{
  "id": 146597,
  "title": "External Discussion Thread",
  "url": "/competitions/alaska2-image-steganalysis/discussion/146597",
  "author_name": "Addison Howard",
  "post_date": "2020-04-27T19:09:44.681000",
  "votes": 1,
  "comment_count": 21,
  "views": 0,
  "content": "<p>Post links to your external data sources here before the deadline specified in the rules. Once it has been posted, you do not need to post it again.</p>",
  "messages": [
    {
      "id": 927143,
      "postDate": "2020-07-13T08:19:16.130Z",
      "content": "<p><a href=\"https://github.com/iduta/pyconv\">https://github.com/iduta/pyconv</a></p>",
      "rawMarkdown": "https://github.com/iduta/pyconv",
      "votes": 1
    },
    {
      "id": 846621,
      "postDate": "2020-05-13T22:33:54.630Z",
      "content": "<p><a href=\"/addisonhoward\">@addisonhoward</a> Is it permitted to use external source data like iStego, HUGO, Alaska1? Actual data, not just public, trained weights. Rules as they stand indicate yes. Hoping answer is no.</p>",
      "rawMarkdown": "@addisonhoward Is it permitted to use external source data like iStego, HUGO, Alaska1? Actual data, not just public, trained weights. Rules as they stand indicate yes. Hoping answer is no.\n",
      "votes": 2,
      "replies": [
        {
          "id": 849446,
          "postDate": "2020-05-15T18:55:53.797Z",
          "content": "<p>Dear Robga,\nThe use of external source data is not formally forbidden. \nI do not know as much as you guys on Deep Learning. But I claim to know quite a lot on steganalysis and trying to use external dataset is very likely to badly hurt. This for two reasons:\n- First, many prior works (including some from our group) show that the way image have been developed heavily influence the detection of hidden data. We have developed the image in a realistic way, yet we keep the recipe secret such that no one has the advantage to be able to reproduce it\n- Second, we did not explain how we select the number of bit to be embed in each image. Similarly, we do no specify exactly the embedding scheme (those have been modified to embed in color images while designed for grayscale).\nAll in all, one will likely produce images; cover and stego, that differ enough from the training / testing set to hurt more than help</p>\n\n<p>However, I would strongly suggest if one want to go into the direction of data augmentation, to have a look at : <a href=\"https://hal-lirmm.ccsd.cnrs.fr/lirmm-02559838/file/IHMMSec-2016_Yedroudj_Chaumont_Comby_Amara_Bas_Pixels-off.pdf\">this paper on data augmentation</a>, that can easily be implemented using the testing set provided.</p>",
          "rawMarkdown": "Dear Robga,\nThe use of external source data is not formally forbidden. \nI do not know as much as you guys on Deep Learning. But I claim to know quite a lot on steganalysis and trying to use external dataset is very likely to badly hurt. This for two reasons:\n- First, many prior works (including some from our group) show that the way image have been developed heavily influence the detection of hidden data. We have developed the image in a realistic way, yet we keep the recipe secret such that no one has the advantage to be able to reproduce it\n- Second, we did not explain how we select the number of bit to be embed in each image. Similarly, we do no specify exactly the embedding scheme (those have been modified to embed in color images while designed for grayscale).\nAll in all, one will likely produce images; cover and stego, that differ enough from the training / testing set to hurt more than help\n\nHowever, I would strongly suggest if one want to go into the direction of data augmentation, to have a look at : [this paper on data augmentation](https://hal-lirmm.ccsd.cnrs.fr/lirmm-02559838/file/IHMMSec-2016_Yedroudj_Chaumont_Comby_Amara_Bas_Pixels-off.pdf), that can easily be implemented using the testing set provided.\n",
          "votes": 4
        },
        {
