{ "id": 307752, "title": "Resources provided by hosts", "url": "/competitions/birdclef-2022/discussion/307752", "author_name": "Stefan Kahl", "post_date": "2022-02-15T13:52:18.039000", "votes": 43, "comment_count": 30, "views": 0, "content": "

Thanks for participating in this year’s BirdCLEF competition. Just like previous editions, this year’s challenge focuses on the science of recognizing birds by sound. There has been significant progress in this area of research in recent years, but we are not there yet. This is especially true for rare and endangered bird species. In many cases, only very few training recordings exists for those birds, yet, they're significantly more important to monitor due to their elevated conservation status.

\n

Species portraits

\n

In this competition, we focus on birds that are endemic to the chain of islands of Hawaii. We collected soundscape recordings and expert labels for 10 Hawaiian bird species, and our host Amanda prepared a notebook that introduces each of the birds, so you get a better understanding of what we're looking for.

\n

Sample submissions

\n

As in previous editions, this year's test data will be hidden and submitting a solution to this competition can be tricky. That's why we prepared another notebook that briefly introduces the submission process.

\n

Other resources

\n

Please make sure to check out previous editions either here on Kaggle (2020, 2021) or by browsing the archive of BirdCLEF working notes and overview papers.

", "messages": [ { "id": 1691604, "postDate": "2022-02-15T13:52:18.040Z", "content": "

Thanks for participating in this year’s BirdCLEF competition. Just like previous editions, this year’s challenge focuses on the science of recognizing birds by sound. There has been significant progress in this area of research in recent years, but we are not there yet. This is especially true for rare and endangered bird species. In many cases, only very few training recordings exists for those birds, yet, they're significantly more important to monitor due to their elevated conservation status.

\n

Species portraits

\n

In this competition, we focus on birds that are endemic to the chain of islands of Hawaii. We collected soundscape recordings and expert labels for 10 Hawaiian bird species, and our host Amanda prepared a notebook that introduces each of the birds, so you get a better understanding of what we're looking for.

\n

Sample submissions

\n

As in previous editions, this year's test data will be hidden and submitting a solution to this competition can be tricky. That's why we prepared another notebook that briefly introduces the submission process.

\n

Other resources

\n

Please make sure to check out previous editions either here on Kaggle (2020, 2021) or by browsing the archive of BirdCLEF working notes and overview papers.

", "rawMarkdown": "Thanks for participating in this year’s BirdCLEF competition. Just like previous editions, this year’s challenge focuses on the science of recognizing birds by sound. There has been significant progress in this area of research in recent years, but we are not there yet. This is especially true for rare and endangered bird species. In many cases, only very few training recordings exists for those birds, yet, they're significantly more important to monitor due to their elevated conservation status.\n\n**Species portraits**\n\nIn this competition, we focus on birds that are endemic to the chain of islands of Hawaii. We collected soundscape recordings and expert labels for 10 Hawaiian bird species, and our host [Amanda](https://www.kaggle.com/amandanavine) prepared a [notebook that introduces each of the birds](https://www.kaggle.com/amandanavine/hawaiian-bird-species), so you get a better understanding of what we're looking for.\n\n**Sample submissions**\n\nAs in previous editions, this year's test data will be hidden and submitting a solution to this competition can be tricky. That's why we prepared another [notebook that briefly introduces the submission process](https://www.kaggle.com/stefankahl/how-to-submit-to-birdclef-2022).\n\n**Other resources**\n\nPlease make sure to check out previous editions either here on Kaggle ([2020](https://www.kaggle.com/c/birdsong-recognition), [2021](https://www.kaggle.com/c/birdclef-2021/overview/description)) or by browsing the archive of BirdCLEF [working notes and overview papers](https://scholar.google.com/scholar?q=birdclef).", "votes": 42 }, { "id": 1777943, "postDate": "2022-05-05T00:00:45.037Z", "content": "

Hi @stefankahl ,
\nmy recent submission encountered Notebook Threw Exception without any details. However, the notebook runs smoothly on the publicly available test dataset. Could you please help me check what is going on?

