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": 956812, "postDate": "2020-08-03T19:57:09.560Z", "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": 6 }, { "id": 1022625, "postDate": "2020-09-22T16:35:07.857Z", "content": "https://github.com/junyanz/pytorch-CycleGAN-and-pix2pix
\nhttps://github.com/magicleap/SuperPointPretrainedNetwork
\nhttps://github.com/fidler-lab/defgrid-release
\nhttps://github.com/aritra0593/Reinforced-Feature-Points
https://github.com/rwightman/gen-efficientnet-pytorch
\nhttps://github.com/Cadene/pretrained-models.pytorch
\nhttps://github.com/clovaai/rexnet
https://github.com/tensorflow/models/tree/master/research/object_detection
\nhttps://tfhub.dev/s?module-type=image-object-detection
Can we use full version of GLDv2 dataset?
", "rawMarkdown": "Can we use full version of GLDv2 dataset? ", "votes": 1, "replies": [ { "id": 978821, "postDate": "2020-08-20T12:17:30.180Z", "content": "It should be allowed.
", "rawMarkdown": "It should be allowed." } ] }, { "id": 1021591, "postDate": "2020-09-22T02:43:11.163Z", "content": "ADE-20k pretrained models: https://github.com/CSAILVision/semantic-segmentation-pytorch
\nImage matching: https://github.com/rpautrat/SuperPoint
\nhttps://github.com/cvg/Hierarchical-Localization
Wow, exciting! I've been wondering since the beginning if someone's gonna use SuperPoint and/or SuperGlue in this competition 😯 Can't wait to read your team's write-up (if you're going to write one) 😊
", "rawMarkdown": "Wow, exciting! I've been wondering since the beginning if someone's gonna use SuperPoint and/or SuperGlue in this competition 😯 Can't wait to read your team's write-up (if you're going to write one) 😊" } ] }, { "id": 982173, "postDate": "2020-08-23T05:56:05.253Z", "content": "can we use http://places2.csail.mit.edu/ if we open source our trained model. see the problem is it says non-commercial research, but if we open source the model, wouldn't that make it indirect?
", "rawMarkdown": "can we use http://places2.csail.mit.edu/ if we open source our trained model. see the problem is it says non-commercial research, but if we open source the model, wouldn't that make it indirect?", "votes": 2, "replies": [ { "id": 1001134, "postDate": "2020-09-07T04:57:35.870Z", "content": "have a look at this notebook by @rsmits
\nlooks like you can use this kaggle dataset
", "rawMarkdown": "have a look at [this notebook](https://www.kaggle.com/rsmits/tf-keras-effnet-b2-non-landmark-removal/data) by @rsmits \n\nlooks like you can use [this kaggle dataset](https://www.kaggle.com/rsmits/keras-vgg16-places365)" } ] }, { "id": 1030798, "postDate": "2020-09-29T01:31:30.620Z", "content": "https://github.com/tensorflow/models/tree/master/research/delf
\nhttps://www.kaggle.com/camaskew/delg-saved-models
\nhttps://www.kaggle.com/camaskew/baseline-landmark-retrieval-model
https://www.kaggle.com/camaskew/delg-saved-models
\nhttps://www.kaggle.com/keras/inceptionresnetv2
\nhttps://www.kaggle.com/keras/xception
https://www.kaggle.com/camaskew/baseline-landmark-retrieval-model
\nhttps://www.kaggle.com/camaskew/delg-saved-models
\nhttps://github.com/tensorflow/models/tree/master/research/delf
\nhttps://www.kaggle.com/c/landmark-retrieval-2020/data
https://github.com/lukemelas/EfficientNet-PyTorch
\nhttps://github.com/zylo117/Yet-Another-EfficientDet-Pytorch
https://github.com/pytorch/vision/tree/master/torchvision/models
\nhttps://github.com/qubvel/efficientnet
\nhttps://www.tensorflow.org/api_docs/python/tf/keras/applications
\nhttps://github.com/zhanghang1989/ResNeSt
\nhttps://github.com/magicleap/SuperGluePretrainedNetwork
\nhttps://www.kaggle.com/c/landmark-recognition-challenge
\nhttps://storage.googleapis.com/openimages/web/index.html
\nhttps://cocodataset.org/#home
https://github.com/PaddlePaddle/Research/tree/master/CV/landmark
", "rawMarkdown": "https://github.com/PaddlePaddle/Research/tree/master/CV/landmark" }, { "id": 1020546, "postDate": "2020-09-21T08:40:53.293Z", "content": "Pretrained model for frontal face detection (open-cv): https://github.com/opencv/opencv/blob/master/data/haarcascades/haarcascade_frontalface_default.xml
", "rawMarkdown": "Pretrained model for frontal face detection (open-cv): https://github.com/opencv/opencv/blob/master/data/haarcascades/haarcascade_frontalface_default.xml" }, { "id": 1019465, "postDate": "2020-09-20T13:09:25.060Z", "content": "MAGSAC
