![\"whales\"](\"https://i.imgur.com/Ep0MvEw.jpg\")
In this competition, you’re challenged to build an algorithm to identify individual whales in images. You’ll analyze Happywhale's database of over 25,000 images, gathered from research institutions and public contributors. By contributing, you’ll help to open rich fields of understanding for marine mammal population dynamics around the globe.
\n\nA Playground version of this competition was hosted a few months ago. This competition contains even more images and individual whales than the last. While you can certainly use the images in the previous competition if you wish, submissions for the Playground competition are disabled for the duration of this competition. We recognize the chance of leakage from the Playground competition, since some images have been re-used, and have worked to minimize the possibility. The value that this competition brings to the scientific community, though, far outstrips the risks.
\n\nHave fun, and let the best model win!
", "messages": [ { "id": 430673, "postDate": "2018-11-30T20:38:13.860Z", "content": "In this competition, you’re challenged to build an algorithm to identify individual whales in images. You’ll analyze Happywhale's database of over 25,000 images, gathered from research institutions and public contributors. By contributing, you’ll help to open rich fields of understanding for marine mammal population dynamics around the globe.
\n\nA Playground version of this competition was hosted a few months ago. This competition contains even more images and individual whales than the last. While you can certainly use the images in the previous competition if you wish, submissions for the Playground competition are disabled for the duration of this competition. We recognize the chance of leakage from the Playground competition, since some images have been re-used, and have worked to minimize the possibility. The value that this competition brings to the scientific community, though, far outstrips the risks.
\n\nHave fun, and let the best model win!
", "rawMarkdown": "In this competition, you’re challenged to build an algorithm to identify individual whales in images. You’ll analyze Happywhale's database of over 25,000 images, gathered from research institutions and public contributors. By contributing, you’ll help to open rich fields of understanding for marine mammal population dynamics around the globe.\n\nA [Playground version of this competition](https://www.kaggle.com/c/whale-categorization-playground) was hosted a few months ago. This competition contains even more images and individual whales than the last. While you can certainly use the images in the previous competition if you wish, submissions for the Playground competition are disabled for the duration of this competition. We recognize the chance of leakage from the Playground competition, since some images have been re-used, and have worked to minimize the possibility. The value that this competition brings to the scientific community, though, far outstrips the risks.\n\nHave fun, and let the best model win!", "votes": 18 }, { "id": 430694, "postDate": "2018-11-30T21:10:55.317Z", "content": "Awesome! I missed a few whale challenges. I try to not skip this one! But first doodle recognition has to finish and I need to erase the doodles from my desktop :)
", "rawMarkdown": "Awesome! I missed a few whale challenges. I try to not skip this one! But first doodle recognition has to finish and I need to erase the doodles from my desktop :)", "votes": 7, "replies": [ { "id": 430705, "postDate": "2018-11-30T21:33:27.480Z", "content": "I'll allow it. :-)
", "rawMarkdown": "I'll allow it. :-)", "votes": 1 } ] }, { "id": 452557, "postDate": "2019-01-08T22:12:26.967Z", "content": "I'm getting into the competition a little late but I have to say is things like this that got me excited to dive into ML.
", "rawMarkdown": "I'm getting into the competition a little late but I have to say is things like this that got me excited to dive into ML. ", "votes": 4, "replies": [ { "id": 453221, "postDate": "2019-01-09T21:43:23.753Z", "content": "\n\n\ndive into ML.
\n
I sea what you did there.
", "rawMarkdown": "> dive into ML.\n\nI sea what you did there.", "votes": 3 } ] }, { "id": 444857, "postDate": "2018-12-25T01:50:36.540Z", "content": "Hey @inversion , It has been a great deal of fun so far!
", "rawMarkdown": "Hey @inversion , It has been a great deal of fun so far!", "votes": 1 }, { "id": 432998, "postDate": "2018-12-04T14:52:34.043Z", "content": "Hello. As competition description says we'll work with Happywhale’s database. But if so, is it legal to download all images from the database and just submit right labels of the whales? I gues not. But how you protect competition against this leak?
