{"cells":[{"metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","trusted":true},"cell_type":"code","source":"# This Python 3 environment comes with many helpful analytics libraries installed\n# It is defined by the kaggle/python Docker image: https://github.com/kaggle/docker-python\n# For example, here's several helpful packages to load\n\nimport numpy as np # linear algebra\nimport pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)\n\n# Input data files are available in the read-only \"../input/\" directory\n# For example, running this (by clicking run or pressing Shift+Enter) will list all files under the input directory\n\nimport os\nfor dirname, _, filenames in os.walk('/kaggle/input'):\n    for filename in filenames:\n        print(os.path.join(dirname, filename))\n\n# You can write up to 5GB to the current directory (/kaggle/working/) that gets preserved as output when you create a version using \"Save & Run All\" \n# You can also write temporary files to /kaggle/temp/, but they won't be saved outside of the current session","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"d629ff2d2480ee46fbb7e2d37f6b5fab8052498a","_cell_guid":"79c7e3d0-c299-4dcb-8224-4455121ee9b0","trusted":true},"cell_type":"code","source":"# 此库用于迁移学习\n!pip install pretrainedmodels","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"from PIL import Image\nimport matplotlib.pyplot as plt\nimport numpy as np\nimport pandas as pd\n\nimport torch\nimport fastai\nfrom fastai import vision\n\nimport pretrainedmodels as pm","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"%matplotlib inline","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"!ls /kaggle/input/cassava-disease","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"!unzip ../input/cassava-disease/train.zip","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"!unzip ../input/cassava-disease/test.zip","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"!ls ./train","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"import glob\nimport os\nimport shutil\nfrom typing import List\n\nimport numpy as np\nimport pandas as pd\nfrom sklearn import model_selection\n\nPREFIX = \"./data\"\nSEED = 42\nTEST_SIZE = 0.2\n\n# 转为pd.DataFrame类型\ndef filepaths_to_dataframe(paths: List[str]) -> pd.DataFrame:\n    \"\"\"Converts filepaths to a Pandas DataFrame.\"\"\"\n    results = {\"label\": [], \"filename\": []}\n    for path in paths:\n#         print(path.split('/'))\n        _, _, _label, _ = path.split('/')\n        results[\"label\"].append(_label)\n        results[\"filename\"].append(path)\n    df = (pd.DataFrame\n            .from_dict(results))\n    return df\n\n\ndef make_interim_training_data(prefix: str, df: pd.DataFrame) -> None:\n    if not os.path.isdir(f\"{prefix}/interim/train\"):\n        os.makedirs(f\"{prefix}/interim/train\")\n\n    for _, row in df.iterrows():\n        label, path = row\n        filename = (os.path\n                      .basename(path))\n        if not os.path.isdir(f\"{prefix}/interim/train/{label}\"):\n            os.mkdir(f\"{prefix}/interim/train/{label}\")\n        shutil.copy(path, f\"{prefix}/interim/train/{label}/{filename}\")\n\n        \ndef make_interim_validation_data(prefix: str, df: pd.DataFrame) -> None:\n    if not os.path.isdir(f\"{prefix}/interim/valid\"):\n        os.makedirs(f\"{prefix}/interim/valid\")\n    for _, row in df.iterrows():\n        label, path = row\n        filename = (os.path\n                      .basename(path))\n        if not os.path.isdir(f\"{prefix}/interim/valid/{label}\"):\n            os.mkdir(f\"{prefix}/interim/valid/{label}\")            \n        shutil.copy(path, f\"{prefix}/interim/valid/{label}/{filename}\")\n\n        \nfilepaths = glob.glob(f\"./train/*/*.jpg\", recursive=True)\ndf = filepaths_to_dataframe(filepaths)\nprng = np.random.RandomState(SEED)\n\ntraining_df, validation_df = model_selection.train_test_split(df,\n                                                              test_size=TEST_SIZE,\n                                                              random_state=prng,\n                                                              stratify=df[\"label\"])\n    \nif not os.path.isdir(PREFIX):\n    os.mkdir(PREFIX)\nmake_interim_training_data(PREFIX, training_df)\nmake_interim_validation_data(PREFIX, validation_df)","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"划分文件数量的直方图","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"fig, axes = plt.subplots(1, 3, sharex=True, sharey=True, figsize=(15, 8))\n\n_ = df.loc[:, \"label\"].value_counts().plot.bar(ax=axes[0], title=\"Raw\")\n\n_ = training_df.loc[:, \"label\"].value_counts().plot.bar(ax=axes[1], title=\"Training\")\n\n_ = validation_df.loc[:, \"label\"].value_counts().plot.bar(ax=axes[2], title=\"Validation\")","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"def set_seed(seed):\n    np.random.seed(seed)\n    torch.manual_seed(seed)\n    torch.cuda.manual_seed_all(seed)\n    torch.backends.cudnn.deterministic = True\n\nset_seed(42)","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"创建 ImageDataBunch\n为了创建训练数据集，我们使用标准的数据扩充技术。