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": 815578, "postDate": "2020-04-21T16:46:27.580Z", "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.", "votes": 11 }, { "id": 824859, "postDate": "2020-04-28T16:22:14.287Z", "content": "If you are looking for additional freely available data of prostate tissue: take a look at the PESO dataset. The dataset consists of whole slide images of prostatectomies with precise annotations of the epithelium.
", "rawMarkdown": "If you are looking for additional freely available data of prostate tissue: take a look at the [PESO dataset](https://zenodo.org/record/1485967). The dataset consists of whole slide images of prostatectomies with precise annotations of the epithelium. ", "votes": 5, "replies": [ { "id": 829213, "postDate": "2020-05-01T15:44:30.263Z", "rawMarkdown": "" }, { "id": 906274, "postDate": "2020-06-29T07:14:23.180Z", "content": "@wouterbulten i suppose it dsnt has grades defined. i has only given the masks
", "rawMarkdown": "@wouterbulten i suppose it dsnt has grades defined. i has only given the masks " }, { "id": 906288, "postDate": "2020-06-29T07:27:37.213Z", "content": "Indeed, no Gleason grades. It can be used for pre-training or if you want to go for a segmentation method.
", "rawMarkdown": "Indeed, no Gleason grades. It can be used for pre-training or if you want to go for a segmentation method. ", "votes": 1 } ] }, { "id": 816567, "postDate": "2020-04-22T12:35:50.910Z", "content": "Maybe I've just not read all the terms and conditions, but am I allowed to train a model on proprietary hardware and upload the weights for inference? Or do I need to disclose the model+weights? Or only the weights?
", "rawMarkdown": "Maybe I've just not read all the terms and conditions, but am I allowed to train a model on proprietary hardware and upload the weights for inference? Or do I need to disclose the model+weights? Or only the weights?", "votes": 3, "replies": [ { "id": 820141, "postDate": "2020-04-25T07:16:20.897Z", "rawMarkdown": "", "isDeleted": true }, { "id": 878837, "postDate": "2020-06-08T22:48:26.860Z", "rawMarkdown": "", "votes": 1, "isDeleted": true }, { "id": 879151, "postDate": "2020-06-09T08:33:47.837Z", "content": "Yes you are allowed to train models locally on your own hardware, and upload these models to be used in an inference kernel. Note that rules regarding external data still apply. Also, winning solutions need to make the training code open source, see https://www.kaggle.com/c/prostate-cancer-grade-assessment/rules
", "rawMarkdown": "Yes you are allowed to train models locally on your own hardware, and upload these models to be used in an inference kernel. Note that rules regarding external data still apply. Also, winning solutions need to make the training code open source, see https://www.kaggle.com/c/prostate-cancer-grade-assessment/rules", "votes": 1 }, { "id": 879245, "postDate": "2020-06-09T10:36:01.383Z", "rawMarkdown": "", "isDeleted": true }, { "id": 879314, "postDate": "2020-06-09T11:45:39.623Z", "content": "If those weights are from models trained on the competition data, then you don't have to make them public. You have to make it public when:
\n\ncan we also make use of Gleason2019 challenge data.. it has got over now .
", "rawMarkdown": "can we also make use of Gleason2019 challenge data.. it has got over now . " } ] }, { "id": 929999, "postDate": "2020-07-15T06:09:34.570Z", "content": "https://github.com/joe-siyuan-qiao/WeightStandardization\nhttps://github.com/facebookresearch/semi-supervised-ImageNet1K-models\nhttps://github.com/lukemelas/EfficientNet-PyTorch\nhttps://github.com/qubvel/segmentation_models.pytorch\nhttps://github.com/facebookresearch/pycls/\nhttps://zenodo.org/record/1485967\nhttps://github.com/rwightman/pytorch-image-models/\nhttps://github.com/google-research/big_transfer\nhttps://github.com/qubvel/efficientnet\nhttps://github.com/Cadene/pretrained-models.pytorch
", "rawMarkdown": "https://github.com/joe-siyuan-qiao/WeightStandardization\nhttps://github.com/facebookresearch/semi-supervised-ImageNet1K-models\nhttps://github.com/lukemelas/EfficientNet-PyTorch\nhttps://github.com/qubvel/segmentation_models.pytorch\nhttps://github.com/facebookresearch/pycls/\nhttps://zenodo.org/record/1485967\nhttps://github.com/rwightman/pytorch-image-models/\nhttps://github.com/google-research/big_transfer\nhttps://github.com/qubvel/efficientnet\nhttps://github.com/Cadene/pretrained-models.pytorch", "votes": 1 }, { "id": 917615, "postDate": "2020-07-06T16:33:07.740Z", "content": "https://github.com/yhhhli/RegNet-Pytorch
", "rawMarkdown": "https://github.com/yhhhli/RegNet-Pytorch", "votes": 1 }, { "id": 899623, "postDate": "2020-06-24T10:38:31.840Z", "content": "https://github.com/osmr/imgclsmob/tree/master/pytorch\nhttps://github.com/facebookresearch/pycls/\nhttps://github.com/iduta/pyconv
", "rawMarkdown": "https://github.com/osmr/imgclsmob/tree/master/pytorch\nhttps://github.com/facebookresearch/pycls/\nhttps://github.com/iduta/pyconv", "votes": 1 }, { "id": 825860, "postDate": "2020-04-29T09:41:09.700Z", "content": "Almost all possible weights of Keras.\n- https://www.kaggle.com/ipythonx/keras-pretrained-imagenet-weights\n- https://www.kaggle.com/ipythonx/efficientnet-keras-noisystudent-weights-b0b7
", "rawMarkdown": "Almost all possible weights of Keras.\n- https://www.kaggle.com/ipythonx/keras-pretrained-imagenet-weights\n- https://www.kaggle.com/ipythonx/efficientnet-keras-noisystudent-weights-b0b7", "votes": 1 }, { "id": 824671, "postDate": "2020-04-28T14:19:49.163Z", "content": "This google paper is using the data from here. According to the paper it's freely available, but looks like only for institutions. So I guess this data is excluded from the competition, right? @wouterbulten could you please comment? Many thanks!