", "rawMarkdown": "Hello. As competition description says we'll work with Happywhale’s database. But if so, is it legal to download all images from the database and just submit right labels of the whales? I gues not. But how you protect competition against this leak?", "votes": 1, "replies": [ { "id": 434291, "postDate": "2018-12-06T06:55:51.410Z", "content": "Hey there,
\n\nThe labeled test images are not available from the Happywhale site, in addition - the rules prohibit hand-labeling of the dataset. Should it be determined any winner hand-labeled in order to win, they will be disqualified.
\n\nThanks!
", "rawMarkdown": "Hey there,\n\nThe labeled test images are not available from the Happywhale site, in addition - the rules prohibit hand-labeling of the dataset. Should it be determined any winner hand-labeled in order to win, they will be disqualified.\n\nThanks!", "votes": 5 }, { "id": 434334, "postDate": "2018-12-06T08:07:54.537Z", "content": "Great than! Thanks
", "rawMarkdown": "Great than! Thanks", "votes": 1 }, { "id": 455012, "postDate": "2019-01-12T18:42:38.010Z", "rawMarkdown": "", "isDeleted": true } ] }, { "id": 431854, "postDate": "2018-12-03T02:31:52.400Z", "content": "Remember when I was at the last playground competition, I was stuck at 0.49x score. This time is a good chance to play more :)
", "rawMarkdown": "Remember when I was at the last playground competition, I was stuck at 0.49x score. This time is a good chance to play more :)", "votes": 1 }, { "id": 430685, "postDate": "2018-11-30T20:55:14.430Z", "content": "So you want to be able to spot \"Bob\" the whale if he is in a photo; spot \"Sandy\" the whale if she is in a different photo. I haven't down loaded the data yet... is there more than one whale in a photo?
", "rawMarkdown": "So you want to be able to spot \"Bob\" the whale if he is in a photo; spot \"Sandy\" the whale if she is in a different photo. I haven't down loaded the data yet... is there more than one whale in a photo?", "votes": 1, "replies": [ { "id": 430689, "postDate": "2018-11-30T20:58:54.157Z", "content": "Its a single whale every photo, but there are many classes. Plus some whales have only one image to train and some have 34. Data is not representative.
", "rawMarkdown": "Its a single whale every photo, but there are many classes. Plus some whales have only one image to train and some have 34. Data is not representative. ", "votes": -1 }, { "id": 430690, "postDate": "2018-11-30T21:01:50.310Z", "content": "LOL . . . I used \"Bob\" when I was explaining the comp to a colleague. Yes, that's right. Almost all of the whales images are of a single whale fluke.
\n\nYou can also use the data explorer on the Data tab to check out the images before downloading. Click on the image folder (e.g., train.zip).
\n\n
So... there are many species, some species with only one picture. You also want to identify Bob whale from Jimmy the whale.... which implies that they have more than one picture someplace in the dataset, correct? And you want to determine what species they are. Does every \"pick a name for a whale\" have an image in the test set as well?
", "rawMarkdown": "So... there are many species, some species with only one picture. You also want to identify Bob whale from Jimmy the whale.... which implies that they have more than one picture someplace in the dataset, correct? And you want to determine what species they are. Does every \"pick a name for a whale\" have an image in the test set as well?" }, { "id": 430707, "postDate": "2018-11-30T21:39:11.433Z", "content": "@trog All the whales are Humpback's. You're predicting individuals. The data tab has more into about what is in Train/Test.
", "rawMarkdown": "@trog All the whales are Humpback's. You're predicting individuals. The data tab has more into about what is in Train/Test.", "votes": 2 }, { "id": 433871, "postDate": "2018-12-05T15:28:58.033Z", "content": "In the test images there is an example with two whales: 0ec7b958d.jpg.\nThere is also a few that are composite photos, fyi:\n3cae6cc29.jpg\n9aecef5dc.jpg\nc4e1fe45f.jpg\nfbff4e7b1.jpg
", "rawMarkdown": "In the test images there is an example with two whales: 0ec7b958d.jpg.\nThere is also a few that are composite photos, fyi:\n3cae6cc29.jpg\n9aecef5dc.jpg\nc4e1fe45f.jpg\nfbff4e7b1.jpg\n", "votes": 2 }, { "id": 435148, "postDate": "2018-12-07T15:31:26.883Z", "content": "Would composites of the same whale in one image affect either the precision or recall?