定义用于数据扩充的转换的所有参数都保留默认值(除非另外指定)。","execution_count":null},{"metadata":{},"cell_type":"markdown","source":"!ls ./data/interim/train/","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"_transform_kwargs = {\"do_flip\": True,\n                     \"flip_vert\": True,  # 默认是 False 用于数据增强，详情fastai文档\n                     \"max_rotate\": 180,  # 默认是 10\n                     \"max_zoom\": 1.2,    # 默认是 1.1\n                     \"max_lighting\": 0.2,\n                     \"max_warp\": 0.2,\n                     \"p_affine\": 0.75,\n                     \"p_lighting\": 0.7,\n                    }\n        \n_transforms = vision.get_transforms(**_transform_kwargs)\n\n_data_bunch_kwargs = {\"path\": \"./data/interim/\",\n                      \"train\": \"train\",\n                      \"valid\": \"valid\",\n                      \"bs\": 16,\n                      \"size\": 448,\n                      \"ds_tfms\": _transforms,\n                      \"test\": \"../../test\",  # 无需复制到./data即可访问测试数据\n                     }\n\nimage_data_bunch = vision.ImageDataBunch.from_folder(**_data_bunch_kwargs).normalize()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"image_data_bunch.train_ds # train 集数据","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"image_data_bunch.valid_ds","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"image_data_bunch.test_ds","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"查看数据\n了解图像外观","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"#image_data_bunch.show_batch(figsize=(20,20))","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"拟合模型","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"from torchvision.models import *\nimport pretrainedmodels\n\nfrom fastai.vision import *\nfrom fastai.vision.models import *\nfrom fastai.vision.learner import model_meta\n\n# from utils import *\nimport sys","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"\ndef identity(x): return x\n\ndef nasnetamobile(pretrained=False):\n    pretrained = 'imagenet' if pretrained else None\n    model = pretrainedmodels.nasnetamobile(pretrained=pretrained, num_classes=1000)  \n    model.logits = identity\n    return nn.Sequential(model)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"model_meta[nasnetamobile] =  { 'cut': noop, \n                               'split': lambda m: (list(m[0][0].children())[8], m[1]) }","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# _base_arch = lambda arg: pm.se_resnext101_32x4d(num_classes=1000, pretrained=\"imagenet\") # 使用se_resnext101_32x4d\n# _base_arch = lambda arg: pm.pnasnet5large(num_classes=1000, pretrained='imagenet') # 使用inceptionv4\n_base_arch = lambda arg: pm.nasnetamobile(num_classes=1000, pretrained='imagenet') # 使用nasnetalarge\nlearner = vision.cnn_learner(image_data_bunch,\n                             base_arch=nasnetamobile,\n                             pretrained=True,                                                                                                 \n                             metrics=[vision.error_rate,vision.accuracy], # 添加指标\n                             model_dir=\"/kaggle/working/models/VGG\",\n                             opt_func=optim.Adam)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"模型结构","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"learner.summary()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# lr_find(learn:Learner, start_lr:Floats=1e-07, end_lr:Floats=10, num_it:int=100, stop_div:bool=True, wd:float=None)\nlearner.lr_find()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"learner.recorder.plot()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"learner.recorder.plot_lr(show_moms=True)","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"采用1cycle policy策略\n参考：A disciplined approach to neural network hyper-parameters: Part 1 -- learning rate, batch size, momentum, and weight decay","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"# 从数据中提取最优的学习率。\ndef find_optimal_lr(recorder):\n    optimal_lr = 0\n    minimum_loss = float(\"inf\")\n    for loss, lr in zip(recorder.losses, recorder.lrs):\n        if loss < minimum_loss:\n            optimal_lr = lr\n            minimum_loss = loss\n    return optimal_lr, minimum_loss\n","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"# 使用其他优化器","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"from fastai.script import *\nfrom fastai.vision import *\nfrom fastai.callbacks import *\nfrom fastai.distributed import *\nfrom fastai.callbacks.tracker import *\ntorch.backends.cudnn.benchmark = True\nimport time\ndef fit_sgd_warm(learn, n_cycles, lr, mom, cycle_len, cycle_mult):\n    n = len(learn.data.train_dl)\n    phases = [(TrainingPhase(n * (cycle_len * cycle_mult**i))\n                 .schedule_hp('lr', lr, anneal=annealing_cos)\n                 .schedule_hp('mom', mom)) for i in range(n_cycles)]\n    sched = GeneralScheduler(learn, phases)\n    learn.callbacks.append(sched)\n    if cycle_mult != 1:\n        total_epochs = int(cycle_len * (1 - (cycle_mult)**n_cycles)/(1-cycle_mult)) \n    else: total_epochs = n_cycles * cycle_len\n    learn.fit(total_epochs)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"fit_sgd_warm(learner, 15, 1e-5, 0.9, 1, 2)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"_save_model_kwargs = {\"every\": \"improvement\",\n                      \"monitor\": \"valid_loss\",\n                      \"name\": \"best-model-stage-1\"}\n_save_model = (fastai.callbacks\n                     .SaveModelCallback(learner, **_save_model_kwargs))\nlearner.fit(15, callbacks=[_save_model])","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"# End","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"# 定义一个存储“最佳”模型状态的回调函数。