          "id": 854817,
          "postDate": "2020-05-20T11:04:34.190Z",
          "rawMarkdown": "",
          "isDeleted": true
        }
      ]
    },
    {
      "id": 838069,
      "postDate": "2020-05-08T09:11:42.360Z",
      "content": "<p><a href=\"https://www.kaggle.com/c/imagenet-object-localization-challenge\">https://www.kaggle.com/c/imagenet-object-localization-challenge</a>\n<a href=\"https://github.com/rwightman/pytorch-image-models\">https://github.com/rwightman/pytorch-image-models</a></p>",
      "rawMarkdown": "https://www.kaggle.com/c/imagenet-object-localization-challenge\nhttps://github.com/rwightman/pytorch-image-models"
    },
    {
      "id": 930250,
      "postDate": "2020-07-15T09:52:57.043Z",
      "content": "<p><a href=\"https://github.com/lukemelas/EfficientNet-PyTorch\">https://github.com/lukemelas/EfficientNet-PyTorch</a>\n<a href=\"https://github.com/yhhhli/RegNet-Pytorch\">https://github.com/yhhhli/RegNet-Pytorch</a>\n<a href=\"https://github.com/facebookresearch/pycls\">https://github.com/facebookresearch/pycls</a>\n<a href=\"https://github.com/clovaai/rexnet\">https://github.com/clovaai/rexnet</a>\n<a href=\"https://github.com/rwightman/pytorch-image-models\">https://github.com/rwightman/pytorch-image-models</a></p>",
      "rawMarkdown": "https://github.com/lukemelas/EfficientNet-PyTorch\nhttps://github.com/yhhhli/RegNet-Pytorch\nhttps://github.com/facebookresearch/pycls\nhttps://github.com/clovaai/rexnet\nhttps://github.com/rwightman/pytorch-image-models"
    },
    {
      "id": 928819,
      "postDate": "2020-07-14T08:36:43.583Z",
      "content": "<p><a href=\"https://github.com/lukemelas/EfficientNet-PyTorch\">https://github.com/lukemelas/EfficientNet-PyTorch</a></p>",
      "rawMarkdown": "https://github.com/lukemelas/EfficientNet-PyTorch"
    },
    {
      "id": 928095,
      "postDate": "2020-07-13T18:49:05.957Z",
      "content": "<p><a href=\"http://places2.csail.mit.edu\">http://places2.csail.mit.edu</a></p>",
      "rawMarkdown": "http://places2.csail.mit.edu"
    },
    {
      "id": 923322,
      "postDate": "2020-07-10T18:10:31.763Z",
      "content": "<p><a href=\"https://storage.googleapis.com/openimages/web/index.html\">https://storage.googleapis.com/openimages/web/index.html</a></p>",
      "rawMarkdown": "https://storage.googleapis.com/openimages/web/index.html"
    },
    {
      "id": 922211,
      "postDate": "2020-07-09T22:52:14.133Z",
      "content": "<p><a href=\"https://github.com/rwightman/pytorch-image-models\">https://github.com/rwightman/pytorch-image-models</a>\n<a href=\"https://github.com/rwightman/gen-efficientnet-pytorch\">https://github.com/rwightman/gen-efficientnet-pytorch</a>\n<a href=\"https://github.com/lukemelas/EfficientNet-PyTorch\">https://github.com/lukemelas/EfficientNet-PyTorch</a>\n<a href=\"https://github.com/clovaai/rexnet\">https://github.com/clovaai/rexnet</a></p>",
      "rawMarkdown": "https://github.com/rwightman/pytorch-image-models\nhttps://github.com/rwightman/gen-efficientnet-pytorch\nhttps://github.com/lukemelas/EfficientNet-PyTorch\nhttps://github.com/clovaai/rexnet"
    },
    {
      "id": 919410,
      "postDate": "2020-07-07T20:47:24.297Z",
      "content": "<p><a href=\"https://github.com/rwightman/pytorch-image-models\">https://github.com/rwightman/pytorch-image-models</a>\n<a href=\"https://github.com/Cadene/pretrained-models.pytorch\">https://github.com/Cadene/pretrained-models.pytorch</a>\n<a href=\"https://github.com/yhhhli/RegNet-Pytorch\">https://github.com/yhhhli/RegNet-Pytorch</a>\n<a href=\"https://github.com/facebookresearch/pycls\">https://github.com/facebookresearch/pycls</a>\n<a href=\"https://github.com/lukemelas/EfficientNet-PyTorch\">https://github.com/lukemelas/EfficientNet-PyTorch</a></p>",
      "rawMarkdown": "https://github.com/rwightman/pytorch-image-models\nhttps://github.com/Cadene/pretrained-models.pytorch\nhttps://github.com/yhhhli/RegNet-Pytorch\nhttps://github.com/facebookresearch/pycls\nhttps://github.com/lukemelas/EfficientNet-PyTorch"