", "rawMarkdown": "Would composites of the same whale in one image affect either the precision or recall?" }, { "id": 465427, "postDate": "2019-02-03T04:41:15.113Z", "rawMarkdown": "" } ] }, { "id": 475659, "postDate": "2019-02-21T03:39:52.613Z", "content": "Hi! Why the # of competitors decrease from 2000+ to only 1414? Are there people being kicked out of the competition?@inversion
", "rawMarkdown": "Hi! Why the # of competitors decrease from 2000+ to only 1414? Are there people being kicked out of the competition?@inversion", "replies": [ { "id": 475673, "postDate": "2019-02-21T03:57:53.447Z", "content": "There is a leaderboard re-calc in progress. Will be back to normal soon!
", "rawMarkdown": "There is a leaderboard re-calc in progress. Will be back to normal soon!" } ] }, { "id": 455216, "postDate": "2019-01-13T09:38:23.043Z", "content": "@inversion, Why there is some data that is present on HappyWhale.com, but not in this competition (e.g. Vivaldi whale images)?
", "rawMarkdown": "@inversion, Why there is some data that is present on HappyWhale.com, but not in this competition (e.g. Vivaldi whale images)?", "replies": [ { "id": 455983, "postDate": "2019-01-14T22:24:37.503Z", "content": "There were a number of considerations for deciding which images to include in the contest. I don't have the particulars on this particular whale.
", "rawMarkdown": "There were a number of considerations for deciding which images to include in the contest. I don't have the particulars on this particular whale." } ] }, { "id": 454978, "postDate": "2019-01-12T17:00:47.430Z", "content": "Hi @inversion , could I know that is it legal to:\n- training a bounding box model and hand modify the bad predictions for training set?\n- training a bounding box model and hand modify the bad predictions for test set?\n- training a bounding box model and hand modify the bad predictions for test set, and open source the modified results before the merge deadline?
", "rawMarkdown": "Hi @inversion , could I know that is it legal to:\n- training a bounding box model and hand modify the bad predictions for training set?\n- training a bounding box model and hand modify the bad predictions for test set?\n- training a bounding box model and hand modify the bad predictions for test set, and open source the modified results before the merge deadline?", "replies": [ { "id": 455980, "postDate": "2019-01-14T22:21:51.580Z", "content": "You cannot make any hand modifications to Test predictions, which would include bounding boxes. The model should work on an unseen image without any manual intervention.
\n\nYou may, however, modify anything you want for the training set.
", "rawMarkdown": "You cannot make any hand modifications to Test predictions, which would include bounding boxes. The model should work on an unseen image without any manual intervention.\n\nYou may, however, modify anything you want for the training set." }, { "id": 456038, "postDate": "2019-01-15T02:39:54.157Z", "content": "Thanks inversion. I'll follow this rule.
", "rawMarkdown": "Thanks inversion. I'll follow this rule." } ] }, { "id": 454526, "postDate": "2019-01-11T19:30:40.227Z", "content": "The Data Description said that there are 3,000+ whale Ids. What I found in the training set are 5005 unique Ids. Please let me know if any wrong. Thank you!
", "rawMarkdown": "The Data Description said that there are 3,000+ whale Ids. What I found in the training set are 5005 unique Ids. Please let me know if any wrong. Thank you!\n", "replies": [ { "id": 455293, "postDate": "2019-01-13T13:53:40.223Z", "content": "Hi Zheng,
\n\nIt doesn't said there are 3,000+ whale Ids. The Data Description said there are a few examples for each of 3,000+ whale Ids. Your question is answered here by Lu Yang. https://www.kaggle.com/c/humpback-whale-identification/discussion/76062
", "rawMarkdown": "Hi Zheng,\n\nIt doesn't said there are 3,000+ whale Ids. The Data Description said there are **a few examples** for each of 3,000+ whale Ids. Your question is answered here by Lu Yang. https://www.kaggle.com/c/humpback-whale-identification/discussion/76062" }, { "id": 474593, "postDate": "2019-02-19T15:48:47.020Z", "content": "Got it! Thank you!
