![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F252874%2Ffc1b6ca8223a469782bc0243cf224779%2FCapture.JPG?generation=1594090552724623&alt=media\")
Per the competition rules, post links to your external data sources here before the deadline specified. Once it has been posted, you do not need to post it again.
", "messages": [ { "id": 912781, "postDate": "2020-07-02T18:07:06.180Z", "content": "Considering the experience from the DFDC competition. I had a suggestion for a new feature or a widget/form (for a lack of a better word)\nIt should be an interface which will allow participants to enter a link & name of the external dataset. Once given they can be analyzed for suitability for the competition. This will create a green-list of datasets & highlight ones which cannot be used. So whenever a new participant submits the name of the dataset , it automatically gets checked with the green-list & there are no gray areas in the competition.
\n\nThis will also remove the current in-efficient means of submitting the information through a message on a forum thread. I have seen the organizers trying to respond to the same queries again & again. Also for the participant it gives an easy interface to check for green-listed datasets.
", "rawMarkdown": "Considering the experience from the DFDC competition. I had a suggestion for a new feature or a widget/form (for a lack of a better word)\nIt should be an interface which will allow participants to enter a link & name of the external dataset. Once given they can be analyzed for suitability for the competition. This will create a green-list of datasets & highlight ones which cannot be used. So whenever a new participant submits the name of the dataset , it automatically gets checked with the green-list & there are no gray areas in the competition.\n\nThis will also remove the current *in-efficient* means of submitting the information through a message on a forum thread. I have seen the organizers trying to respond to the same queries again & again. Also for the participant it gives an easy interface to check for green-listed datasets.", "votes": 4, "replies": [ { "id": 917799, "postDate": "2020-07-06T18:49:55.250Z", "content": "Thank you for this suggestion, we appreciate the creativity of the Kaggle community.
", "rawMarkdown": "Thank you for this suggestion, we appreciate the creativity of the Kaggle community. " }, { "id": 918148, "postDate": "2020-07-07T02:59:31.603Z", "content": "Thanks for the kind words @maggiemd !\nWould love to see this suggestion being implemented. The current forum based approach is highly inefficient and leads to a lot of confusion. It would be good if kaggle can come up with a structured approach for gathering this information (external data threads)
\n\nA simple external data submission widget, could solve a lot of heartaches -
\n\n`\n
`\nOnce submitted the dataset can be evaluated in terms of suitability & bracketed into a green, red & gray buckets. \n- Green: It's good to go!\n- Red: Cannot be used by participants\n- Gray: Organizers have reservations OR still evaluating
", "rawMarkdown": "Thanks for the kind words @maggiemd !\nWould love to see this suggestion being implemented. The current forum based approach is highly inefficient and leads to a lot of confusion. It would be good if kaggle can come up with a structured approach for gathering this information (**external data threads**)\n\nA simple external data submission widget, could solve a lot of heartaches - \n\n`\n\n\n`\nOnce submitted the dataset can be evaluated in terms of suitability & bracketed into a green, red & gray buckets. \n- Green: It's good to go!\n- Red: Cannot be used by participants\n- Gray: Organizers have reservations OR still evaluating \n\n\n" } ] }, { "id": 964945, "postDate": "2020-08-10T09:31:51.133Z", "content": "TF imagenet pretrained weights.\nhttps://github.com/tensorflow/tensorflow/tree/v2.2.0/tensorflow/python/keras/applications\nhttps://github.com/tensorflow/models/tree/master/official/vision/image_classification\nhttps://github.com/tensorflow/tpu
DELG pretrained models.\nHost baseline from Google Landmark Retrieval 2020: https://www.kaggle.com/camaskew/baseline-landmark-retrieval-model\nHost baseline from Google Landmark Recognition 2020: https://www.kaggle.com/camaskew/delg-saved-models\nDELG official models: https://github.com/tensorflow/models/tree/master/research/delf
Well, this may be obvious but since nobody said it:\nThe baseline model provided by the organizers: https://www.kaggle.com/camaskew/baseline-landmark-retrieval-model
", "rawMarkdown": "Well, this may be obvious but since nobody said it:\nThe baseline model provided by the organizers: https://www.kaggle.com/camaskew/baseline-landmark-retrieval-model", "votes": 1, "replies": [ { "id": 961767, "postDate": "2020-08-07T13:35:58.603Z", "content": "Did you finetune the baseline model ? I still can't do that 😣
", "rawMarkdown": "Did you finetune the baseline model ? I still can't do that 😣" } ] }, { "id": 909902, "postDate": "2020-06-30T23:00:29.870Z", "content": "Per the competition rules, post links to your external data sources here before the deadline specified. Once it has been posted, you do not need to post it again.