", "rawMarkdown": "[This](https://www.nature.com/articles/s41746-019-0112-2.epdf?) google paper is using the data from [here](https://portal.gdc.cancer.gov/projects/TCGA-PRAD). According to the paper it's freely available, but looks like only for institutions. So I guess this data is excluded from the competition, right? @wouterbulten could you please comment? Many thanks!", "votes": 1, "replies": [ { "id": 824836, "postDate": "2020-04-28T16:11:29.103Z", "content": "Unfortunately, using that dataset is not allowed as it is not publicly available to everyone.
\n\nEdit: freely accessible slides are allowed, see my reply below.
", "rawMarkdown": "Unfortunately, using that dataset is not allowed as it is not publicly available to everyone.\n\nEdit: freely accessible slides are allowed, see my reply below." }, { "id": 825047, "postDate": "2020-04-28T18:39:52.047Z", "content": "thank you for the fast and clear answer!
", "rawMarkdown": "thank you for the fast and clear answer!" }, { "id": 825435, "postDate": "2020-04-29T02:19:20.267Z", "rawMarkdown": "", "isDeleted": true }, { "id": 827279, "postDate": "2020-04-30T07:43:58.217Z", "content": "We had an internal discussion and the freely accessible slides can be used for the competition. My apologies for the confusion. Some parts of the TCGA data is not freely accessible unless you apply for access, those parts are not allowed for the competition.
", "rawMarkdown": "We had an internal discussion and the freely accessible slides can be used for the competition. My apologies for the confusion. Some parts of the TCGA data is not freely accessible unless you apply for access, those parts are not allowed for the competition. ", "votes": 4 }, { "id": 883660, "postDate": "2020-06-12T20:08:41.440Z", "content": "Are there Gleason (or ISUP) scores associated with these slides? Or did Google take the slides, put them before a panel of experts and then use those expert scores? Are the scores released anywhere?
", "rawMarkdown": "Are there Gleason (or ISUP) scores associated with these slides? Or did Google take the slides, put them before a panel of experts and then use those expert scores? Are the scores released anywhere?" }, { "id": 886613, "postDate": "2020-06-15T06:55:33.500Z", "content": "The TCGA dataset does contain the Gleason score. In Google's paper where they have used this dataset, they asked a panel of pathologists to annotate and grade the tumor regions. Those additional annotations are not public as far as I know.
", "rawMarkdown": "The TCGA dataset does contain the Gleason score. In Google's paper where they have used this dataset, they asked a panel of pathologists to annotate and grade the tumor regions. Those additional annotations are not public as far as I know. " }, { "id": 887182, "postDate": "2020-06-15T14:23:17.067Z", "content": "@wouterbulten I am not sure about there not being any Gleason scores available for TCGA. There are bcr xml files available for about 500 cases, some of which have a Gleason score, but not all of them. \nI still have to see whether these Gleason score properly match with the ID's on the diagnostic slides that are freely available for download.
\n\nThe page can can be found here (I find the gdc site very confusing):
", "rawMarkdown": "@wouterbulten I am not sure about there not being any Gleason scores available for TCGA. There are bcr xml files available for about 500 cases, some of which have a Gleason score, but not all of them. \nI still have to see whether these Gleason score properly match with the ID's on the diagnostic slides that are freely available for download.\n\nThe page can can be found [here](https://portal.gdc.cancer.gov/repository?facetTab=cases&filters=%7B%22op%22%3A%22and%22%2C%22content%22%3A%5B%7B%22op%22%3A%22in%22%2C%22content%22%3A%7B%22field%22%3A%22cases.primary_site%22%2C%22value%22%3A%5B%22prostate%20gland%22%5D%7D%7D%2C%7B%22op%22%3A%22in%22%2C%22content%22%3A%7B%22field%22%3A%22files.access%22%2C%22value%22%3A%5B%22open%22%5D%7D%7D%2C%7B%22op%22%3A%22in%22%2C%22content%22%3A%7B%22field%22%3A%22files.data_format%22%2C%22value%22%3A%5B%22bcr%20xml%22%5D%7D%7D%5D%7D) (I find the gdc site very confusing): " }, { "id": 888251, "postDate": "2020-06-16T08:20:45.153Z", "content": "I think you have misread my reply. The TCGA set does have associated Gleason scores, in terms of case-level labels. As far as I know there are no region level annotations of the tumor itself.
", "rawMarkdown": "I think you have misread my reply. The TCGA set does have associated Gleason scores, in terms of case-level labels. As far as I know there are no region level annotations of the tumor itself. " }, { "id": 888427, "postDate": "2020-06-16T10:45:16.733Z", "content": "Oops I misread the sentence completely, my apologies.
", "rawMarkdown": "Oops I misread the sentence completely, my apologies." } ] }, { "id": 822380, "postDate": "2020-04-26T21:23:03.613Z", "content": "Gland Segmentation Challenge Data\nhttps://warwick.ac.uk/fac/sci/dcs/research/tia/glascontest/download/
", "rawMarkdown": "Gland Segmentation Challenge Data\nhttps://warwick.ac.uk/fac/sci/dcs/research/tia/glascontest/download/", "votes": 1, "replies": [ { "id": 855795, "postDate": "2020-05-21T08:08:38.397Z", "content": "In addition to that, there is also the CRAG dataset:
\n\nhttps://warwick.ac.uk/fac/sci/dcs/research/tia/data/mildnet/
", "rawMarkdown": "In addition to that, there is also the CRAG dataset:\n\nhttps://warwick.ac.uk/fac/sci/dcs/research/tia/data/mildnet/" } ] }, { "id": 819798, "postDate": "2020-04-24T21:52:50.550Z", "content": "https://jgamper.github.io/PanNukeDataset/
", "rawMarkdown": "https://jgamper.github.io/PanNukeDataset/", "votes": 2 }, { "id": 944377, "postDate": "2020-07-25T04:18:11.320Z", "content": "Nice!
