model.pth","metadata":{"id":"DGjKUQCEj-0g","outputId":"df8058e1-7acd-4127-b095-5bb73e29d8e1"},"outputs":[],"execution_count":null},{"cell_type":"code","source":"# Предсказываем test, то есть 250 неразмеченных картинок\nmodel.load_state_dict(torch.load('best_model.pth'))\nmodel.to(DEVICE)\ntest_preds = predict(model, test_loader, DEVICE)","metadata":{"id":"KESwy9OQlpGh","outputId":"3d9cfccd-92ce-49db-8373-eff17a7b1269"},"outputs":[],"execution_count":null},{"cell_type":"code","source":"# \"Индексы\" файлов в тесте. По сути список 0, ..., 249,\n# но реализация более обобщённая\ntest_file_ids = [\n    int(os.path.splitext(os.path.basename(p))[0])\n    for p in test_data.paths\n]\n\n# Сопоставляем class_XX и XX\nclass_names = base_ds.classes\ntest_labels = [\n    int(class_names[idx].split('_')[1])\n    for idx in test_preds\n]\n\nresults = pd.DataFrame({\n    'index': test_file_ids,\n    'label': test_labels\n})\n\n# Сохраняем предсказания в файл для Kaggle\nresults.to_csv('submission.csv', index=False)\nprint(results.head())","metadata":{"id":"AZdR9b8br6OV","outputId":"3da0b0c2-571a-4760-b56d-20c5e28ee107"},"outputs":[],"execution_count":null},{"cell_type":"code","source":"!kaggle competitions submit butterflies-classification -f submission.csv -m \"My submission\"","metadata":{"id":"f-_O6o0hrEcy","outputId":"59c29432-666e-4471-e94b-fe9ebf6344ae"},"outputs":[],"execution_count":null}]}