", "rawMarkdown": "Per the competition rules, post links to your external data sources here before the deadline specified. Once it has been posted, you do not need to post it again." }, { "id": 973589, "postDate": "2020-08-17T11:40:16.650Z", "content": "The code repository we use:
\nhttps://github.com/PaddlePaddle/PaddleClas
\nhttps://github.com/ym547559398/pymetric
\nhttps://github.com/PaddlePaddle/models
\nhttps://github.com/zhanghang1989/ResNeSt
use data:
\nGLD v1 and GLD v2
\nhttps://www.kaggle.com/c/landmark-retrieval-2019
\nhttps://www.kaggle.com/c/landmark-retrieval-challenge
I believe the last day to comment external data here was the Entry deadline, which was one week ago
", "rawMarkdown": "I believe the last day to comment external data here was the Entry deadline, which was one week ago", "votes": 1 }, { "id": 974156, "postDate": "2020-08-17T18:53:33.973Z", "rawMarkdown": "", "isDeleted": true }, { "id": 974172, "postDate": "2020-08-17T19:08:41.980Z", "content": "The repositories do not contain additional data that was not already mentioned and the data from existing competitions is available to all participants.
", "rawMarkdown": "The repositories do not contain additional data that was not already mentioned and the data from existing competitions is available to all participants. " } ] }, { "id": 932728, "postDate": "2020-07-17T08:40:25.413Z", "content": "why my submission is not submitting it takes too much time
", "rawMarkdown": "why my submission is not submitting it takes too much time", "votes": -1 }, { "id": 969312, "postDate": "2020-08-13T15:55:23.960Z", "content": "imagenet pretrained weights:
\nhttps://github.com/XingangPan/IBN-Net
tensorflow models: https://github.com/tensorflow/models, https://tfhub.dev/, https://www.tensorflow.org/resources/models-datasets
", "rawMarkdown": "tensorflow models: https://github.com/tensorflow/models, https://tfhub.dev/, https://www.tensorflow.org/resources/models-datasets" }, { "id": 965420, "postDate": "2020-08-10T15:59:34.137Z", "content": "https://storage.googleapis.com/openimages/web/index.html\nhttps://github.com/lukemelas/EfficientNet-PyTorch\nhttps://github.com/filipradenovic/cnnimageretrieval-pytorch\nhttps://github.com/almazan/deep-image-retrieval\nhttps://github.com/PaddlePaddle/models\nhttps://github.com/PaddlePaddle/Research\nhttps://pytorch.org/docs/stable/torchvision/index.html
", "rawMarkdown": "https://storage.googleapis.com/openimages/web/index.html\nhttps://github.com/lukemelas/EfficientNet-PyTorch\nhttps://github.com/filipradenovic/cnnimageretrieval-pytorch\nhttps://github.com/almazan/deep-image-retrieval\nhttps://github.com/PaddlePaddle/models\nhttps://github.com/PaddlePaddle/Research\nhttps://pytorch.org/docs/stable/torchvision/index.html" }, { "id": 965279, "postDate": "2020-08-10T14:14:00.937Z", "content": "https://pypi.org/project/pytorchcv/\nhttps://pytorch.org/docs/stable/torchvision/models.html
", "rawMarkdown": "https://pypi.org/project/pytorchcv/\nhttps://pytorch.org/docs/stable/torchvision/models.html" }, { "id": 965157, "postDate": "2020-08-10T12:24:44.623Z", "content": "tf keras imagenet pretrained efficientnet https://github.com/qubvel/efficientnet
", "rawMarkdown": "tf keras imagenet pretrained efficientnet https://github.com/qubvel/efficientnet" }, { "id": 964561, "postDate": "2020-08-10T02:22:37.520Z", "content": "https://github.com/QiaoranC/tf_ResNeSt_RegNet_model\nhttps://github.com/zhanghang1989/ResNeSt\nhttps://github.com/facebookresearch/pycls\nhttps://github.com/Cadene/pretrained-models.pytorch
", "rawMarkdown": "https://github.com/QiaoranC/tf_ResNeSt_RegNet_model\nhttps://github.com/zhanghang1989/ResNeSt\nhttps://github.com/facebookresearch/pycls\nhttps://github.com/Cadene/pretrained-models.pytorch", "replies": [ { "id": 965516, "postDate": "2020-08-10T17:29:48.073Z", "content": "https://arxiv.org/abs/1812.01584\nhttp://storage.googleapis.com/delf/d2r_frcnn_20190411.tar.gz
", "rawMarkdown": "https://arxiv.org/abs/1812.01584\nhttp://storage.googleapis.com/delf/d2r_frcnn_20190411.tar.gz" } ] }, { "id": 964252, "postDate": "2020-08-09T17:31:41.047Z", "content": "https://github.com/tensorflow/models/tree/master/research/
\nhttps://www.kaggle.com/google/google-landmarks-dataset
\nhttps://github.com/osmr/imgclsmob/
\nhttps://github.com/rwightman/pytorch-image-models
Not sure if this counts as external data but I converted about half the dataset into tfrecords. Each example contains triplets for a model using triplet loss.