    },
    {
      "id": 916121,
      "postDate": "2020-07-05T11:38:00.600Z",
      "content": "<p><a href=\"https://github.com/tensorflow/models/tree/master/research/slim\">https://github.com/tensorflow/models/tree/master/research/slim</a></p>",
      "rawMarkdown": "https://github.com/tensorflow/models/tree/master/research/slim"
    },
    {
      "id": 914393,
      "postDate": "2020-07-03T20:25:48.817Z",
      "content": "<p><a href=\"https://github.com/tensorflow/tpu/tree/master/models/official/efficientnet\">https://github.com/tensorflow/tpu/tree/master/models/official/efficientnet</a></p>",
      "rawMarkdown": "https://github.com/tensorflow/tpu/tree/master/models/official/efficientnet"
    },
    {
      "id": 910669,
      "postDate": "2020-07-01T09:37:59.867Z",
      "content": "<p><a href=\"http://loki.disi.unitn.it/RAISE/download.html\">http://loki.disi.unitn.it/RAISE/download.html</a></p>",
      "rawMarkdown": "http://loki.disi.unitn.it/RAISE/download.html"
    },
    {
      "id": 904704,
      "postDate": "2020-06-27T20:29:10.180Z",
      "content": "<p><a href=\"http://image-net.org/download\">http://image-net.org/download</a></p>",
      "rawMarkdown": "http://image-net.org/download"
    },
    {
      "id": 904520,
      "postDate": "2020-06-27T17:24:14.247Z",
      "content": "<p><a href=\"https://github.com/YoongiKim/CIFAR-10-images\">https://github.com/YoongiKim/CIFAR-10-images</a></p>",
      "rawMarkdown": "https://github.com/YoongiKim/CIFAR-10-images"
    },
    {
      "id": 901452,
      "postDate": "2020-06-25T13:47:49.850Z",
      "content": "<p><a href=\"https://github.com/qubvel/efficientnet\">https://github.com/qubvel/efficientnet</a></p>",
      "rawMarkdown": "https://github.com/qubvel/efficientnet"
    },
    {
      "id": 900671,
      "postDate": "2020-06-25T02:29:08.540Z",
      "content": "<p><a href=\"https://pytorch.org/docs/stable/torchvision/models.html\">https://pytorch.org/docs/stable/torchvision/models.html</a>\n<a href=\"https://github.com/Cadene/pretrained-models.pytorch\">https://github.com/Cadene/pretrained-models.pytorch</a>\n<a href=\"https://github.com/lukemelas/EfficientNet-PyTorch\">https://github.com/lukemelas/EfficientNet-PyTorch</a></p>",
      "rawMarkdown": "https://pytorch.org/docs/stable/torchvision/models.html\nhttps://github.com/Cadene/pretrained-models.pytorch\nhttps://github.com/lukemelas/EfficientNet-PyTorch"
    },
    {
      "id": 864105,
      "postDate": "2020-05-27T19:47:46.677Z",
      "content": "<p><a href=\"https://github.com/rwightman/gen-efficientnet-pytorch\">https://github.com/rwightman/gen-efficientnet-pytorch</a></p>",
      "rawMarkdown": "[https://github.com/rwightman/gen-efficientnet-pytorch](https://github.com/rwightman/gen-efficientnet-pytorch)"
    },
    {
      "id": 825861,
      "postDate": "2020-04-29T09:42:46Z",
      "content": "<ul>\n<li><a href=\"https://github.com/facebookresearch/FixRes\">https://github.com/facebookresearch/FixRes</a></li>\n<li><a href=\"https://github.com/d-li14/ghostnet.pytorch\">https://github.com/d-li14/ghostnet.pytorch</a></li>\n</ul>\n\n<p>Pretrained weights of GhostNet, Pytorch \n<a href=\"https://www.kaggle.com/ipythonx/ghostnetpretrained\">https://www.kaggle.com/ipythonx/ghostnetpretrained</a></p>",
      "rawMarkdown": "- https://github.com/facebookresearch/FixRes\n- https://github.com/d-li14/ghostnet.pytorch\n\nPretrained weights of GhostNet, Pytorch \nhttps://www.kaggle.com/ipythonx/ghostnetpretrained"
    },
    {
      "id": 823627,
      "postDate": "2020-04-27T19:09:44.683Z",
      "content": "<p>Post links to your external data sources here before the deadline specified in the rules. Once it has been posted, you do not need to post it again.</p>",
      "rawMarkdown": "Post links to your external data sources here before the deadline specified in the rules. Once it has been posted, you do not need to post it again."