", "rawMarkdown": "Nice!" }, { "id": 937382, "postDate": "2020-07-21T01:27:23.287Z", "content": "https://github.com/deroneriksson/python-wsi-preprocessing/tree/master/deephistopath/wsi
", "rawMarkdown": "https://github.com/deroneriksson/python-wsi-preprocessing/tree/master/deephistopath/wsi" }, { "id": 935738, "postDate": "2020-07-19T15:42:43.417Z", "content": "https://github.com/lukemelas/EfficientNet-PyTorch
", "rawMarkdown": "https://github.com/lukemelas/EfficientNet-PyTorch" }, { "id": 935737, "postDate": "2020-07-19T15:42:19.250Z", "content": "https://github.com/qubvel/ttach
", "rawMarkdown": "https://github.com/qubvel/ttach" }, { "id": 933926, "postDate": "2020-07-18T06:12:16.913Z", "content": "https://github.com/lukemelas/EfficientNet-PyTorch\nhttps://github.com/Cadene/pretrained-models.pytorch
", "rawMarkdown": "https://github.com/lukemelas/EfficientNet-PyTorch\nhttps://github.com/Cadene/pretrained-models.pytorch\n" }, { "id": 933143, "postDate": "2020-07-17T14:18:39.243Z", "content": "https://github.com/machine-perception-robotics-group/attention_branch_network
", "rawMarkdown": "https://github.com/machine-perception-robotics-group/attention_branch_network" }, { "id": 931058, "postDate": "2020-07-15T23:48:41.220Z", "content": "https://github.com/DigitalSlideArchive/HistomicsTK\nhttps://github.com/digantamisra98/Mish\nhttps://github.com/tensorflow/tpu/tree/master/models/official/mnasnet/mixnet\nThe openly available prostate slides of\nhttps://portal.gdc.cancer.gov/repository?facetTab=cases&filters=%7B%22op%22%3A%22and%22%2C%22content%22%3A%5B%7B%22op%22%3A%22in%22%2C%22content%22%3A%7B%22field%22%3A%22cases.primary_site%22%2C%22value%22%3A%5B%22prostate%20gland%22%5D%7D%7D%2C%7B%22op%22%3A%22in%22%2C%22content%22%3A%7B%22field%22%3A%22cases.project.program.name%22%2C%22value%22%3A%5B%22TCGA%22%5D%7D%7D%2C%7B%22op%22%3A%22in%22%2C%22content%22%3A%7B%22field%22%3A%22files.access%22%2C%22value%22%3A%5B%22open%22%5D%7D%7D%2C%7B%22op%22%3A%22in%22%2C%22content%22%3A%7B%22field%22%3A%22files.experimental_strategy%22%2C%22value%22%3A%5B%22Diagnostic%20Slide%22%5D%7D%7D%5D%7D
", "rawMarkdown": "https://github.com/DigitalSlideArchive/HistomicsTK\nhttps://github.com/digantamisra98/Mish\nhttps://github.com/tensorflow/tpu/tree/master/models/official/mnasnet/mixnet\nThe openly available prostate slides of\nhttps://portal.gdc.cancer.gov/repository?facetTab=cases&filters=%7B%22op%22%3A%22and%22%2C%22content%22%3A%5B%7B%22op%22%3A%22in%22%2C%22content%22%3A%7B%22field%22%3A%22cases.primary_site%22%2C%22value%22%3A%5B%22prostate%20gland%22%5D%7D%7D%2C%7B%22op%22%3A%22in%22%2C%22content%22%3A%7B%22field%22%3A%22cases.project.program.name%22%2C%22value%22%3A%5B%22TCGA%22%5D%7D%7D%2C%7B%22op%22%3A%22in%22%2C%22content%22%3A%7B%22field%22%3A%22files.access%22%2C%22value%22%3A%5B%22open%22%5D%7D%7D%2C%7B%22op%22%3A%22in%22%2C%22content%22%3A%7B%22field%22%3A%22files.experimental_strategy%22%2C%22value%22%3A%5B%22Diagnostic%20Slide%22%5D%7D%7D%5D%7D\n\n" }, { "id": 931056, "postDate": "2020-07-15T23:45:26.073Z", "content": "https://github.com/lukemelas/EfficientNet-PyTorch\nhttps://pytorch.org/docs/stable/torchvision/models.html\nhttps://github.com/pytorch/pytorch\nhttps://github.com/osmr/imgclsmob/tree/master/pytorch\nhttps://github.com/MarvinTeichmann/pyvision\nhttps://github.com/MarvinTeichmann/TorchLab\nhttps://github.com/ildoonet/pytorch-gradual-warmup-lr.git
\n\nAs well as models/weights trained locally for this challenge
", "rawMarkdown": "https://github.com/lukemelas/EfficientNet-PyTorch\nhttps://pytorch.org/docs/stable/torchvision/models.html\nhttps://github.com/pytorch/pytorch\nhttps://github.com/osmr/imgclsmob/tree/master/pytorch\nhttps://github.com/MarvinTeichmann/pyvision\nhttps://github.com/MarvinTeichmann/TorchLab\nhttps://github.com/ildoonet/pytorch-gradual-warmup-lr.git\n\nAs well as models/weights trained locally for this challenge\n\n" }, { "id": 931029, "postDate": "2020-07-15T22:48:09.040Z", "content": "https://github.com/qubvel/efficientnet\nhttps://github.com/Zhengyushan/adaptive_color_deconvolution
", "rawMarkdown": "https://github.com/qubvel/efficientnet\nhttps://github.com/Zhengyushan/adaptive_color_deconvolution" }, { "id": 929681, "postDate": "2020-07-14T20:59:29.607Z", "content": "https://github.com/joe-siyuan-qiao/WeightStandardization
", "rawMarkdown": "https://github.com/joe-siyuan-qiao/WeightStandardization" }, { "id": 929416, "postDate": "2020-07-14T16:50:50.067Z", "content": "https://github.com/facebookresearch/semi-supervised-ImageNet1K-models
", "rawMarkdown": "https://github.com/facebookresearch/semi-supervised-ImageNet1K-models" }, { "id": 929284, "postDate": "2020-07-14T15:29:52.043Z", "content": "https://github.com/ansleliu/MixNet-PyTorch
", "rawMarkdown": "https://github.com/ansleliu/MixNet-PyTorch" }, { "id": 927694, "postDate": "2020-07-13T14:49:02.757Z", "content": "https://github.com/qubvel/ttach\nhttps://github.com/Peter554/StainTools\nhttps://github.com/schaugf/HEnorm_python\nhttps://github.com/tcxxxx/WSI-analysis\nhttps://github.com/osmr/imgclsmob\nhttps://github.com/Cadene/pretrained-models.pytorch\nhttps://github.com/pytorch/vision\nhttps://github.com/lukemelas/EfficientNet-PyTorch