\nhttps://www.kaggle.com/mattbast/google-landmarks-2020-tfrecords
\nhttps://www.kaggle.com/mattbast/google-landmarks-2020-triplet-loss-tfrecords-pt2
\nhttps://www.kaggle.com/mattbast/google-landmarks-2020-triplet-loss-tfrecords-pt3
\nhttps://www.kaggle.com/mattbast/google-landmarks-2020-triplet-loss-tfrecords-pt4
\nhttps://www.kaggle.com/mattbast/google-landmarks-2020-triplet-loss-tfrecords-pt5
\nhttps://www.kaggle.com/mattbast/google-landmarks-2020-triplet-loss-tfrecords-pt6
I would like to try keras pretrained models.\nKeras Applications\nhttps://keras.io/api/applications/\nClassification models Zoo - Keras (and TensorFlow Keras)\nhttps://github.com/qubvel/classification_models
", "rawMarkdown": "I would like to try keras pretrained models.\nKeras Applications\nhttps://keras.io/api/applications/\nClassification models Zoo - Keras (and TensorFlow Keras)\nhttps://github.com/qubvel/classification_models" }, { "id": 963055, "postDate": "2020-08-08T16:23:38.477Z", "content": "Already mentioned, but with modified links.\nFull GLD v2 Dataset, https://github.com/cvdfoundation/google-landmark\nGLD v1 Dataset, https://www.kaggle.com/google/google-landmarks-dataset
", "rawMarkdown": "Already mentioned, but with modified links.\nFull GLD v2 Dataset, https://github.com/cvdfoundation/google-landmark\nGLD v1 Dataset, https://www.kaggle.com/google/google-landmarks-dataset" }, { "id": 952779, "postDate": "2020-07-31T08:06:27.143Z", "content": "I would like to try GLD-v1 and full GLD-v2.\nhttps://www.kaggle.com/c/landmark-retrieval-challenge\nhttps://www.kaggle.com/c/landmark-recognition-challenge\nhttps://www.kaggle.com/c/landmark-retrieval-2019\nhttps://www.kaggle.com/c/landmark-recognition-2019
", "rawMarkdown": "I would like to try GLD-v1 and full GLD-v2.\nhttps://www.kaggle.com/c/landmark-retrieval-challenge\nhttps://www.kaggle.com/c/landmark-recognition-challenge\nhttps://www.kaggle.com/c/landmark-retrieval-2019\nhttps://www.kaggle.com/c/landmark-recognition-2019" }, { "id": 948586, "postDate": "2020-07-28T04:41:09.260Z", "content": "Datasets from the CVPR 2020 (Long-Term Visual Localization)
\n\nhttps://www.visuallocalization.net/datasets/
\n\nIs GLD-v1 allowed?gld-v1 kaggle dataset
\n\nIn dataset link, license term says \"The images listed in this dataset are publicly available on the web, and may have different licenses. Google does not own their copyright.\"
", "rawMarkdown": "Is GLD-v1 allowed?[gld-v1 kaggle dataset](https://www.kaggle.com/google/google-landmarks-dataset)\n\nIn dataset link, license term says \"The images listed in this dataset are publicly available on the web, and may have different licenses. Google does not own their copyright.\"" }, { "id": 939516, "postDate": "2020-07-22T09:32:45.107Z", "content": "External dataset:
\nWEBVISION DATASET 2.0: https://data.vision.ee.ethz.ch/cvl/webvision//dataset2018.html
\nSimCLRV2: https://github.com/google-research/simclr
Considering the experience from the DFDC competition. I had a suggestion for a new feature or a widget/form (for a lack of a better word)\nIt should be an interface which will allow participants to enter a link & name of the external dataset. Once given they can be analyzed for suitability for the competition. This will create a green-list of datasets & highlight ones which cannot be used. So whenever a new participant submits the name of the dataset , it automatically gets checked with the green-list & there are no gray areas in the competition.