", "rawMarkdown": "Hi @stefankahl ,\nmy recent submission encountered Notebook Threw Exception without any details. However, the notebook runs smoothly on the publicly available test dataset. Could you please help me check what is going on?", "votes": 1 }, { "id": 1777381, "postDate": "2022-05-04T15:08:00.290Z", "content": "

@stefankahl @tomdenton @amandanavine

\n

Thank you for organizing this interesting competition.
\nI have a question about the evaluation metric.

\n

Is species weighting, as well as positive and negative weighting, calculated on the entire test data?
\nOr is it calculated independently on public and private data?

\n

Thanks

", "rawMarkdown": "@stefankahl @tomdenton @amandanavine \n\nThank you for organizing this interesting competition.\nI have a question about the evaluation metric.\n\nIs species weighting, as well as positive and negative weighting, calculated on the entire test data?\nOr is it calculated independently on public and private data?\n\nThanks", "votes": 1, "replies": [ { "id": 1777526, "postDate": "2022-05-04T16:31:35.123Z", "content": "

We generally don't disclose information related to the public vs private datasets while the competition is running.

", "rawMarkdown": "We generally don't disclose information related to the public vs private datasets while the competition is running.", "votes": 1 } ] }, { "id": 1692315, "postDate": "2022-02-16T00:58:59.803Z", "content": "

Thank you for organizing again a speech recognition competition!
\nJust a question about the data, can we use the data from previous competition ?

", "rawMarkdown": "Thank you for organizing again a speech recognition competition!\nJust a question about the data, can we use the data from previous competition ?", "votes": 1, "replies": [ { "id": 1692427, "postDate": "2022-02-16T03:15:57.517Z", "content": "

Hello Shiro! You may use data other than the provided competition data (external data) to develop and test your submissions. However, you must ensure the external data is publicly available and equally accessible to use by all participants of the competition (See https://www.kaggle.com/c/birdclef-2022/rules for more details).

", "rawMarkdown": "Hello Shiro! You may use data other than the provided competition data (external data) to develop and test your submissions. However, you must ensure the external data is publicly available and equally accessible to use by all participants of the competition (See https://www.kaggle.com/c/birdclef-2022/rules for more details).", "votes": 5 }, { "id": 1693779, "postDate": "2022-02-17T00:11:13.233Z", "content": "

Thank for the answer @amandanavine
\nI have a quick question then, have you seen some overlap with previous competitions ? I suspect some of them, some sample have the same name in some other competition with same classes. Just the format of the file is different ogg vs mp3.

", "rawMarkdown": "Thank for the answer @amandanavine \nI have a quick question then, have you seen some overlap with previous competitions ? I suspect some of them, some sample have the same name in some other competition with same classes. Just the format of the file is different ogg vs mp3. ", "votes": 1 }, { "id": 1693817, "postDate": "2022-02-17T01:08:16.203Z", "content": "

The bulk of the training data consists of recordings from xenocanto.org, which have been used in previous competitions.

", "rawMarkdown": " The bulk of the training data consists of recordings from xenocanto.org, which have been used in previous competitions. ", "votes": 1 }, { "id": 1694737, "postDate": "2022-02-17T17:17:17.503Z", "content": "

Hi, Shiro;
\nThe species which overlap with the previous competition(s) will have repeated data. The Xeno-Canto identifiers are unique, so xc123456.ogg and xc123456.mp3 will contain the same audio with a different encoding.

\n

(We have re-encoded the audio to 32khz to reduce the overall dataset filesize. Thanks to the Nyquist theorem a 32khz sample rate is plenty to capture just about any bird vocalization, which top out around the 13kHz frequency range… And even for the really high pitched birds like the golden crowned kinglet and brown creeper, most of the vocal activity is still under 11kHz.)