", "rawMarkdown": "[https://github.com/qubvel/ttach](https://github.com/qubvel/ttach)\n[https://github.com/Peter554/StainTools](https://github.com/Peter554/StainTools)\n[https://github.com/schaugf/HEnorm_python](https://github.com/schaugf/HEnorm_python)\n[https://github.com/tcxxxx/WSI-analysis](https://github.com/tcxxxx/WSI-analysis)\n[https://github.com/osmr/imgclsmob](https://github.com/osmr/imgclsmob)\n[https://github.com/Cadene/pretrained-models.pytorch](https://github.com/Cadene/pretrained-models.pytorch)\n[https://github.com/pytorch/vision](https://github.com/pytorch/vision)\n[https://github.com/lukemelas/EfficientNet-PyTorch](https://github.com/lukemelas/EfficientNet-PyTorch)" }, { "id": 926006, "postDate": "2020-07-12T12:32:11.530Z", "content": "https://github.com/lukemelas/EfficientNet-PyTorch\nhttps://github.com/qubvel/segmentation_models.pytorch\nhttps://github.com/facebookresearch/pycls/\nhttps://zenodo.org/record/1485967\nhttps://github.com/rwightman/pytorch-image-models/
", "rawMarkdown": "https://github.com/lukemelas/EfficientNet-PyTorch\nhttps://github.com/qubvel/segmentation_models.pytorch\nhttps://github.com/facebookresearch/pycls/\nhttps://zenodo.org/record/1485967\nhttps://github.com/rwightman/pytorch-image-models/" }, { "id": 924923, "postDate": "2020-07-11T17:51:20.303Z", "content": "https://github.com/qubvel/efficientnet
", "rawMarkdown": "https://github.com/qubvel/efficientnet" }, { "id": 916561, "postDate": "2020-07-05T19:01:33.763Z", "content": "https://github.com/rwightman/pytorch-image-models\nhttps://pytorch.org/hub/facebookresearch_WSL-Images_resnext/
", "rawMarkdown": "https://github.com/rwightman/pytorch-image-models\nhttps://pytorch.org/hub/facebookresearch_WSL-Images_resnext/" }, { "id": 916301, "postDate": "2020-07-05T14:36:02.690Z", "content": "https://github.com/zhanghang1989/ResNeSt\nhttps://github.com/lukemelas/EfficientNet-PyTorch\nhttps://github.com/Cadene/pretrained-models.pytorch\nhttps://github.com/rwightman/pytorch-image-models/
", "rawMarkdown": "https://github.com/zhanghang1989/ResNeSt\nhttps://github.com/lukemelas/EfficientNet-PyTorch\nhttps://github.com/Cadene/pretrained-models.pytorch\nhttps://github.com/rwightman/pytorch-image-models/" }, { "id": 914184, "postDate": "2020-07-03T17:02:22.720Z", "content": "https://github.com/qubvel/classification_models\nhttps://github.com/qubvel/efficientnet
", "rawMarkdown": "https://github.com/qubvel/classification_models\nhttps://github.com/qubvel/efficientnet" }, { "id": 911916, "postDate": "2020-07-02T05:27:20.217Z", "content": "Could we use model trained from any public notebook of the competition for example there is one guy who did public his notebook having score of 87.
\n\nWe use of following notebook to create one model\n https://www.kaggle.com/haqishen/train-efficientnet-b0-w-36-tiles-256-lb0-87/output\nand one public model from follow inference\n https://www.kaggle.com/haqishen/panda-inference-w-36-tiles-256/
\n\nand then ensemble with our models.
", "rawMarkdown": "Could we use model trained from any public notebook of the competition for example there is one guy who did public his notebook having score of 87. \n\nWe use of following notebook to create one model\n https://www.kaggle.com/haqishen/train-efficientnet-b0-w-36-tiles-256-lb0-87/output\nand one public model from follow inference\n https://www.kaggle.com/haqishen/panda-inference-w-36-tiles-256/\n\nand then ensemble with our models.\n" }, { "id": 911915, "postDate": "2020-07-02T05:25:53.470Z", "content": "DataSets and APIs\n1. I am using dataset converted to PNG on AWS as it was not possible to convert in kaggle\n ( No external data was added )\n Followings are links , I have made them public\n - https://www.kaggle.com/rajnishe/panda-cancer-l2-36tile256-p1\n - https://www.kaggle.com/rajnishe/panda-cancer-l2-36tile256-p2\n - https://www.kaggle.com/rajnishe/panda-cancer-l2-sqtile32-v1\n - https://www.kaggle.com/rajnishe/panda-cancer-l2-36tile208-p1\n - https://www.kaggle.com/rajnishe/panda-cancer-l2-36tile208-p2\n -https://www.kaggle.com/micheomaano/tf-record-256-256-48
\n\nhttps://github.com/lukemelas/EfficientNet-PyTorch
", "rawMarkdown": "https://github.com/lukemelas/EfficientNet-PyTorch" }, { "id": 906733, "postDate": "2020-06-29T14:11:13.850Z", "content": "https://github.com/qubvel/efficientnet
", "rawMarkdown": "https://github.com/qubvel/efficientnet" }, { "id": 902765, "postDate": "2020-06-26T10:47:59.237Z", "content": "https://github.com/lukemelas/EfficientNet-PyTorch\nhttps://pytorch.org/docs/stable/torchvision/models.html
", "rawMarkdown": "https://github.com/lukemelas/EfficientNet-PyTorch\nhttps://pytorch.org/docs/stable/torchvision/models.html" }, { "id": 887214, "postDate": "2020-06-15T14:45:53.190Z", "content": "Hi. Can I use Whole Slide Images from TCGA or just the images like no annotations or anything?\nThe Cancer Genome Atlas is considered as \"data from wild\".