\n\nThis will also remove the current in-efficient means of submitting the information through a message on a forum thread. I have seen the organizers trying to respond to the same queries again & again. Also for the participant it gives an easy interface to check for green-listed datasets.
", "votes": 4, "replies": [ { "id": 917799, "author_name": "Maggie", "author_url": "", "post_date": "2020-07-06T18:49:55.250000", "content": "Thank you for this suggestion, we appreciate the creativity of the Kaggle community.
", "votes": 0, "replies": [] }, { "id": 918148, "author_name": "SkyLord", "author_url": "", "post_date": "2020-07-07T02:59:31.603000", "content": "Thanks for the kind words @maggiemd !\nWould love to see this suggestion being implemented. The current forum based approach is highly inefficient and leads to a lot of confusion. It would be good if kaggle can come up with a structured approach for gathering this information (external data threads)
\n\nA simple external data submission widget, could solve a lot of heartaches -
\n\n`\n
`\nOnce submitted the dataset can be evaluated in terms of suitability & bracketed into a green, red & gray buckets. \n- Green: It's good to go!\n- Red: Cannot be used by participants\n- Gray: Organizers have reservations OR still evaluating
", "votes": 0, "replies": [] } ] }, { "id": 964945, "author_name": "Fumihiro Kaneko", "author_url": "", "post_date": "2020-08-10T09:31:51.133000", "content": "TF imagenet pretrained weights.\nhttps://github.com/tensorflow/tensorflow/tree/v2.2.0/tensorflow/python/keras/applications\nhttps://github.com/tensorflow/models/tree/master/official/vision/image_classification\nhttps://github.com/tensorflow/tpu
DELG pretrained models.\nHost baseline from Google Landmark Retrieval 2020: https://www.kaggle.com/camaskew/baseline-landmark-retrieval-model\nHost baseline from Google Landmark Recognition 2020: https://www.kaggle.com/camaskew/delg-saved-models\nDELG official models: https://github.com/tensorflow/models/tree/master/research/delf
Well, this may be obvious but since nobody said it:\nThe baseline model provided by the organizers: https://www.kaggle.com/camaskew/baseline-landmark-retrieval-model
", "votes": 1, "replies": [ { "id": 961767, "author_name": "toshi_k", "author_url": "", "post_date": "2020-08-07T13:35:58.603000", "content": "Did you finetune the baseline model ? I still can't do that 😣
", "votes": 0, "replies": [] } ] }, { "id": 973589, "author_name": "smileimg", "author_url": "", "post_date": "2020-08-17T11:40:16.650000", "content": "The code repository we use:
\nhttps://github.com/PaddlePaddle/PaddleClas
\nhttps://github.com/ym547559398/pymetric
\nhttps://github.com/PaddlePaddle/models
\nhttps://github.com/zhanghang1989/ResNeSt
use data:
\nGLD v1 and GLD v2
\nhttps://www.kaggle.com/c/landmark-retrieval-2019
\nhttps://www.kaggle.com/c/landmark-retrieval-challenge
I believe the last day to comment external data here was the Entry deadline, which was one week ago
", "votes": 1, "replies": [] }, { "id": 974156, "author_name": "", "author_url": "", "post_date": "2020-08-17T18:53:33.973000", "content": "", "votes": 0, "replies": [] }, { "id": 974172, "author_name": "Maggie", "author_url": "", "post_date": "2020-08-17T19:08:41.980000", "content": "The repositories do not contain additional data that was not already mentioned and the data from existing competitions is available to all participants.