", "rawMarkdown": "Hi, Shiro;\nThe species which overlap with the previous competition(s) will have repeated data. The Xeno-Canto identifiers are unique, so xc123456.ogg and xc123456.mp3 will contain the same audio with a different encoding. \n\n(We have re-encoded the audio to 32khz to reduce the overall dataset filesize. Thanks to the Nyquist theorem a 32khz sample rate is plenty to capture just about any bird vocalization, which top out around the 13kHz frequency range... And even for the really high pitched birds like the golden crowned kinglet and brown creeper, most of the vocal activity is still under 11kHz.)", "votes": 3 }, { "id": 1698637, "postDate": "2022-02-20T14:52:28.897Z", "content": "

@amandanavine @tomdenton Thank you for your reply. I checked few samples and they seems the same. However, I saw that the primary/secondary labels have different typologie name between the Cornell Competition and Birdclef One. I think it is a more \"scientific name\" such as Limnodromus scolopaceus_Long-billed Dowitcher . I suppose you have a mapping name which translate \"Limnodromus scolopaceus_Long-billed Dowitcher\" into \"lobdow\"? I would like to convert the primary/secondary labels from cornell competition into the Birdclef format. Is it possible to get the same mapping ?

", "rawMarkdown": "@amandanavine @tomdenton Thank you for your reply. I checked few samples and they seems the same. However, I saw that the primary/secondary labels have different typologie name between the Cornell Competition and Birdclef One. I think it is a more \"scientific name\" such as Limnodromus scolopaceus_Long-billed Dowitcher . I suppose you have a mapping name which translate \"Limnodromus scolopaceus_Long-billed Dowitcher\" into \"lobdow\"? I would like to convert the primary/secondary labels from cornell competition into the Birdclef format. Is it possible to get the same mapping ?\n" }, { "id": 1702838, "postDate": "2022-02-24T01:36:04.550Z", "content": "

Just make the list of missing code name, is there a way to get their code name? :

\n
{'amazilia beryllina',\n  'amazilia cyanocephala',\n  'amazilia violiceps',\n  'amphispiza quinquestriata',\n  'atthis heloisa',\n  'campylopterus curvipennis',\n  'caracara cheriway',\n  'chlorostilbon ricordii',\n  'corvus caurinus',\n  'euphonia elegantissima',\n  'hylocharis leucotis',\n  'hylocharis xantusii',\n  'phalacrocorax brasilianus',\n  'sylvia cantillans',\n  'sylvia communis',\n  'sylvia curruca',\n  'sylvia melanocephala',\n  'sylvia mystacea',\n  'tetrao tetrix'}\n
", "rawMarkdown": "Just make the list of missing code name, is there a way to get their code name? : \n\n```\n{'amazilia beryllina',\n 'amazilia cyanocephala',\n 'amazilia violiceps',\n 'amphispiza quinquestriata',\n 'atthis heloisa',\n 'campylopterus curvipennis',\n 'caracara cheriway',\n 'chlorostilbon ricordii',\n 'corvus caurinus',\n 'euphonia elegantissima',\n 'hylocharis leucotis',\n 'hylocharis xantusii',\n 'phalacrocorax brasilianus',\n 'sylvia cantillans',\n 'sylvia communis',\n 'sylvia curruca',\n 'sylvia melanocephala',\n 'sylvia mystacea',\n 'tetrao tetrix'}\n```" }, { "id": 1704657, "postDate": "2022-02-25T17:34:03.793Z", "content": "

Here's what I've got for those species:

\n
berhum,     amazilia beryllina\nazchum1,     amazilia cyanocephala\nvichum,     amazilia violiceps\nfisspa,     amphispiza quinquestriata\nbumhum,     atthis heloisa\nwetsab1,     campylopterus curvipennis\ncrecar1,     caracara cheriway\ncubeme1,     chlorostilbon ricordii\nnorcro,     corvus caurinus\neleeup1,     euphonia elegantissima\nwhehum,     hylocharis leucotis\nxanhum,     hylocharis xantusii\nneocor,     phalacrocorax brasilianus\nsubwar1,     sylvia cantillans\ngrewhi1,     sylvia communis\nleswhi1,     sylvia curruca\nsarwar1,     sylvia melanocephala\nmenwar1,     sylvia mystacea\nblagro1,     tetrao tetrix\n
\n

These are the e-bird codes, so you can generally get the code by googling the species, finding its page at ebird.org and looking at the code in the address bar. Like this one:
\nhttps://ebird.org/species/azchum1