", "rawMarkdown": "Hi. Can I use Whole Slide Images from TCGA or just the images like no annotations or anything?\nThe Cancer Genome Atlas is considered as \"data from wild\". ", "replies": [ { "id": 889166, "postDate": "2020-06-16T20:07:22.473Z", "content": "Kindly clear me with this point. It would be a great help.....!
", "rawMarkdown": "Kindly clear me with this point. It would be a great help.....!" } ] }, { "id": 886343, "postDate": "2020-06-15T00:23:25.190Z", "content": "https://github.com/rwightman/pytorch-image-models/\nhttps://github.com/Cadene/pretrained-models.pytorch\nhttps://github.com/open-mmlab/mmcv
", "rawMarkdown": "https://github.com/rwightman/pytorch-image-models/\nhttps://github.com/Cadene/pretrained-models.pytorch\nhttps://github.com/open-mmlab/mmcv" }, { "id": 879473, "postDate": "2020-06-09T13:55:56.013Z", "content": "Hi I am new in kaggle contest Can I use my process output data as input to another new kernel to train the model and submit my prediction
", "rawMarkdown": "Hi I am new in kaggle contest Can I use my process output data as input to another new kernel to train the model and submit my prediction", "replies": [ { "id": 879500, "postDate": "2020-06-09T14:17:42.530Z", "content": "Could you elaborate on what you mean with \"process output data\"?
", "rawMarkdown": "Could you elaborate on what you mean with \"process output data\"?" }, { "id": 879521, "postDate": "2020-06-09T14:30:35.913Z", "content": "Yes, I process this contest data and save it as a zip file now I want to use the process data as input by open a new kernel. Can I do that? Please suggest to me.
", "rawMarkdown": "Yes, I process this contest data and save it as a zip file now I want to use the process data as input by open a new kernel. Can I do that? Please suggest to me." }, { "id": 879629, "postDate": "2020-06-09T15:40:41.810Z", "content": "Yes, that's allowed given that you don't include private external data. Other people have made specific derivatives of the dataset (under the same license). For example exporting tiles, or specific levels of the images. If you are one of the competitions winners, this pre-processing step is part of your solution and needs to be made open source.
\n\nWhat's not allowed (as an example): Processing the dataset with a private deep learning system (trained on private data) and using the output to train a system for the competition.
", "rawMarkdown": "Yes, that's allowed given that you don't include private external data. Other people have made specific derivatives of the dataset (under the same license). For example exporting tiles, or specific levels of the images. If you are one of the competitions winners, this pre-processing step is part of your solution and needs to be made open source.\n\nWhat's not allowed (as an example): Processing the dataset with a private deep learning system (trained on private data) and using the output to train a system for the competition. \n" }, { "id": 879635, "postDate": "2020-06-09T15:42:57.130Z", "content": "Thanks, for reply
", "rawMarkdown": "Thanks, for reply " } ] }, { "id": 872975, "postDate": "2020-06-03T16:55:58.560Z", "content": "https://github.com/creafz/pytorch-cnn-finetune\nhttps://github.com/zhanghang1989/ResNeSt
", "rawMarkdown": "https://github.com/creafz/pytorch-cnn-finetune\nhttps://github.com/zhanghang1989/ResNeSt" }, { "id": 871235, "postDate": "2020-06-02T08:04:01.017Z", "content": "Is it allowed to use data from this dataset: https://gleason2019.grand-challenge.org/Home/ ?
", "rawMarkdown": "Is it allowed to use data from this dataset: https://gleason2019.grand-challenge.org/Home/ ?\n", "replies": [ { "id": 871249, "postDate": "2020-06-02T08:11:03.270Z", "content": "Unfortunately, the website does not state the license of the dataset. The rules of the challenge state that \"Participating teams are not allowed to share the data\" which would indicate that you must be part of the Gleason2019 challenge to use the data.
\n\nIf you would like to use this dataset, please check with the organizers (https://gleason2019.grand-challenge.org/Contact/) whether the data is publicly available and under what license.
", "rawMarkdown": "Unfortunately, the website does not state the license of the dataset. The rules of the challenge state that \"Participating teams are not allowed to share the data\" which would indicate that you must be part of the Gleason2019 challenge to use the data.\n\nIf you would like to use this dataset, please check with the organizers (https://gleason2019.grand-challenge.org/Contact/) whether the data is publicly available and under what license.", "votes": 3 }, { "id": 878762, "postDate": "2020-06-08T20:03:27.817Z", "content": "Any word on if this data is usable? From what I can see the data is publicly available
", "rawMarkdown": "Any word on if this data is usable? From what I can see the data is publicly available", "votes": 1 } ] }, { "id": 869947, "postDate": "2020-06-01T11:50:21.517Z", "content": "https://github.com/Cadene/pretrained-models.pytorch\nhttps://github.com/rwightman/pytorch-image-models\nhttps://github.com/lukemelas/EfficientNet-PyTorch
", "rawMarkdown": "https://github.com/Cadene/pretrained-models.pytorch\nhttps://github.com/rwightman/pytorch-image-models\nhttps://github.com/lukemelas/EfficientNet-PyTorch", "replies": [ { "id": 930308, "postDate": "2020-07-15T10:51:46.807Z", "content": "https://pytorch.org/docs/stable/torchvision/models.html\nhttps://github.com/zhanghang1989/ResNeSt
", "rawMarkdown": "https://pytorch.org/docs/stable/torchvision/models.html\nhttps://github.com/zhanghang1989/ResNeSt" } ] }, { "id": 816232, "postDate": "2020-04-22T07:38:35.870Z", "content": "Might consider using this one, but it seems like 3TB....\nhttps://camelyon17.grand-challenge.org/Data/
", "rawMarkdown": "Might consider using this one, but it seems like 3TB....\nhttps://camelyon17.grand-challenge.org/Data/", "replies": [ { "id": 816258, "postDate": "2020-04-22T08:08:23.867Z", "content": "This is for lymph node metastasis, not the same tissue type. But the solutions may of course be of interest.