\n", "rawMarkdown": "**MAGSAC**\n- https://github.com/danini/magsac\n- https://github.com/ducha-aiki/pymagsac" }, { "id": 1018513, "postDate": "2020-09-19T18:11:36.297Z", "content": "https://www.kaggle.com/google/google-landmarks-dataset
\nhttps://github.com/tensorflow/models/tree/master/research/delf
\nhttps://github.com/eric-yyjau/pytorch-superpoint
Full GLv2 dataset: https://github.com/cvdfoundation/google-landmark
\nPre-trained models: https://github.com/rwightman/pytorch-image-models
\nConfiguration: https://github.com/rbgirshick/yacs
\nrapids.ai CUML: https://github.com/rapidsai/cuml
Wikipedia categories for GLR 2020
", "rawMarkdown": "[Wikipedia categories for GLR 2020](https://www.kaggle.com/steubk/wikipedia-categories-for-glr-2020) " }, { "id": 989813, "postDate": "2020-08-29T06:13:06.917Z", "content": "EfficientNet-PyTorch:
\nhttps://github.com/lukemelas/EfficientNet-PyTorch
If I train a model on one notebook (NB-1) and create another script file (SF-1) and load that model. Is that considered as a pre-trained model? and do I need to post that?
\nI train a model, save, and commit. How do I retrieve this after a few commits or load it onto another script file?
\n@maggiemd
Hi, I have a question about pre learned models.
\nIs it allowed to learn models using the train data once and save the models, and then when submitting, simply load the saved models from inputs.
\nIn other words, training part is done outside of the submitted kernel, and only prediction will be done.
\nI would really appreciate if someone can help me out with this question!
Yes. You can train the model anywhere you want, be it kaggle kernels or you personal system. For submission you need to take in the test data as inputs as generate the submission file as output. Hope it helps.
", "rawMarkdown": "Yes. You can train the model anywhere you want, be it kaggle kernels or you personal system. For submission you need to take in the test data as inputs as generate the submission file as output. Hope it helps.", "votes": 1 } ] }, { "id": 974532, "postDate": "2020-08-18T01:16:33.233Z", "content": "Can I use ImageNet Dataset?
\n", "rawMarkdown": "Can I use ImageNet Dataset?\n\nhttp://www.image-net.org/", "replies": [ { "id": 975776, "postDate": "2020-08-18T13:12:26.953Z", "rawMarkdown": "", "isDeleted": true } ] }, { "id": 1019599, "postDate": "2020-09-20T14:31:21.080Z", "rawMarkdown": "", "isDeleted": true }, { "id": 966934, "postDate": "2020-08-11T19:16:15.077Z", "rawMarkdown": "", "isDeleted": true } ], "comments": [ { "id": 1022625, "author_name": "Dieter", "author_url": "", "post_date": "2020-09-22T16:35:07.857000", "content": "https://github.com/junyanz/pytorch-CycleGAN-and-pix2pix
\nhttps://github.com/magicleap/SuperPointPretrainedNetwork
\nhttps://github.com/fidler-lab/defgrid-release
\nhttps://github.com/aritra0593/Reinforced-Feature-Points
https://github.com/rwightman/gen-efficientnet-pytorch
\nhttps://github.com/Cadene/pretrained-models.pytorch
\nhttps://github.com/clovaai/rexnet
https://github.com/tensorflow/models/tree/master/research/object_detection
\nhttps://tfhub.dev/s?module-type=image-object-detection
Can we use full version of GLDv2 dataset?
", "votes": 1, "replies": [ { "id": 978821, "author_name": "Chandan Verma", "author_url": "", "post_date": "2020-08-20T12:17:30.180000", "content": "It should be allowed.
", "votes": 0, "replies": [] } ] }, { "id": 1021591, "author_name": "NguyenThanhNhan", "author_url": "", "post_date": "2020-09-22T02:43:11.163000", "content": "ADE-20k pretrained models: https://github.com/CSAILVision/semantic-segmentation-pytorch
\nImage matching: https://github.com/rpautrat/SuperPoint
\nhttps://github.com/cvg/Hierarchical-Localization
Wow, exciting! I've been wondering since the beginning if someone's gonna use SuperPoint and/or SuperGlue in this competition 😯 Can't wait to read your team's write-up (if you're going to write one) 😊
", "votes": 0, "replies": [] } ] }, { "id": 982173, "author_name": "Stephen Liang", "author_url": "", "post_date": "2020-08-23T05:56:05.253000", "content": "can we use http://places2.csail.mit.edu/ if we open source our trained model. see the problem is it says non-commercial research, but if we open source the model, wouldn't that make it indirect?
