{
  "id": 101952,
  "title": "Kernel threw exception",
  "url": "/competitions/aptos2019-blindness-detection/discussion/101952",
  "author_name": "leixiang@AInnovation",
  "post_date": "2019-07-30T03:18:05.739000",
  "votes": -1,
  "comment_count": 0,
  "views": 0,
  "content": "<p>i why many times however still kernel threw exception error, could anyone help me check the code?</p>\n\n<p>import cv2\nimport numpy as np\nimport pandas as pd</p>\n\n<h1>fast ai</h1>\n\n<p>import fastai\nfrom fastai import *\nfrom fastai.vision import *\nfrom fastai.callbacks import *\nfrom fastai.basic_train import *\nfrom fastai.vision.learner import *</p>\n\n<p>from tqdm import tqdm\nfrom os.path import isfile\nfrom torchvision import transforms\nfrom torch.utils.data import Dataset</p>\n\n<p>print(os.listdir('../input/aptos2019-blindness-detection/'))</p>\n\n<h1>dataset definition</h1>\n\n<p>class MyDataset(Dataset):\n    def <strong>init</strong>(self, df_data, image_folder, transform=None):\n        self.df = df_data.values\n        self.image_folder = image_folder\n        self.transform = transform</p>\n\n<pre><code>def __len__(self):\n    return len(self.df)\n\ndef __getitem__(self, idx):\n    image_name, label = self.df[idx]\n    # image preprocessing\n    image_path = os.path.join(self.image_folder, image_name + '.png')\n    image = cv2.imread(image_path)\n    if self.transform:\n        image = self.transform(image)\n\n    return image, label\n</code></pre>\n\n<h1>test image transforms(to tensor and normalization)</h1>\n\n<p>image_size = 330\ncrop_size = 300\ntest_transform = transforms.Compose([\n    transforms.ToPILImage(),\n    transforms.Resize((image_size, image_size)),\n    # center crop\n    transforms.CenterCrop(crop_size),\n    transforms.ToTensor(),\n    transforms.Normalize(mean=(0.485, 0.456, 0.406), std=(0.229, 0.224, 0.225))])</p>\n\n<p>test_csv = pd.read_csv('../input/aptos2019-blindness-detection/test.csv')\ntest_csv['diagnosis'] = -1\ntest_folder = '../input/aptos2019-blindness-detection/test_images/'\ntest_data = MyDataset(test_csv, test_folder, transform=test_transform)\ntest_loader = torch.utils.data.DataLoader(test_data, batch_size=16, shuffle=False)</p>\n\n<h1>model definition</h1>\n\n<p>learn = load_learner('../input/regressionmodel2/')\nmodel = learn.model</p>\n\n<h1>for param in model.parameters():</h1>\n\n<h1>param.requires_grad = False</h1>\n\n<h2>simple TTA</h2>\n\n<h1>prediction</h1>\n\n<p>preds_1 = predict(model, testloader=test_loader)</p>\n\n<h1>preds_2 = np.array(predict(model, testloader=test_loader))</h1>\n\n<h1>preds_3 = np.array(predict(model, testloader=test_loader))</h1>\n\n<h1>preds_4 = np.array(predict(model, testloader=test_loader))</h1>\n\n<h1>preds_5 = np.array(predict(model, testloader=test_loader))</h1>\n\n<h1>roundoff</h1>\n\n<p>preds_1 = kappa_opt.predict(preds_1)</p>\n\n<h1>print(preds_1)</h1>\n\n<h1>preds_2 = kappa_opt.predict(preds_2)</h1>\n\n<h1>preds_3 = kappa_opt.predict(preds_3)</h1>\n\n<h1>preds_4 = kappa_opt.predict(preds_4)</h1>\n\n<h1>preds_5 = kappa_opt.predict(preds_5)</h1>\n\n<p>#</p>\n\n<h1>simple model ensemble</h1>\n\n<h1>preds_avg = (preds_1 + preds_2 + preds_3 + preds_4 + preds_5)</h1>\n\n<h1>submission</h1>\n\n<p>sample_sub = pd.read_csv('../input/aptos2019-blindness-detection/sample_submission.csv')\nsample_sub.diagnosis = preds_1\nsample_sub.diagnosis = sample_sub['diagnosis'].astype(int)</p>\n\n<h1>print(sample_sub.head(10))</h1>\n\n<p>sample_sub.to_csv('submission.csv', index=False)</p>",
  "messages": [
    {
      "id": 588020,
      "postDate": "2019-07-30T03:18:05.740Z",
      "content": "<p>i why many times however still kernel threw exception error, could anyone help me check the code?