", "rawMarkdown": "Here's what I've got for those species:\n```\nberhum, \tamazilia beryllina\nazchum1, \tamazilia cyanocephala\nvichum, \tamazilia violiceps\nfisspa, \tamphispiza quinquestriata\nbumhum, \tatthis heloisa\nwetsab1, \tcampylopterus curvipennis\ncrecar1, \tcaracara cheriway\ncubeme1, \tchlorostilbon ricordii\nnorcro, \tcorvus caurinus\neleeup1, \teuphonia elegantissima\nwhehum, \thylocharis leucotis\nxanhum, \thylocharis xantusii\nneocor, \tphalacrocorax brasilianus\nsubwar1, \tsylvia cantillans\ngrewhi1, \tsylvia communis\nleswhi1, \tsylvia curruca\nsarwar1, \tsylvia melanocephala\nmenwar1, \tsylvia mystacea\nblagro1, \ttetrao tetrix\n```\nThese are the e-bird codes, so you can generally get the code by googling the species, finding its page at ebird.org and looking at the code in the address bar. Like this one:\nhttps://ebird.org/species/azchum1", "votes": 3 } ] }, { "id": 1697085, "postDate": "2022-02-19T11:07:51.620Z", "content": "

Is scraping the https://xeno-canto.org/ website allowed ?

", "rawMarkdown": "Is scraping the https://xeno-canto.org/ website allowed ?", "votes": -1, "replies": [ { "id": 1697089, "postDate": "2022-02-19T11:12:04.053Z", "content": "

@stefankahl had actually posted this last year.

\n

I think the website might have been updated since then, from the same thread, GM Rohan Rao had posted:

\n
\n

For those interested, the complete extended datasets (23,620 additional recordings) are publicly available for all 264 species (split in two by first alphabet due to Kaggle's 20GB limitation):

\n
\n

https://www.kaggle.com/rohanrao/xeno-canto-bird-recordings-extended-a-m
\nhttps://www.kaggle.com/rohanrao/xeno-canto-bird-recordings-extended-n-z

", "rawMarkdown": "@stefankahl had actually posted [this](https://www.kaggle.com/c/birdsong-recognition/discussion/160293) last year. \n\nI think the website might have been updated since then, from [the same thread](https://www.kaggle.com/c/birdsong-recognition/discussion/160293#894803), GM Rohan Rao had posted:\n\n> For those interested, the complete extended datasets (23,620 additional recordings) are publicly available for all 264 species (split in two by first alphabet due to Kaggle's 20GB limitation): \n\nhttps://www.kaggle.com/rohanrao/xeno-canto-bird-recordings-extended-a-m \nhttps://www.kaggle.com/rohanrao/xeno-canto-bird-recordings-extended-n-z", "votes": 6 } ] }, { "id": 1774236, "postDate": "2022-05-02T01:56:35.187Z", "content": "

Hi @tomdenton
\nyou display the leaderboard to the third decimal place?
\nLooking at the crowded leaderboards, it seems that the display up to the second decimal place is not enough. Consider Thank you.

", "rawMarkdown": "Hi @tomdenton \nyou display the leaderboard to the third decimal place?\nLooking at the crowded leaderboards, it seems that the display up to the second decimal place is not enough. Consider Thank you." }, { "id": 1733514, "postDate": "2022-03-24T11:02:36.247Z", "content": "
\n

In this competition, we focus on birds that are endemic to the chain of islands of Hawaii. We collected soundscape recordings and expert labels for 10 Hawaiian bird species,

\n
\n

So it`s clear that the test_soundscape recordings are all recorded on islands of Hawaii, but how about the train_audio recordings, were all of them recorded on islands of Hawaii as well?

", "rawMarkdown": "> In this competition, we focus on birds that are endemic to the chain of islands of Hawaii. We collected soundscape recordings and expert labels for 10 Hawaiian bird species,\n\nSo it`s clear that the test_soundscape recordings are all recorded on islands of Hawaii, but how about the train_audio recordings, were all of them recorded on islands of Hawaii as well?", "replies": [ { "id": 1733616, "postDate": "2022-03-24T12:52:46.513Z", "content": "

Since training data are downloaded from xeno-canto, it's not generally true, but I think almost all of the endemic species are only recorded in Hawaii.