", "rawMarkdown": "Got it! Thank you!" } ] }, { "id": 443888, "postDate": "2018-12-22T17:04:24.797Z", "content": "I was wondering how many individual humpbacks live in the world. Apparently, about 80 000, not too many... https://en.wikipedia.org/wiki/Humpback_whale
", "rawMarkdown": "I was wondering how many individual humpbacks live in the world. Apparently, about 80 000, not too many... https://en.wikipedia.org/wiki/Humpback_whale ", "replies": [ { "id": 450805, "postDate": "2019-01-05T20:00:03.240Z", "content": "Wow.
", "rawMarkdown": "Wow." } ] }, { "id": 443218, "postDate": "2018-12-21T08:30:51.087Z", "content": "Please comment something about duplicate labels fro the same whales: https://www.kaggle.com/c/humpback-whale-identification/discussion/75352
", "rawMarkdown": "Please comment something about duplicate labels fro the same whales: https://www.kaggle.com/c/humpback-whale-identification/discussion/75352" }, { "id": 437376, "postDate": "2018-12-11T19:37:57.417Z", "content": "What an interesting competition ! Good luck to everyone !
", "rawMarkdown": "What an interesting competition ! Good luck to everyone ! \n" }, { "id": 431933, "postDate": "2018-12-03T05:41:15.710Z", "content": "Can I write Matlab code to implement the algorithm of Whale identification?
", "rawMarkdown": "Can I write Matlab code to implement the algorithm of Whale identification?", "replies": [ { "id": 431954, "postDate": "2018-12-03T06:47:12.933Z", "content": "Yes, you could do that since you would be submitting a solution set and not the code itself.
", "rawMarkdown": "Yes, you could do that since you would be submitting a solution set and not the code itself. " }, { "id": 432256, "postDate": "2018-12-03T16:15:09.263Z", "content": "Thanks for reply.
", "rawMarkdown": "Thanks for reply." } ] }, { "id": 430857, "postDate": "2018-12-01T04:08:30.983Z", "content": "The api links are giving me a 404 - Not Found on google colab . Anyone else facing this issue ?
", "rawMarkdown": "The api links are giving me a 404 - Not Found on google colab . Anyone else facing this issue ?", "replies": [ { "id": 431586, "postDate": "2018-12-02T14:52:15.410Z", "rawMarkdown": "", "isDeleted": true } ] }, { "id": 431081, "postDate": "2018-12-01T15:51:08.090Z", "rawMarkdown": "", "votes": 4, "isDeleted": true } ], "comments": [ { "id": 430694, "author_name": "beluga", "author_url": "", "post_date": "2018-11-30T21:10:55.317000", "content": "Awesome! I missed a few whale challenges. I try to not skip this one! But first doodle recognition has to finish and I need to erase the doodles from my desktop :)
", "votes": 7, "replies": [ { "id": 430705, "author_name": "inversion", "author_url": "", "post_date": "2018-11-30T21:33:27.480000", "content": "I'll allow it. :-)
", "votes": 1, "replies": [] } ] }, { "id": 452557, "author_name": "Nic Ollis", "author_url": "", "post_date": "2019-01-08T22:12:26.967000", "content": "I'm getting into the competition a little late but I have to say is things like this that got me excited to dive into ML.
", "votes": 4, "replies": [ { "id": 453221, "author_name": "Michael Tam", "author_url": "", "post_date": "2019-01-09T21:43:23.753000", "content": "\n\n\ndive into ML.
\n
I sea what you did there.
", "votes": 3, "replies": [] } ] }, { "id": 444857, "author_name": "Jhonatan da Silva", "author_url": "", "post_date": "2018-12-25T01:50:36.540000", "content": "Hey @inversion , It has been a great deal of fun so far!