", "votes": 2, "replies": [ { "id": 1001134, "author_name": "jho_miro", "author_url": "", "post_date": "2020-09-07T04:57:35.870000", "content": "have a look at this notebook by @rsmits
\nlooks like you can use this kaggle dataset
", "votes": 0, "replies": [] } ] }, { "id": 1030798, "author_name": "Sunghyun Jun", "author_url": "", "post_date": "2020-09-29T01:31:30.620000", "content": "https://github.com/tensorflow/models/tree/master/research/delf
\nhttps://www.kaggle.com/camaskew/delg-saved-models
\nhttps://www.kaggle.com/camaskew/baseline-landmark-retrieval-model
https://www.kaggle.com/camaskew/delg-saved-models
\nhttps://www.kaggle.com/keras/inceptionresnetv2
\nhttps://www.kaggle.com/keras/xception
https://www.kaggle.com/camaskew/baseline-landmark-retrieval-model
\nhttps://www.kaggle.com/camaskew/delg-saved-models
\nhttps://github.com/tensorflow/models/tree/master/research/delf
\nhttps://www.kaggle.com/c/landmark-retrieval-2020/data
https://github.com/lukemelas/EfficientNet-PyTorch
\nhttps://github.com/zylo117/Yet-Another-EfficientDet-Pytorch
https://github.com/pytorch/vision/tree/master/torchvision/models
\nhttps://github.com/qubvel/efficientnet
\nhttps://www.tensorflow.org/api_docs/python/tf/keras/applications
\nhttps://github.com/zhanghang1989/ResNeSt
\nhttps://github.com/magicleap/SuperGluePretrainedNetwork
\nhttps://www.kaggle.com/c/landmark-recognition-challenge
\nhttps://storage.googleapis.com/openimages/web/index.html
\nhttps://cocodataset.org/#home
https://github.com/PaddlePaddle/Research/tree/master/CV/landmark
", "votes": 0, "replies": [] }, { "id": 1020546, "author_name": "Alex", "author_url": "", "post_date": "2020-09-21T08:40:53.293000", "content": "Pretrained model for frontal face detection (open-cv): https://github.com/opencv/opencv/blob/master/data/haarcascades/haarcascade_frontalface_default.xml
", "votes": 0, "replies": [] }, { "id": 1019465, "author_name": "Mark Peng", "author_url": "", "post_date": "2020-09-20T13:09:25.060000", "content": "MAGSAC
\n", "votes": 0, "replies": [] }, { "id": 1018513, "author_name": "bestfitting", "author_url": "", "post_date": "2020-09-19T18:11:36.297000", "content": "https://www.kaggle.com/google/google-landmarks-dataset
\nhttps://github.com/tensorflow/models/tree/master/research/delf
\nhttps://github.com/eric-yyjau/pytorch-superpoint
Full GLv2 dataset: https://github.com/cvdfoundation/google-landmark
\nPre-trained models: https://github.com/rwightman/pytorch-image-models
\nConfiguration: https://github.com/rbgirshick/yacs
\nrapids.ai CUML: https://github.com/rapidsai/cuml
Wikipedia categories for GLR 2020
", "votes": 0, "replies": [] }, { "id": 989813, "author_name": "Der Informatiker", "author_url": "", "post_date": "2020-08-29T06:13:06.917000", "content": "EfficientNet-PyTorch:
\nhttps://github.com/lukemelas/EfficientNet-PyTorch
If I train a model on one notebook (NB-1) and create another script file (SF-1) and load that model. Is that considered as a pre-trained model? and do I need to post that?
\nI train a model, save, and commit. How do I retrieve this after a few commits or load it onto another script file?
\n@maggiemd
Hi, I have a question about pre learned models.
\nIs it allowed to learn models using the train data once and save the models, and then when submitting, simply load the saved models from inputs.
\nIn other words, training part is done outside of the submitted kernel, and only prediction will be done.
\nI would really appreciate if someone can help me out with this question!
Yes. You can train the model anywhere you want, be it kaggle kernels or you personal system. For submission you need to take in the test data as inputs as generate the submission file as output. Hope it helps.
", "votes": 1, "replies": [] } ] }, { "id": 974532, "author_name": "Chanran Kim", "author_url": "", "post_date": "2020-08-18T01:16:33.233000", "content": "Can I use ImageNet Dataset?