", "rawMarkdown": "This is for lymph node metastasis, not the same tissue type. But the solutions may of course be of interest." }, { "id": 816383, "postDate": "2020-04-22T10:37:15.743Z", "content": "Oh, whoops! Thanks for the heads up. That prevented me from downloading 3TB of useless data🤕
", "rawMarkdown": "Oh, whoops! Thanks for the heads up. That prevented me from downloading 3TB of useless data🤕 " } ] }, { "id": 837859, "postDate": "2020-05-08T05:05:40.853Z", "rawMarkdown": "", "isDeleted": true } ], "comments": [ { "id": 824859, "author_name": "Wouter Bulten", "author_url": "", "post_date": "2020-04-28T16:22:14.287000", "content": "If you are looking for additional freely available data of prostate tissue: take a look at the PESO dataset. The dataset consists of whole slide images of prostatectomies with precise annotations of the epithelium.
", "votes": 5, "replies": [ { "id": 829213, "author_name": "j&j_ML", "author_url": "", "post_date": "2020-05-01T15:44:30.263000", "content": "", "votes": 0, "replies": [] }, { "id": 906274, "author_name": "Jaideep", "author_url": "", "post_date": "2020-06-29T07:14:23.180000", "content": "@wouterbulten i suppose it dsnt has grades defined. i has only given the masks
", "votes": 0, "replies": [] }, { "id": 906288, "author_name": "Wouter Bulten", "author_url": "", "post_date": "2020-06-29T07:27:37.213000", "content": "Indeed, no Gleason grades. It can be used for pre-training or if you want to go for a segmentation method.
", "votes": 1, "replies": [] } ] }, { "id": 816567, "author_name": "deepcsquid", "author_url": "", "post_date": "2020-04-22T12:35:50.910000", "content": "Maybe I've just not read all the terms and conditions, but am I allowed to train a model on proprietary hardware and upload the weights for inference? Or do I need to disclose the model+weights? Or only the weights?
", "votes": 3, "replies": [ { "id": 820141, "author_name": "", "author_url": "", "post_date": "2020-04-25T07:16:20.897000", "content": "", "votes": 0, "replies": [] }, { "id": 878837, "author_name": "", "author_url": "", "post_date": "2020-06-08T22:48:26.860000", "content": "", "votes": 1, "replies": [] }, { "id": 879151, "author_name": "Wouter Bulten", "author_url": "", "post_date": "2020-06-09T08:33:47.837000", "content": "Yes you are allowed to train models locally on your own hardware, and upload these models to be used in an inference kernel. Note that rules regarding external data still apply. Also, winning solutions need to make the training code open source, see https://www.kaggle.com/c/prostate-cancer-grade-assessment/rules
", "votes": 1, "replies": [] }, { "id": 879245, "author_name": "", "author_url": "", "post_date": "2020-06-09T10:36:01.383000", "content": "", "votes": 0, "replies": [] }, { "id": 879314, "author_name": "Wouter Bulten", "author_url": "", "post_date": "2020-06-09T11:45:39.623000", "content": "If those weights are from models trained on the competition data, then you don't have to make them public. You have to make it public when:
\n\ncan we also make use of Gleason2019 challenge data.. it has got over now .
", "votes": 0, "replies": [] } ] }, { "id": 929999, "author_name": "hirune924", "author_url": "", "post_date": "2020-07-15T06:09:34.570000", "content": "https://github.com/joe-siyuan-qiao/WeightStandardization\nhttps://github.com/facebookresearch/semi-supervised-ImageNet1K-models\nhttps://github.com/lukemelas/EfficientNet-PyTorch\nhttps://github.com/qubvel/segmentation_models.pytorch\nhttps://github.com/facebookresearch/pycls/\nhttps://zenodo.org/record/1485967\nhttps://github.com/rwightman/pytorch-image-models/\nhttps://github.com/google-research/big_transfer\nhttps://github.com/qubvel/efficientnet\nhttps://github.com/Cadene/pretrained-models.pytorch
", "votes": 1, "replies": [] }, { "id": 917615, "author_name": "", "author_url": "", "post_date": "2020-07-06T16:33:07.740000", "content": "https://github.com/yhhhli/RegNet-Pytorch
", "votes": 1, "replies": [] }, { "id": 899623, "author_name": "Dieter", "author_url": "", "post_date": "2020-06-24T10:38:31.840000", "content": "https://github.com/osmr/imgclsmob/tree/master/pytorch\nhttps://github.com/facebookresearch/pycls/\nhttps://github.com/iduta/pyconv
", "votes": 1, "replies": [] }, { "id": 825860, "author_name": "Innat", "author_url": "", "post_date": "2020-04-29T09:41:09.700000", "content": "Almost all possible weights of Keras.\n- https://www.kaggle.com/ipythonx/keras-pretrained-imagenet-weights\n- https://www.kaggle.com/ipythonx/efficientnet-keras-noisystudent-weights-b0b7
", "votes": 1, "replies": [] }, { "id": 824671, "author_name": "abzaliev", "author_url": "", "post_date": "2020-04-28T14:19:49.163000", "content": "This google paper is using the data from here. According to the paper it's freely available, but looks like only for institutions. So I guess this data is excluded from the competition, right? @wouterbulten could you please comment? Many thanks!