", "votes": 1, "replies": [] }, { "id": 432998, "author_name": "pi-null-mezon", "author_url": "", "post_date": "2018-12-04T14:52:34.043000", "content": "Hello. As competition description says we'll work with Happywhale’s database. But if so, is it legal to download all images from the database and just submit right labels of the whales? I gues not. But how you protect competition against this leak?
", "votes": 1, "replies": [ { "id": 434291, "author_name": "Addison Howard", "author_url": "", "post_date": "2018-12-06T06:55:51.410000", "content": "Hey there,
\n\nThe labeled test images are not available from the Happywhale site, in addition - the rules prohibit hand-labeling of the dataset. Should it be determined any winner hand-labeled in order to win, they will be disqualified.
\n\nThanks!
", "votes": 5, "replies": [] }, { "id": 434334, "author_name": "pi-null-mezon", "author_url": "", "post_date": "2018-12-06T08:07:54.537000", "content": "Great than! Thanks
", "votes": 1, "replies": [] }, { "id": 455012, "author_name": "", "author_url": "", "post_date": "2019-01-12T18:42:38.010000", "content": "", "votes": 0, "replies": [] } ] }, { "id": 431854, "author_name": "cab", "author_url": "", "post_date": "2018-12-03T02:31:52.400000", "content": "Remember when I was at the last playground competition, I was stuck at 0.49x score. This time is a good chance to play more :)
", "votes": 1, "replies": [] }, { "id": 430685, "author_name": "trog", "author_url": "", "post_date": "2018-11-30T20:55:14.430000", "content": "So you want to be able to spot \"Bob\" the whale if he is in a photo; spot \"Sandy\" the whale if she is in a different photo. I haven't down loaded the data yet... is there more than one whale in a photo?
", "votes": 1, "replies": [ { "id": 430689, "author_name": "Piyush Mahatkar", "author_url": "", "post_date": "2018-11-30T20:58:54.157000", "content": "Its a single whale every photo, but there are many classes. Plus some whales have only one image to train and some have 34. Data is not representative.
", "votes": -1, "replies": [] }, { "id": 430690, "author_name": "inversion", "author_url": "", "post_date": "2018-11-30T21:01:50.310000", "content": "LOL . . . I used \"Bob\" when I was explaining the comp to a colleague. Yes, that's right. Almost all of the whales images are of a single whale fluke.
\n\nYou can also use the data explorer on the Data tab to check out the images before downloading. Click on the image folder (e.g., train.zip).
\n\n
So... there are many species, some species with only one picture. You also want to identify Bob whale from Jimmy the whale.... which implies that they have more than one picture someplace in the dataset, correct? And you want to determine what species they are. Does every \"pick a name for a whale\" have an image in the test set as well?
", "votes": 0, "replies": [] }, { "id": 430707, "author_name": "inversion", "author_url": "", "post_date": "2018-11-30T21:39:11.433000", "content": "@trog All the whales are Humpback's. You're predicting individuals. The data tab has more into about what is in Train/Test.
", "votes": 2, "replies": [] }, { "id": 433871, "author_name": "raifer", "author_url": "", "post_date": "2018-12-05T15:28:58.033000", "content": "In the test images there is an example with two whales: 0ec7b958d.jpg.\nThere is also a few that are composite photos, fyi:\n3cae6cc29.jpg\n9aecef5dc.jpg\nc4e1fe45f.jpg\nfbff4e7b1.jpg
", "votes": 2, "replies": [] }, { "id": 435148, "author_name": "Simeon Trieu", "author_url": "", "post_date": "2018-12-07T15:31:26.883000", "content": "Would composites of the same whale in one image affect either the precision or recall?
", "votes": 0, "replies": [] }, { "id": 465427, "author_name": "hengck23", "author_url": "", "post_date": "2019-02-03T04:41:15.113000", "content": "", "votes": 0, "replies": [] } ] }, { "id": 475659, "author_name": "Dilapsky Lee", "author_url": "", "post_date": "2019-02-21T03:39:52.613000", "content": "Hi! Why the # of competitors decrease from 2000+ to only 1414? Are there people being kicked out of the competition?@inversion
", "votes": 0, "replies": [ { "id": 475673, "author_name": "inversion", "author_url": "", "post_date": "2019-02-21T03:57:53.447000", "content": "There is a leaderboard re-calc in progress. Will be back to normal soon!