\n", "votes": 0, "replies": [ { "id": 975776, "author_name": "", "author_url": "", "post_date": "2020-08-18T13:12:26.953000", "content": "", "votes": 0, "replies": [] } ] }, { "id": 1019599, "author_name": "", "author_url": "", "post_date": "2020-09-20T14:31:21.080000", "content": "", "votes": 0, "replies": [] }, { "id": 966934, "author_name": "", "author_url": "", "post_date": "2020-08-11T19:16:15.077000", "content": "", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "956812": "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.", "1022625": "https://github.com/junyanz/pytorch-CycleGAN-and-pix2pix\nhttps://github.com/magicleap/SuperPointPretrainedNetwork\nhttps://github.com/fidler-lab/defgrid-release\nhttps://github.com/aritra0593/Reinforced-Feature-Points\n", "1022419": "https://github.com/rwightman/gen-efficientnet-pytorch\nhttps://github.com/Cadene/pretrained-models.pytorch\nhttps://github.com/clovaai/rexnet", "1022357": "[https://github.com/tensorflow/models/tree/master/research/object_detection](url)\n[https://tfhub.dev/s?module-type=image-object-detection](url)", "978815": "Can we use full version of GLDv2 dataset? ", "1021591": "ADE-20k pretrained models: https://github.com/CSAILVision/semantic-segmentation-pytorch\nImage matching: https://github.com/rpautrat/SuperPoint\nhttps://github.com/cvg/Hierarchical-Localization\n", "982173": "can we use http://places2.csail.mit.edu/ if we open source our trained model. see the problem is it says non-commercial research, but if we open source the model, wouldn't that make it indirect?", "1030798": "[https://github.com/tensorflow/models/tree/master/research/delf](url)\n[https://www.kaggle.com/camaskew/delg-saved-models](url)\n[https://www.kaggle.com/camaskew/baseline-landmark-retrieval-model](url)", "1030342": "https://www.kaggle.com/camaskew/delg-saved-models\nhttps://www.kaggle.com/keras/inceptionresnetv2\nhttps://www.kaggle.com/keras/xception", "1023044": "https://www.kaggle.com/camaskew/baseline-landmark-retrieval-model\nhttps://www.kaggle.com/camaskew/delg-saved-models\nhttps://github.com/tensorflow/models/tree/master/research/delf\nhttps://www.kaggle.com/c/landmark-retrieval-2020/data", "1022999": "https://github.com/lukemelas/EfficientNet-PyTorch\nhttps://github.com/zylo117/Yet-Another-EfficientDet-Pytorch", "1022242": "https://github.com/pytorch/vision/tree/master/torchvision/models\nhttps://github.com/qubvel/efficientnet\nhttps://www.tensorflow.org/api_docs/python/tf/keras/applications\nhttps://github.com/zhanghang1989/ResNeSt\nhttps://github.com/magicleap/SuperGluePretrainedNetwork\nhttps://www.kaggle.com/c/landmark-recognition-challenge\nhttps://storage.googleapis.com/openimages/web/index.html\nhttps://cocodataset.org/#home", "1021047": "https://github.com/PaddlePaddle/Research/tree/master/CV/landmark", "1020546": "Pretrained model for frontal face detection (open-cv): https://github.com/opencv/opencv/blob/master/data/haarcascades/haarcascade_frontalface_default.xml", "1019465": "**MAGSAC**\n- https://github.com/danini/magsac\n- https://github.com/ducha-aiki/pymagsac", "1018513": "https://www.kaggle.com/google/google-landmarks-dataset\nhttps://github.com/tensorflow/models/tree/master/research/delf\nhttps://github.com/eric-yyjau/pytorch-superpoint", "1004817": "Full GLv2 dataset: https://github.com/cvdfoundation/google-landmark\nPre-trained models: https://github.com/rwightman/pytorch-image-models\nConfiguration: https://github.com/rbgirshick/yacs\nrapids.ai CUML: https://github.com/rapidsai/cuml", "1000615": "[Wikipedia categories for GLR 2020](https://www.kaggle.com/steubk/wikipedia-categories-for-glr-2020) ", "989813": "EfficientNet-PyTorch:\nhttps://github.com/lukemelas/EfficientNet-PyTorch", "988364": "If I train a model on one notebook (NB-1) and create another script file (SF-1) and load that model. Is that considered as a pre-trained model? and do I need to post that?\nI train a model, save, and commit. How do I retrieve this after a few commits or load it onto another script file?\n@maggiemd ", "985143": "Hi, I have a question about pre learned models.\nIs it allowed to learn models using the train data once and save the models, and then when submitting, simply load the saved models from inputs.\nIn other words, training part is done outside of the submitted kernel, and only prediction will be done.\nI would really appreciate if someone can help me out with this question!\n", "974532": "Can I use ImageNet Dataset?\n\nhttp://www.image-net.org/", "1019599": "", "966934": "" } }