", "votes": 1, "replies": [ { "id": 824836, "author_name": "Wouter Bulten", "author_url": "", "post_date": "2020-04-28T16:11:29.103000", "content": "Unfortunately, using that dataset is not allowed as it is not publicly available to everyone.
\n\nEdit: freely accessible slides are allowed, see my reply below.
", "votes": 0, "replies": [] }, { "id": 825047, "author_name": "abzaliev", "author_url": "", "post_date": "2020-04-28T18:39:52.047000", "content": "thank you for the fast and clear answer!
", "votes": 0, "replies": [] }, { "id": 825435, "author_name": "", "author_url": "", "post_date": "2020-04-29T02:19:20.267000", "content": "", "votes": 0, "replies": [] }, { "id": 827279, "author_name": "Wouter Bulten", "author_url": "", "post_date": "2020-04-30T07:43:58.217000", "content": "We had an internal discussion and the freely accessible slides can be used for the competition. My apologies for the confusion. Some parts of the TCGA data is not freely accessible unless you apply for access, those parts are not allowed for the competition.
", "votes": 4, "replies": [] }, { "id": 883660, "author_name": "Josh Varty", "author_url": "", "post_date": "2020-06-12T20:08:41.440000", "content": "Are there Gleason (or ISUP) scores associated with these slides? Or did Google take the slides, put them before a panel of experts and then use those expert scores? Are the scores released anywhere?
", "votes": 0, "replies": [] }, { "id": 886613, "author_name": "Wouter Bulten", "author_url": "", "post_date": "2020-06-15T06:55:33.500000", "content": "The TCGA dataset does contain the Gleason score. In Google's paper where they have used this dataset, they asked a panel of pathologists to annotate and grade the tumor regions. Those additional annotations are not public as far as I know.
", "votes": 0, "replies": [] }, { "id": 887182, "author_name": "Stephan", "author_url": "", "post_date": "2020-06-15T14:23:17.067000", "content": "@wouterbulten I am not sure about there not being any Gleason scores available for TCGA. There are bcr xml files available for about 500 cases, some of which have a Gleason score, but not all of them. \nI still have to see whether these Gleason score properly match with the ID's on the diagnostic slides that are freely available for download.
\n\nThe page can can be found here (I find the gdc site very confusing):
", "votes": 0, "replies": [] }, { "id": 888251, "author_name": "Wouter Bulten", "author_url": "", "post_date": "2020-06-16T08:20:45.153000", "content": "I think you have misread my reply. The TCGA set does have associated Gleason scores, in terms of case-level labels. As far as I know there are no region level annotations of the tumor itself.
", "votes": 0, "replies": [] }, { "id": 888427, "author_name": "Stephan", "author_url": "", "post_date": "2020-06-16T10:45:16.733000", "content": "Oops I misread the sentence completely, my apologies.
", "votes": 0, "replies": [] } ] }, { "id": 822380, "author_name": "Larxel", "author_url": "", "post_date": "2020-04-26T21:23:03.613000", "content": "Gland Segmentation Challenge Data\nhttps://warwick.ac.uk/fac/sci/dcs/research/tia/glascontest/download/
", "votes": 1, "replies": [ { "id": 855795, "author_name": "SimonGraham", "author_url": "", "post_date": "2020-05-21T08:08:38.397000", "content": "In addition to that, there is also the CRAG dataset:
\n\nhttps://warwick.ac.uk/fac/sci/dcs/research/tia/data/mildnet/
", "votes": 0, "replies": [] } ] }, { "id": 819798, "author_name": "kvigly", "author_url": "", "post_date": "2020-04-24T21:52:50.550000", "content": "https://jgamper.github.io/PanNukeDataset/
", "votes": 2, "replies": [] }, { "id": 944377, "author_name": "dalayrrewnp", "author_url": "", "post_date": "2020-07-25T04:18:11.320000", "content": "Nice!
", "votes": 0, "replies": [] }, { "id": 937382, "author_name": "jwelliav", "author_url": "", "post_date": "2020-07-21T01:27:23.287000", "content": "https://github.com/deroneriksson/python-wsi-preprocessing/tree/master/deephistopath/wsi
", "votes": 0, "replies": [] }, { "id": 935738, "author_name": "wangyunpeng_bio", "author_url": "", "post_date": "2020-07-19T15:42:43.417000", "content": "https://github.com/lukemelas/EfficientNet-PyTorch
", "votes": 0, "replies": [] }, { "id": 935737, "author_name": "wangyunpeng_bio", "author_url": "", "post_date": "2020-07-19T15:42:19.250000", "content": "https://github.com/qubvel/ttach
", "votes": 0, "replies": [] }, { "id": 933926, "author_name": "YUYUTA", "author_url": "", "post_date": "2020-07-18T06:12:16.913000", "content": "https://github.com/lukemelas/EfficientNet-PyTorch\nhttps://github.com/Cadene/pretrained-models.pytorch
", "votes": 0, "replies": [] }, { "id": 933143, "author_name": "Inoichan", "author_url": "", "post_date": "2020-07-17T14:18:39.243000", "content": "https://github.com/machine-perception-robotics-group/attention_branch_network