", "votes": 0, "replies": [] } ] }, { "id": 455216, "author_name": "Vladislav Shakhrai", "author_url": "", "post_date": "2019-01-13T09:38:23.043000", "content": "@inversion, Why there is some data that is present on HappyWhale.com, but not in this competition (e.g. Vivaldi whale images)?
", "votes": 0, "replies": [ { "id": 455983, "author_name": "inversion", "author_url": "", "post_date": "2019-01-14T22:24:37.503000", "content": "There were a number of considerations for deciding which images to include in the contest. I don't have the particulars on this particular whale.
", "votes": 0, "replies": [] } ] }, { "id": 454978, "author_name": "Yiheng Wang", "author_url": "", "post_date": "2019-01-12T17:00:47.430000", "content": "Hi @inversion , could I know that is it legal to:\n- training a bounding box model and hand modify the bad predictions for training set?\n- training a bounding box model and hand modify the bad predictions for test set?\n- training a bounding box model and hand modify the bad predictions for test set, and open source the modified results before the merge deadline?
", "votes": 0, "replies": [ { "id": 455980, "author_name": "inversion", "author_url": "", "post_date": "2019-01-14T22:21:51.580000", "content": "You cannot make any hand modifications to Test predictions, which would include bounding boxes. The model should work on an unseen image without any manual intervention.
\n\nYou may, however, modify anything you want for the training set.
", "votes": 0, "replies": [] }, { "id": 456038, "author_name": "Yiheng Wang", "author_url": "", "post_date": "2019-01-15T02:39:54.157000", "content": "Thanks inversion. I'll follow this rule.
", "votes": 0, "replies": [] } ] }, { "id": 454526, "author_name": "Erica Zheng", "author_url": "", "post_date": "2019-01-11T19:30:40.227000", "content": "The Data Description said that there are 3,000+ whale Ids. What I found in the training set are 5005 unique Ids. Please let me know if any wrong. Thank you!
", "votes": 0, "replies": [ { "id": 455293, "author_name": "toshi_k", "author_url": "", "post_date": "2019-01-13T13:53:40.223000", "content": "Hi Zheng,
\n\nIt doesn't said there are 3,000+ whale Ids. The Data Description said there are a few examples for each of 3,000+ whale Ids. Your question is answered here by Lu Yang. https://www.kaggle.com/c/humpback-whale-identification/discussion/76062
", "votes": 0, "replies": [] }, { "id": 474593, "author_name": "Erica Zheng", "author_url": "", "post_date": "2019-02-19T15:48:47.020000", "content": "Got it! Thank you!
", "votes": 0, "replies": [] } ] }, { "id": 443888, "author_name": "Blonde", "author_url": "", "post_date": "2018-12-22T17:04:24.797000", "content": "I was wondering how many individual humpbacks live in the world. Apparently, about 80 000, not too many... https://en.wikipedia.org/wiki/Humpback_whale
", "votes": 0, "replies": [ { "id": 450805, "author_name": "johnny", "author_url": "", "post_date": "2019-01-05T20:00:03.240000", "content": "Wow.
", "votes": 0, "replies": [] } ] }, { "id": 443218, "author_name": "pi-null-mezon", "author_url": "", "post_date": "2018-12-21T08:30:51.087000", "content": "Please comment something about duplicate labels fro the same whales: https://www.kaggle.com/c/humpback-whale-identification/discussion/75352
", "votes": 0, "replies": [] }, { "id": 437376, "author_name": "george_tsk", "author_url": "", "post_date": "2018-12-11T19:37:57.417000", "content": "What an interesting competition ! Good luck to everyone !
", "votes": 0, "replies": [] }, { "id": 431933, "author_name": "ASIT BARMAN", "author_url": "", "post_date": "2018-12-03T05:41:15.710000", "content": "Can I write Matlab code to implement the algorithm of Whale identification?