    },
    {
      "id": 823730,
      "postDate": "2020-04-27T20:58:47.970Z",
      "rawMarkdown": "",
      "votes": 2,
      "isDeleted": true
    }
  ],
  "comments": [
    {
      "id": 927143,
      "author_name": "bestfitting",
      "author_url": "",
      "post_date": "2020-07-13T08:19:16.130000",
      "content": "<p><a href=\"https://github.com/iduta/pyconv\">https://github.com/iduta/pyconv</a></p>",
      "votes": 1,
      "replies": []
    },
    {
      "id": 846621,
      "author_name": "robga",
      "author_url": "",
      "post_date": "2020-05-13T22:33:54.630000",
      "content": "<p><a href=\"/addisonhoward\">@addisonhoward</a> Is it permitted to use external source data like iStego, HUGO, Alaska1? Actual data, not just public, trained weights. Rules as they stand indicate yes. Hoping answer is no.</p>",
      "votes": 2,
      "replies": [
        {
          "id": 849446,
          "author_name": "Rémi Cogranne",
          "author_url": "",
          "post_date": "2020-05-15T18:55:53.797000",
          "content": "<p>Dear Robga,\nThe use of external source data is not formally forbidden. \nI do not know as much as you guys on Deep Learning. But I claim to know quite a lot on steganalysis and trying to use external dataset is very likely to badly hurt. This for two reasons:\n- First, many prior works (including some from our group) show that the way image have been developed heavily influence the detection of hidden data. We have developed the image in a realistic way, yet we keep the recipe secret such that no one has the advantage to be able to reproduce it\n- Second, we did not explain how we select the number of bit to be embed in each image. Similarly, we do no specify exactly the embedding scheme (those have been modified to embed in color images while designed for grayscale).\nAll in all, one will likely produce images; cover and stego, that differ enough from the training / testing set to hurt more than help</p>\n\n<p>However, I would strongly suggest if one want to go into the direction of data augmentation, to have a look at : <a href=\"https://hal-lirmm.ccsd.cnrs.fr/lirmm-02559838/file/IHMMSec-2016_Yedroudj_Chaumont_Comby_Amara_Bas_Pixels-off.pdf\">this paper on data augmentation</a>, that can easily be implemented using the testing set provided.</p>",
          "votes": 4,
          "replies": []
        },
        {
          "id": 854817,
          "author_name": "",
          "author_url": "",
          "post_date": "2020-05-20T11:04:34.190000",
          "content": "",
          "votes": 0,
          "replies": []
        }
      ]
    },
    {
      "id": 838069,
      "author_name": "Eugene Khvedchenya",
      "author_url": "",
      "post_date": "2020-05-08T09:11:42.360000",
      "content": "<p><a href=\"https://www.kaggle.com/c/imagenet-object-localization-challenge\">https://www.kaggle.com/c/imagenet-object-localization-challenge</a>\n<a href=\"https://github.com/rwightman/pytorch-image-models\">https://github.com/rwightman/pytorch-image-models</a></p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 930250,
      "author_name": "Heroseo",
      "author_url": "",
      "post_date": "2020-07-15T09:52:57.043000",
      "content": "<p><a href=\"https://github.com/lukemelas/EfficientNet-PyTorch\">https://github.com/lukemelas/EfficientNet-PyTorch</a>\n<a href=\"https://github.com/yhhhli/RegNet-Pytorch\">https://github.com/yhhhli/RegNet-Pytorch</a>\n<a href=\"https://github.com/facebookresearch/pycls\">https://github.com/facebookresearch/pycls</a>\n<a href=\"https://github.com/clovaai/rexnet\">https://github.com/clovaai/rexnet</a>\n<a href=\"https://github.com/rwightman/pytorch-image-models\">https://github.com/rwightman/pytorch-image-models</a></p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 928819,
      "author_name": "BokingChen",
      "author_url": "",
      "post_date": "2020-07-14T08:36:43.583000",