", "votes": 0, "replies": [] }, { "id": 931058, "author_name": "Stephan", "author_url": "", "post_date": "2020-07-15T23:48:41.220000", "content": "https://github.com/DigitalSlideArchive/HistomicsTK\nhttps://github.com/digantamisra98/Mish\nhttps://github.com/tensorflow/tpu/tree/master/models/official/mnasnet/mixnet\nThe openly available prostate slides of\nhttps://portal.gdc.cancer.gov/repository?facetTab=cases&filters=%7B%22op%22%3A%22and%22%2C%22content%22%3A%5B%7B%22op%22%3A%22in%22%2C%22content%22%3A%7B%22field%22%3A%22cases.primary_site%22%2C%22value%22%3A%5B%22prostate%20gland%22%5D%7D%7D%2C%7B%22op%22%3A%22in%22%2C%22content%22%3A%7B%22field%22%3A%22cases.project.program.name%22%2C%22value%22%3A%5B%22TCGA%22%5D%7D%7D%2C%7B%22op%22%3A%22in%22%2C%22content%22%3A%7B%22field%22%3A%22files.access%22%2C%22value%22%3A%5B%22open%22%5D%7D%7D%2C%7B%22op%22%3A%22in%22%2C%22content%22%3A%7B%22field%22%3A%22files.experimental_strategy%22%2C%22value%22%3A%5B%22Diagnostic%20Slide%22%5D%7D%7D%5D%7D
", "votes": 0, "replies": [] }, { "id": 931056, "author_name": "MarvMind", "author_url": "", "post_date": "2020-07-15T23:45:26.073000", "content": "https://github.com/lukemelas/EfficientNet-PyTorch\nhttps://pytorch.org/docs/stable/torchvision/models.html\nhttps://github.com/pytorch/pytorch\nhttps://github.com/osmr/imgclsmob/tree/master/pytorch\nhttps://github.com/MarvinTeichmann/pyvision\nhttps://github.com/MarvinTeichmann/TorchLab\nhttps://github.com/ildoonet/pytorch-gradual-warmup-lr.git
\n\nAs well as models/weights trained locally for this challenge
", "votes": 0, "replies": [] }, { "id": 931029, "author_name": "Duc-Kinh Le Tran", "author_url": "", "post_date": "2020-07-15T22:48:09.040000", "content": "https://github.com/qubvel/efficientnet\nhttps://github.com/Zhengyushan/adaptive_color_deconvolution
", "votes": 0, "replies": [] }, { "id": 929681, "author_name": "Eek The Cat", "author_url": "", "post_date": "2020-07-14T20:59:29.607000", "content": "https://github.com/joe-siyuan-qiao/WeightStandardization
", "votes": 0, "replies": [] }, { "id": 929416, "author_name": "Jaideep", "author_url": "", "post_date": "2020-07-14T16:50:50.067000", "content": "https://github.com/facebookresearch/semi-supervised-ImageNet1K-models
", "votes": 0, "replies": [] }, { "id": 929284, "author_name": "fergusoci", "author_url": "", "post_date": "2020-07-14T15:29:52.043000", "content": "https://github.com/ansleliu/MixNet-PyTorch
", "votes": 0, "replies": [] }, { "id": 927694, "author_name": "Yovin Yahathugoda", "author_url": "", "post_date": "2020-07-13T14:49:02.757000", "content": "https://github.com/qubvel/ttach\nhttps://github.com/Peter554/StainTools\nhttps://github.com/schaugf/HEnorm_python\nhttps://github.com/tcxxxx/WSI-analysis\nhttps://github.com/osmr/imgclsmob\nhttps://github.com/Cadene/pretrained-models.pytorch\nhttps://github.com/pytorch/vision\nhttps://github.com/lukemelas/EfficientNet-PyTorch
", "votes": 0, "replies": [] }, { "id": 926006, "author_name": "MPWARE", "author_url": "", "post_date": "2020-07-12T12:32:11.530000", "content": "https://github.com/lukemelas/EfficientNet-PyTorch\nhttps://github.com/qubvel/segmentation_models.pytorch\nhttps://github.com/facebookresearch/pycls/\nhttps://zenodo.org/record/1485967\nhttps://github.com/rwightman/pytorch-image-models/
", "votes": 0, "replies": [] }, { "id": 924923, "author_name": "Elahi", "author_url": "", "post_date": "2020-07-11T17:51:20.303000", "content": "https://github.com/qubvel/efficientnet
", "votes": 0, "replies": [] }, { "id": 916561, "author_name": "Alexander Izvorski", "author_url": "", "post_date": "2020-07-05T19:01:33.763000", "content": "https://github.com/rwightman/pytorch-image-models\nhttps://pytorch.org/hub/facebookresearch_WSL-Images_resnext/
", "votes": 0, "replies": [] }, { "id": 916301, "author_name": "Syn", "author_url": "", "post_date": "2020-07-05T14:36:02.690000", "content": "https://github.com/zhanghang1989/ResNeSt\nhttps://github.com/lukemelas/EfficientNet-PyTorch\nhttps://github.com/Cadene/pretrained-models.pytorch\nhttps://github.com/rwightman/pytorch-image-models/
", "votes": 0, "replies": [] }, { "id": 914184, "author_name": "Richard Xiao", "author_url": "", "post_date": "2020-07-03T17:02:22.720000", "content": "https://github.com/qubvel/classification_models\nhttps://github.com/qubvel/efficientnet
", "votes": 0, "replies": [] }, { "id": 911916, "author_name": "Rajnish Chauhan", "author_url": "", "post_date": "2020-07-02T05:27:20.217000", "content": "Could we use model trained from any public notebook of the competition for example there is one guy who did public his notebook having score of 87.
\n\nWe use of following notebook to create one model\n https://www.kaggle.com/haqishen/train-efficientnet-b0-w-36-tiles-256-lb0-87/output\nand one public model from follow inference\n https://www.kaggle.com/haqishen/panda-inference-w-36-tiles-256/
\n\nand then ensemble with our models.
", "votes": 0, "replies": [] }, { "id": 911915, "author_name": "Rajnish Chauhan", "author_url": "", "post_date": "2020-07-02T05:25:53.470000", "content": "DataSets and APIs\n1. I am using dataset converted to PNG on AWS as it was not possible to convert in kaggle\n ( No external data was added )\n Followings are links , I have made them public\n - https://www.kaggle.com/rajnishe/panda-cancer-l2-36tile256-p1\n - https://www.kaggle.com/rajnishe/panda-cancer-l2-36tile256-p2\n - https://www.kaggle.com/rajnishe/panda-cancer-l2-sqtile32-v1\n - https://www.kaggle.com/rajnishe/panda-cancer-l2-36tile208-p1\n - https://www.kaggle.com/rajnishe/panda-cancer-l2-36tile208-p2\n -https://www.kaggle.com/micheomaano/tf-record-256-256-48
\n\n