", "votes": 0, "replies": [ { "id": 431954, "author_name": "Anirudha Sundaresan", "author_url": "", "post_date": "2018-12-03T06:47:12.933000", "content": "Yes, you could do that since you would be submitting a solution set and not the code itself.
", "votes": 0, "replies": [] }, { "id": 432256, "author_name": "ASIT BARMAN", "author_url": "", "post_date": "2018-12-03T16:15:09.263000", "content": "Thanks for reply.
", "votes": 0, "replies": [] } ] }, { "id": 430857, "author_name": "sumit", "author_url": "", "post_date": "2018-12-01T04:08:30.983000", "content": "The api links are giving me a 404 - Not Found on google colab . Anyone else facing this issue ?
", "votes": 0, "replies": [ { "id": 431586, "author_name": "", "author_url": "", "post_date": "2018-12-02T14:52:15.410000", "content": "", "votes": 0, "replies": [] } ] }, { "id": 431081, "author_name": "", "author_url": "", "post_date": "2018-12-01T15:51:08.090000", "content": "", "votes": 4, "replies": [] } ], "raw_markdown_by_id": { "430673": "In this competition, you’re challenged to build an algorithm to identify individual whales in images. You’ll analyze Happywhale's database of over 25,000 images, gathered from research institutions and public contributors. By contributing, you’ll help to open rich fields of understanding for marine mammal population dynamics around the globe.\n\nA [Playground version of this competition](https://www.kaggle.com/c/whale-categorization-playground) was hosted a few months ago. This competition contains even more images and individual whales than the last. While you can certainly use the images in the previous competition if you wish, submissions for the Playground competition are disabled for the duration of this competition. We recognize the chance of leakage from the Playground competition, since some images have been re-used, and have worked to minimize the possibility. The value that this competition brings to the scientific community, though, far outstrips the risks.\n\nHave fun, and let the best model win!", "430694": "Awesome! I missed a few whale challenges. I try to not skip this one! But first doodle recognition has to finish and I need to erase the doodles from my desktop :)", "452557": "I'm getting into the competition a little late but I have to say is things like this that got me excited to dive into ML. ", "444857": "Hey @inversion , It has been a great deal of fun so far!", "432998": "Hello. As competition description says we'll work with Happywhale’s database. But if so, is it legal to download all images from the database and just submit right labels of the whales? I gues not. But how you protect competition against this leak?", "431854": "Remember when I was at the last playground competition, I was stuck at 0.49x score. This time is a good chance to play more :)", "430685": "So you want to be able to spot \"Bob\" the whale if he is in a photo; spot \"Sandy\" the whale if she is in a different photo. I haven't down loaded the data yet... is there more than one whale in a photo?", "475659": "Hi! Why the # of competitors decrease from 2000+ to only 1414? Are there people being kicked out of the competition?@inversion", "455216": "@inversion, Why there is some data that is present on HappyWhale.com, but not in this competition (e.g. Vivaldi whale images)?", "454978": "Hi @inversion , could I know that is it legal to:\n- training a bounding box model and hand modify the bad predictions for training set?\n- training a bounding box model and hand modify the bad predictions for test set?\n- training a bounding box model and hand modify the bad predictions for test set, and open source the modified results before the merge deadline?", "454526": "The Data Description said that there are 3,000+ whale Ids. What I found in the training set are 5005 unique Ids. Please let me know if any wrong. Thank you!\n", "443888": "I was wondering how many individual humpbacks live in the world. Apparently, about 80 000, not too many... https://en.wikipedia.org/wiki/Humpback_whale ", "443218": "Please comment something about duplicate labels fro the same whales: https://www.kaggle.com/c/humpback-whale-identification/discussion/75352", "437376": "What an interesting competition ! Good luck to everyone ! \n", "431933": "Can I write Matlab code to implement the algorithm of Whale identification?", "430857": "The api links are giving me a 404 - Not Found on google colab . Anyone else facing this issue ?", "431081": "" } }