注意:最好的模型在训练结束时重新加载\n_save_model_kwargs = {\"every\": \"improvement\",\n                      \"monitor\": \"valid_loss\",\n                      \"name\": \"best-model-stage-1\"}\n_save_model = (fastai.callbacks\n                     .SaveModelCallback(learner, **_save_model_kwargs))\nlearner.fit_one_cycle(15, callbacks=[_save_model])","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# 定义一个存储“最佳”模型状态的回调函数。注意:最好的模型在训练结束时重新加载\n_save_model_kwargs = {\"every\": \"improvement\",\n                      \"monitor\": \"valid_loss\",\n                      \"name\": \"best-model-stage-1\"}\n_save_model = (fastai.callbacks\n                     .SaveModelCallback(learner, **_save_model_kwargs))\n  \n# 如果验证损失<培训损失，要么是学习率太低，要么是培训时间不够\nlearner.fit_one_cycle(15, callbacks=[_save_model])","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"learner.recorder.plot_losses()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"learner.recorder.metrics[0][1]","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"learner.recorder.plot_lr(show_moms=True)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"def plotAccuracy(learner):\n    fig,ax = plt.subplots(3,1,figsize=(8,12))\n    ax[0].plot(list(range(15)),learner.recorder.val_losses, label='Validation loss')\n    ax[0].plot(list(range(15)),[learner.recorder.losses[i] for i in range(97,15*98,98)], label='Training loss')\n    ax[0].set_xlabel('Epoch')\n    ax[0].set_ylabel('Loss')\n    ax[0].legend(loc='upper right')\n    ax[1].plot(list(range(15)),[learner.recorder.metrics[i][1] for i in range(15)], label='Accuracy')\n    ax[1].set_xlabel('Epoch')\n    ax[1].set_ylabel('Accuracy')\n\n    ax[2].plot(list(range(15)),[learner.recorder.metrics[i][0] for i in range(15)], label='Error_rate')\n    ax[2].set_xlabel('Epoch')\n    ax[2].set_ylabel('Error_rate')","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"plotAccuracy(learner)","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"模型的预测","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"clf_interp = vision.ClassificationInterpretation.from_learner(learner)","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"显示图像top_losses及其预测，实际，损失和实际分类的概率。\n参考文献","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"#clf_interp.plot_top_losses(16, figsize=(20,20))","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"显示混淆矩阵","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"clf_interp.plot_confusion_matrix()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"clf_interp.plot_confusion_matrix(normalize=True)","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"混淆矩阵的最大非对角条目的排序降序列表，表示为实际的，预测的出现次数。","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"clf_interp.most_confused()","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"解冻，微调和学习率\n与上面步骤差不多，主要进行微调提高准确率","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"learner.unfreeze()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"learner.summary()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"learner.lr_find()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"learner.recorder.plot()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# 保存模型\n_save_model_kwargs = {\"every\": \"improvement\",\n                      \"monitor\": \"valid_loss\",\n                      \"name\": \"best-model-stage-2\"}\n_save_model = (fastai.callbacks\n                     .SaveModelCallback(learner, **_save_model_kwargs))\nlearner.fit_one_cycle(15, max_lr=slice(1e-6, 1e-4), callbacks=[_save_model])","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"clf_interp.plot_top_losses(16, figsize=(20,20))","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"learner.recorder.plot_losses()","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"探索微调模型的结果","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"clf_interp = (vision.ClassificationInterpretation\n                    .from_learner(learner))","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"clf_interp.plot_confusion_matrix()","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"Test-Time Augmentation (TTA)\n可将准确率提高若干个百分点，它就是测试时增强（test time augmentation, TTA）。这里会为原始图像造出多个不同版本，包括不同区域裁剪和更改缩放程度等，并将它们输入到模型中；然后对多个版本进行计算得到平均输出，作为图像的最终输出分数。\n\n这种技术很有效，因为原始图像显示的区域可能会缺少一些重要特征，在模型中输入图像的多个版本并取平均值，能解决上述问题。","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"predicted_class_probabilities, _ = learner.TTA(ds_type=fastai.basic_data.DatasetType.Test)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"_predicted_classes = (predicted_class_probabilities.argmax(dim=1)\n                                                   .numpy())\n_class_labels = np.array(['cbb','cbsd','cgm','cmd','healthy'])\n_predicted_class_labels = _class_labels[_predicted_classes]\n\n_filenames = np.array([item.name for item in image_data_bunch.test_ds.items])\n\nsubmission = (pd.DataFrame\n                .from_dict({'Category': _predicted_class_labels,'Id': _filenames}))","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"submission.head()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"submission.to_csv('submission.csv', header=True, index=False)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"shutil.rmtree(PREFIX)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"learner.export() ","execution_count":null,"outputs":[]}],"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"pygments_lexer":"ipython3","nbconvert_exporter":"python","version":"3.6.4","file_extension":".py","codemirror_mode":{"name":"ipython","version":3},"name":"python","mimetype":"text/x-python"}},"nbformat":4,"nbformat_minor":4}