      "content": "<p><a href=\"https://github.com/lukemelas/EfficientNet-PyTorch\">https://github.com/lukemelas/EfficientNet-PyTorch</a></p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 928095,
      "author_name": "Arthur Stsepanenka",
      "author_url": "",
      "post_date": "2020-07-13T18:49:05.957000",
      "content": "<p><a href=\"http://places2.csail.mit.edu\">http://places2.csail.mit.edu</a></p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 923322,
      "author_name": "Anton Chikin",
      "author_url": "",
      "post_date": "2020-07-10T18:10:31.763000",
      "content": "<p><a href=\"https://storage.googleapis.com/openimages/web/index.html\">https://storage.googleapis.com/openimages/web/index.html</a></p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 922211,
      "author_name": "Yifan Xie",
      "author_url": "",
      "post_date": "2020-07-09T22:52:14.133000",
      "content": "<p><a href=\"https://github.com/rwightman/pytorch-image-models\">https://github.com/rwightman/pytorch-image-models</a>\n<a href=\"https://github.com/rwightman/gen-efficientnet-pytorch\">https://github.com/rwightman/gen-efficientnet-pytorch</a>\n<a href=\"https://github.com/lukemelas/EfficientNet-PyTorch\">https://github.com/lukemelas/EfficientNet-PyTorch</a>\n<a href=\"https://github.com/clovaai/rexnet\">https://github.com/clovaai/rexnet</a></p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 919410,
      "author_name": "Psi",
      "author_url": "",
      "post_date": "2020-07-07T20:47:24.297000",
      "content": "<p><a href=\"https://github.com/rwightman/pytorch-image-models\">https://github.com/rwightman/pytorch-image-models</a>\n<a href=\"https://github.com/Cadene/pretrained-models.pytorch\">https://github.com/Cadene/pretrained-models.pytorch</a>\n<a href=\"https://github.com/yhhhli/RegNet-Pytorch\">https://github.com/yhhhli/RegNet-Pytorch</a>\n<a href=\"https://github.com/facebookresearch/pycls\">https://github.com/facebookresearch/pycls</a>\n<a href=\"https://github.com/lukemelas/EfficientNet-PyTorch\">https://github.com/lukemelas/EfficientNet-PyTorch</a></p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 916121,
      "author_name": "Pzero",
      "author_url": "",
      "post_date": "2020-07-05T11:38:00.600000",
      "content": "<p><a href=\"https://github.com/tensorflow/models/tree/master/research/slim\">https://github.com/tensorflow/models/tree/master/research/slim</a></p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 914393,
      "author_name": "عثمان",
      "author_url": "",
      "post_date": "2020-07-03T20:25:48.817000",
      "content": "<p><a href=\"https://github.com/tensorflow/tpu/tree/master/models/official/efficientnet\">https://github.com/tensorflow/tpu/tree/master/models/official/efficientnet</a></p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 910669,
      "author_name": "Andrés Miguel Torrubia Sáez",
      "author_url": "",
      "post_date": "2020-07-01T09:37:59.867000",
      "content": "<p><a href=\"http://loki.disi.unitn.it/RAISE/download.html\">http://loki.disi.unitn.it/RAISE/download.html</a></p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 904704,
      "author_name": "aL_eX",
      "author_url": "",
      "post_date": "2020-06-27T20:29:10.180000",
      "content": "<p><a href=\"http://image-net.org/download\">http://image-net.org/download</a></p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 904520,
      "author_name": "aL_eX",
      "author_url": "",
      "post_date": "2020-06-27T17:24:14.247000",
      "content": "<p><a href=\"https://github.com/YoongiKim/CIFAR-10-images\">https://github.com/YoongiKim/CIFAR-10-images</a></p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 901452,
      "author_name": "Johnny Lee",
      "author_url": "",
      "post_date": "2020-06-25T13:47:49.850000",
      "content": "<p><a href=\"https://github.com/qubvel/efficientnet\">https://github.com/qubvel/efficientnet</a></p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 900671,
      "author_name": "Guanshuo Xu",