</p>\n\n<p>import cv2\nimport numpy as np\nimport pandas as pd</p>\n\n<h1>fast ai</h1>\n\n<p>import fastai\nfrom fastai import *\nfrom fastai.vision import *\nfrom fastai.callbacks import *\nfrom fastai.basic_train import *\nfrom fastai.vision.learner import *</p>\n\n<p>from tqdm import tqdm\nfrom os.path import isfile\nfrom torchvision import transforms\nfrom torch.utils.data import Dataset</p>\n\n<p>print(os.listdir('../input/aptos2019-blindness-detection/'))</p>\n\n<h1>dataset definition</h1>\n\n<p>class MyDataset(Dataset):\n    def <strong>init</strong>(self, df_data, image_folder, transform=None):\n        self.df = df_data.values\n        self.image_folder = image_folder\n        self.transform = transform</p>\n\n<pre><code>def __len__(self):\n    return len(self.df)\n\ndef __getitem__(self, idx):\n    image_name, label = self.df[idx]\n    # image preprocessing\n    image_path = os.path.join(self.image_folder, image_name + '.png')\n    image = cv2.imread(image_path)\n    if self.transform:\n        image = self.transform(image)\n\n    return image, label\n</code></pre>\n\n<h1>test image transforms(to tensor and normalization)</h1>\n\n<p>image_size = 330\ncrop_size = 300\ntest_transform = transforms.Compose([\n    transforms.ToPILImage(),\n    transforms.Resize((image_size, image_size)),\n    # center crop\n    transforms.CenterCrop(crop_size),\n    transforms.ToTensor(),\n    transforms.Normalize(mean=(0.485, 0.456, 0.406), std=(0.229, 0.224, 0.225))])</p>\n\n<p>test_csv = pd.read_csv('../input/aptos2019-blindness-detection/test.csv')\ntest_csv['diagnosis'] = -1\ntest_folder = '../input/aptos2019-blindness-detection/test_images/'\ntest_data = MyDataset(test_csv, test_folder, transform=test_transform)\ntest_loader = torch.utils.data.DataLoader(test_data, batch_size=16, shuffle=False)</p>\n\n<h1>model definition</h1>\n\n<p>learn = load_learner('../input/regressionmodel2/')\nmodel = learn.model</p>\n\n<h1>for param in model.parameters():</h1>\n\n<h1>param.requires_grad = False</h1>\n\n<h2>simple TTA</h2>\n\n<h1>prediction</h1>\n\n<p>preds_1 = predict(model, testloader=test_loader)</p>\n\n<h1>preds_2 = np.array(predict(model, testloader=test_loader))</h1>\n\n<h1>preds_3 = np.array(predict(model, testloader=test_loader))</h1>\n\n<h1>preds_4 = np.array(predict(model, testloader=test_loader))</h1>\n\n<h1>preds_5 = np.array(predict(model, testloader=test_loader))</h1>\n\n<h1>roundoff</h1>\n\n<p>preds_1 = kappa_opt.predict(preds_1)</p>\n\n<h1>print(preds_1)</h1>\n\n<h1>preds_2 = kappa_opt.predict(preds_2)</h1>\n\n<h1>preds_3 = kappa_opt.predict(preds_3)</h1>\n\n<h1>preds_4 = kappa_opt.predict(preds_4)</h1>\n\n<h1>preds_5 = kappa_opt.predict(preds_5)</h1>\n\n<p>#</p>\n\n<h1>simple model ensemble</h1>\n\n<h1>preds_avg = (preds_1 + preds_2 + preds_3 + preds_4 + preds_5)</h1>\n\n<h1>submission</h1>\n\n<p>sample_sub = pd.read_csv('../input/aptos2019-blindness-detection/sample_submission.csv')\nsample_sub.diagnosis = preds_1\nsample_sub.diagnosis = sample_sub['diagnosis'].astype(int)</p>\n\n<h1>print(sample_sub.head(10))</h1>\n\n<p>sample_sub.to_csv('submission.csv', index=False)</p>",
      "rawMarkdown": "i why many times however still kernel threw exception error, could anyone help me check the code?\n\nimport cv2\nimport numpy as np\nimport pandas as pd\n\n# fast ai\nimport fastai\nfrom fastai import *\nfrom fastai.vision import *\nfrom fastai.callbacks import *\nfrom fastai.basic_train import *\nfrom fastai.vision.learner import *\n\nfrom tqdm import tqdm\nfrom os.path import isfile\nfrom torchvision import transforms\nfrom torch.utils.data import Dataset\n\nprint(os.listdir('../input/aptos2019-blindness-detection/'))\n\n# dataset definition\nclass MyDataset(Dataset):\n    def __init__(self, df_data, image_folder, transform=None):\n        self.df = df_data.values\n        self.image_folder = image_folder\n        self.transform = transform\n\n    def __len__(self):\n        return len(self.df)\n\n    def __getitem__(self, idx):\n        image_name, label = self.df[idx]\n        # image preprocessing\n        image_path = os.path.join(self.image_folder, image_name + '.png')\n        image = cv2.imread(image_path)\n        if self.transform:\n            image = self.transform(image)\n\n        return image, label\n\n\n# test image transforms(to tensor and normalization)\nimage_size = 330\ncrop_size = 300\ntest_transform = transforms.Compose([\n    transforms.ToPILImage(),\n    transforms.Resize((image_size, image_size)),\n    # center crop\n    transforms.CenterCrop(crop_size),\n    transforms.ToTensor(),\n    transforms.Normalize(mean=(0.485, 0.456, 0.406), std=(0.229, 0.224, 0.225))])\n\ntest_csv = pd.read_csv('../input/aptos2019-blindness-detection/test.csv')\ntest_csv['diagnosis'] = -1\ntest_folder = '../input/aptos2019-blindness-detection/test_images/'\ntest_data = MyDataset(test_csv, test_folder, transform=test_transform)\ntest_loader = torch.utils.data.DataLoader(test_data, batch_size=16, shuffle=False)\n\n# model definition\nlearn = load_learner('../input/regressionmodel2/')\nmodel = learn.model\n\n# for param in model.parameters():\n#     param.requires_grad = False\n\n## simple TTA\n# prediction\npreds_1 = predict(model, testloader=test_loader)\n# preds_2 = np.array(predict(model, testloader=test_loader))\n# preds_3 = np.array(predict(model, testloader=test_loader))\n# preds_4 = np.array(predict(model, testloader=test_loader))\n# preds_5 = np.array(predict(model, testloader=test_loader))\n\n\n# roundoff\npreds_1 = kappa_opt.predict(preds_1)\n# print(preds_1)\n# preds_2 = kappa_opt.predict(preds_2)\n# preds_3 = kappa_opt.predict(preds_3)\n# preds_4 = kappa_opt.predict(preds_4)\n# preds_5 = kappa_opt.predict(preds_5)\n\n#\n\n# simple model ensemble\n# preds_avg = (preds_1 + preds_2 + preds_3 + preds_4 + preds_5)\n\n# submission\nsample_sub = pd.read_csv('../input/aptos2019-blindness-detection/sample_submission.csv')\nsample_sub.diagnosis = preds_1\nsample_sub.diagnosis = sample_sub['diagnosis'].astype(int)\n\n# print(sample_sub.head(10))\n\nsample_sub.to_csv('submission.csv', index=False)",
      "votes": -1
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "588020": "i why many times however still kernel threw exception error, could anyone help me check the code?\n\nimport cv2\nimport numpy as np\nimport pandas as pd\n\n# fast ai\nimport fastai\nfrom fastai import *\nfrom fastai.vision import *\nfrom fastai.callbacks import *\nfrom fastai.basic_train import *\nfrom fastai.vision.learner import *\n\nfrom tqdm import tqdm\nfrom os.path import isfile\nfrom torchvision import transforms\nfrom torch.utils.data import Dataset\n\nprint(os.listdir('../input/aptos2019-blindness-detection/'))\n\n# dataset definition\nclass MyDataset(Dataset):\n    def __init__(self, df_data, image_folder, transform=None):\n        self.df = df_data.values\n        self.image_folder = image_folder\n        self.transform = transform\n\n    def __len__(self):\n        return len(self.df)\n\n    def __getitem__(self, idx):\n        image_name, label = self.df[idx]\n        # image preprocessing\n        image_path = os.path.join(self.image_folder, image_name + '.png')\n        image = cv2.imread(image_path)\n        if self.transform:\n            image = self.transform(image)\n\n        return image, label\n\n\n# test image transforms(to tensor and normalization)\nimage_size = 330\ncrop_size = 300\ntest_transform = transforms.Compose([\n    transforms.ToPILImage(),\n    transforms.Resize((image_size, image_size)),\n    # center crop\n    transforms.CenterCrop(crop_size),\n    transforms.ToTensor(),\n    transforms.Normalize(mean=(0.485, 0.456, 0.406), std=(0.229, 0.224, 0.225))])\n\ntest_csv = pd.read_csv('../input/aptos2019-blindness-detection/test.csv')\ntest_csv['diagnosis'] = -1\ntest_folder = '../input/aptos2019-blindness-detection/test_images/'\ntest_data = MyDataset(test_csv, test_folder, transform=test_transform)\ntest_loader = torch.utils.data.DataLoader(test_data, batch_size=16, shuffle=False)\n\n# model definition\nlearn = load_learner('../input/regressionmodel2/')\nmodel = learn.model\n\n# for param in model.parameters():\n#     param.requires_grad = False\n\n## simple TTA\n# prediction\npreds_1 = predict(model, testloader=test_loader)\n# preds_2 = np.array(predict(model, testloader=test_loader))\n# preds_3 = np.array(predict(model, testloader=test_loader))\n# preds_4 = np.array(predict(model, testloader=test_loader))\n# preds_5 = np.array(predict(model, testloader=test_loader))\n\n\n# roundoff\npreds_1 = kappa_opt.predict(preds_1)\n# print(preds_1)\n# preds_2 = kappa_opt.predict(preds_2)\n# preds_3 = kappa_opt.predict(preds_3)\n# preds_4 = kappa_opt.predict(preds_4)\n# preds_5 = kappa_opt.predict(preds_5)\n\n#\n\n# simple model ensemble\n# preds_avg = (preds_1 + preds_2 + preds_3 + preds_4 + preds_5)\n\n# submission\nsample_sub = pd.read_csv('../input/aptos2019-blindness-detection/sample_submission.csv')\nsample_sub.diagnosis = preds_1\nsample_sub.diagnosis = sample_sub['diagnosis'].astype(int)\n\n# print(sample_sub.head(10))\n\nsample_sub.to_csv('submission.csv', index=False)"
  }
}