", "votes": 0, "replies": [] } ] }, { "id": 932728, "author_name": "shubhamshukla", "author_url": "", "post_date": "2020-07-17T08:40:25.413000", "content": "why my submission is not submitting it takes too much time
", "votes": -1, "replies": [] }, { "id": 969312, "author_name": "Octo", "author_url": "", "post_date": "2020-08-13T15:55:23.960000", "content": "imagenet pretrained weights:
\nhttps://github.com/XingangPan/IBN-Net
tensorflow models: https://github.com/tensorflow/models, https://tfhub.dev/, https://www.tensorflow.org/resources/models-datasets
", "votes": 0, "replies": [] }, { "id": 965420, "author_name": "Dmitrii Tsybulevskii", "author_url": "", "post_date": "2020-08-10T15:59:34.137000", "content": "https://storage.googleapis.com/openimages/web/index.html\nhttps://github.com/lukemelas/EfficientNet-PyTorch\nhttps://github.com/filipradenovic/cnnimageretrieval-pytorch\nhttps://github.com/almazan/deep-image-retrieval\nhttps://github.com/PaddlePaddle/models\nhttps://github.com/PaddlePaddle/Research\nhttps://pytorch.org/docs/stable/torchvision/index.html
", "votes": 0, "replies": [] }, { "id": 965279, "author_name": "Ern", "author_url": "", "post_date": "2020-08-10T14:14:00.937000", "content": "https://pypi.org/project/pytorchcv/\nhttps://pytorch.org/docs/stable/torchvision/models.html
", "votes": 0, "replies": [] }, { "id": 965157, "author_name": "keetar", "author_url": "", "post_date": "2020-08-10T12:24:44.623000", "content": "tf keras imagenet pretrained efficientnet https://github.com/qubvel/efficientnet
", "votes": 0, "replies": [] }, { "id": 964561, "author_name": "kenji", "author_url": "", "post_date": "2020-08-10T02:22:37.520000", "content": "https://github.com/QiaoranC/tf_ResNeSt_RegNet_model\nhttps://github.com/zhanghang1989/ResNeSt\nhttps://github.com/facebookresearch/pycls\nhttps://github.com/Cadene/pretrained-models.pytorch
", "votes": 0, "replies": [ { "id": 965516, "author_name": "kenji", "author_url": "", "post_date": "2020-08-10T17:29:48.073000", "content": "https://arxiv.org/abs/1812.01584\nhttp://storage.googleapis.com/delf/d2r_frcnn_20190411.tar.gz
", "votes": 0, "replies": [] } ] }, { "id": 964252, "author_name": "bestfitting", "author_url": "", "post_date": "2020-08-09T17:31:41.047000", "content": "https://github.com/tensorflow/models/tree/master/research/
\nhttps://www.kaggle.com/google/google-landmarks-dataset
\nhttps://github.com/osmr/imgclsmob/
\nhttps://github.com/rwightman/pytorch-image-models
Not sure if this counts as external data but I converted about half the dataset into tfrecords. Each example contains triplets for a model using triplet loss.
\nhttps://www.kaggle.com/mattbast/google-landmarks-2020-tfrecords
\nhttps://www.kaggle.com/mattbast/google-landmarks-2020-triplet-loss-tfrecords-pt2
\nhttps://www.kaggle.com/mattbast/google-landmarks-2020-triplet-loss-tfrecords-pt3
\nhttps://www.kaggle.com/mattbast/google-landmarks-2020-triplet-loss-tfrecords-pt4
\nhttps://www.kaggle.com/mattbast/google-landmarks-2020-triplet-loss-tfrecords-pt5
\nhttps://www.kaggle.com/mattbast/google-landmarks-2020-triplet-loss-tfrecords-pt6
I would like to try keras pretrained models.\nKeras Applications\nhttps://keras.io/api/applications/\nClassification models Zoo - Keras (and TensorFlow Keras)\nhttps://github.com/qubvel/classification_models
", "votes": 0, "replies": [] }, { "id": 963055, "author_name": "keetar", "author_url": "", "post_date": "2020-08-08T16:23:38.477000", "content": "Already mentioned, but with modified links.\nFull GLD v2 Dataset, https://github.com/cvdfoundation/google-landmark\nGLD v1 Dataset, https://www.kaggle.com/google/google-landmarks-dataset
", "votes": 0, "replies": [] }, { "id": 952779, "author_name": "toshi_k", "author_url": "", "post_date": "2020-07-31T08:06:27.143000", "content": "I would like to try GLD-v1 and full GLD-v2.\nhttps://www.kaggle.com/c/landmark-retrieval-challenge\nhttps://www.kaggle.com/c/landmark-recognition-challenge\nhttps://www.kaggle.com/c/landmark-retrieval-2019\nhttps://www.kaggle.com/c/landmark-recognition-2019
", "votes": 0, "replies": [] }, { "id": 948586, "author_name": "SkyLord", "author_url": "", "post_date": "2020-07-28T04:41:09.260000", "content": "Datasets from the CVPR 2020 (Long-Term Visual Localization)
\n\nhttps://www.visuallocalization.net/datasets/
\n\nIs GLD-v1 allowed?gld-v1 kaggle dataset
\n\nIn dataset link, license term says \"The images listed in this dataset are publicly available on the web, and may have different licenses. Google does not own their copyright.\"
", "votes": 0, "replies": [] }, { "id": 939516, "author_name": "fISHpAM", "author_url": "", "post_date": "2020-07-22T09:32:45.107000", "content": "External dataset:
\nWEBVISION DATASET 2.0: https://data.vision.ee.ethz.ch/cvl/webvision//dataset2018.html
\nSimCLRV2: https://github.com/google-research/simclr