      "author_url": "",
      "post_date": "2020-06-25T02:29:08.540000",
      "content": "<p><a href=\"https://pytorch.org/docs/stable/torchvision/models.html\">https://pytorch.org/docs/stable/torchvision/models.html</a>\n<a href=\"https://github.com/Cadene/pretrained-models.pytorch\">https://github.com/Cadene/pretrained-models.pytorch</a>\n<a href=\"https://github.com/lukemelas/EfficientNet-PyTorch\">https://github.com/lukemelas/EfficientNet-PyTorch</a></p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 864105,
      "author_name": "yuval reina",
      "author_url": "",
      "post_date": "2020-05-27T19:47:46.677000",
      "content": "<p><a href=\"https://github.com/rwightman/gen-efficientnet-pytorch\">https://github.com/rwightman/gen-efficientnet-pytorch</a></p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 825861,
      "author_name": "Innat",
      "author_url": "",
      "post_date": "2020-04-29T09:42:46",
      "content": "<ul>\n<li><a href=\"https://github.com/facebookresearch/FixRes\">https://github.com/facebookresearch/FixRes</a></li>\n<li><a href=\"https://github.com/d-li14/ghostnet.pytorch\">https://github.com/d-li14/ghostnet.pytorch</a></li>\n</ul>\n\n<p>Pretrained weights of GhostNet, Pytorch \n<a href=\"https://www.kaggle.com/ipythonx/ghostnetpretrained\">https://www.kaggle.com/ipythonx/ghostnetpretrained</a></p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 823730,
      "author_name": "",
      "author_url": "",
      "post_date": "2020-04-27T20:58:47.970000",
      "content": "",
      "votes": 2,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "927143": "https://github.com/iduta/pyconv",
    "846621": "@addisonhoward Is it permitted to use external source data like iStego, HUGO, Alaska1? Actual data, not just public, trained weights. Rules as they stand indicate yes. Hoping answer is no.\n",
    "838069": "https://www.kaggle.com/c/imagenet-object-localization-challenge\nhttps://github.com/rwightman/pytorch-image-models",
    "930250": "https://github.com/lukemelas/EfficientNet-PyTorch\nhttps://github.com/yhhhli/RegNet-Pytorch\nhttps://github.com/facebookresearch/pycls\nhttps://github.com/clovaai/rexnet\nhttps://github.com/rwightman/pytorch-image-models",
    "928819": "https://github.com/lukemelas/EfficientNet-PyTorch",
    "928095": "http://places2.csail.mit.edu",
    "923322": "https://storage.googleapis.com/openimages/web/index.html",
    "922211": "https://github.com/rwightman/pytorch-image-models\nhttps://github.com/rwightman/gen-efficientnet-pytorch\nhttps://github.com/lukemelas/EfficientNet-PyTorch\nhttps://github.com/clovaai/rexnet",
    "919410": "https://github.com/rwightman/pytorch-image-models\nhttps://github.com/Cadene/pretrained-models.pytorch\nhttps://github.com/yhhhli/RegNet-Pytorch\nhttps://github.com/facebookresearch/pycls\nhttps://github.com/lukemelas/EfficientNet-PyTorch",
    "916121": "https://github.com/tensorflow/models/tree/master/research/slim",
    "914393": "https://github.com/tensorflow/tpu/tree/master/models/official/efficientnet",
    "910669": "http://loki.disi.unitn.it/RAISE/download.html",
    "904704": "http://image-net.org/download",
    "904520": "https://github.com/YoongiKim/CIFAR-10-images",
    "901452": "https://github.com/qubvel/efficientnet",
    "900671": "https://pytorch.org/docs/stable/torchvision/models.html\nhttps://github.com/Cadene/pretrained-models.pytorch\nhttps://github.com/lukemelas/EfficientNet-PyTorch",
    "864105": "[https://github.com/rwightman/gen-efficientnet-pytorch](https://github.com/rwightman/gen-efficientnet-pytorch)",
    "825861": "- https://github.com/facebookresearch/FixRes\n- https://github.com/d-li14/ghostnet.pytorch\n\nPretrained weights of GhostNet, Pytorch \nhttps://www.kaggle.com/ipythonx/ghostnetpretrained",
    "823627": "Post links to your external data sources here before the deadline specified in the rules. Once it has been posted, you do not need to post it again.",
    "823730": ""
  }
}