\n

You can check by yourself by doing EDA of longitude and latitude of the recordings like myself[1].

\n", "rawMarkdown": "Since training data are downloaded from xeno-canto, it's not generally true, but I think almost all of the endemic species are only recorded in Hawaii.\n\nYou can check by yourself by doing EDA of longitude and latitude of the recordings like myself[1].\n\n* [1] EDA notebook about geo-distribution: https://www.kaggle.com/code/tatamikenn/birdclef-22-eda-on-the-scored-species?scriptVersionId=91133883#Geo-distribution", "votes": 1 } ] }, { "id": 1730092, "postDate": "2022-03-20T23:09:29.977Z", "content": "

I have a question.

\n
\n

We collected soundscape recordings and expert labels for 10 Hawaiian bird species

\n
\n

Why only 10 species? The number of scored species are 21. Is that mean that the rest of 11 species are processed in a different way?

", "rawMarkdown": "I have a question.\n\n> We collected soundscape recordings and expert labels for 10 Hawaiian bird species\n\nWhy only 10 species? The number of scored species are 21. Is that mean that the rest of 11 species are processed in a different way?", "replies": [ { "id": 1730837, "postDate": "2022-03-21T16:43:25.537Z", "content": "

If I recall correctly, the other 11 were recorded incidentally during attempts to get audio for the 10 that are of primary interest.

", "rawMarkdown": "If I recall correctly, the other 11 were recorded incidentally during attempts to get audio for the 10 that are of primary interest. ", "votes": 2 }, { "id": 1733790, "postDate": "2022-03-24T16:13:36.673Z", "content": "

By the way, as far as the host can answer, do you differentiate the scores between the 10 primary concerned (endangered) species and the other 11 non-endangered species? For example, the 10 primary concerned species are weighted twice as much as the rest.

", "rawMarkdown": "By the way, as far as the host can answer, do you differentiate the scores between the 10 primary concerned (endangered) species and the other 11 non-endangered species? For example, the 10 primary concerned species are weighted twice as much as the rest." }, { "id": 1733864, "postDate": "2022-03-24T17:08:05.350Z", "content": "

Each species was assigned an equal weight.

", "rawMarkdown": "Each species was assigned an equal weight.", "votes": 1 } ] }, { "id": 1702410, "postDate": "2022-02-23T15:28:11.990Z", "content": "

Hello! We are a team that enters a competition for the first time, we have a question regarding the limitation of 9 hours, are they hours of training?

", "rawMarkdown": "Hello! We are a team that enters a competition for the first time, we have a question regarding the limitation of 9 hours, are they hours of training?", "replies": [ { "id": 1702557, "postDate": "2022-02-23T17:38:20.880Z", "content": "

hours of inference

", "rawMarkdown": "hours of inference", "votes": 1 } ] }, { "id": 1774159, "postDate": "2022-05-01T22:55:07.963Z", "content": "

Hi all, i am getting 'Notebook Exceeded Allowed Compute'. As far as I know I am doing prediction one at a time and not saving any of the files I am generating to make the prediction. This is my submission nb. Can anyone help me understand what is going on? thanks

\n

https://www.kaggle.com/code/rumman18/submission-notebook

", "rawMarkdown": "Hi all, i am getting 'Notebook Exceeded Allowed Compute'. As far as I know I am doing prediction one at a time and not saving any of the files I am generating to make the prediction. This is my submission nb. Can anyone help me understand what is going on? thanks\n\nhttps://www.kaggle.com/code/rumman18/submission-notebook", "isDeleted": true }, { "id": 1772195, "postDate": "2022-04-30T01:03:56.177Z", "content": "

Hi all, i have a question about pre trained models. Since internet will be disabled and some of keras pre trained models require internet access to get downloaded - i was wondering how we could go about that? thanks

", "rawMarkdown": "Hi all, i have a question about pre trained models. Since internet will be disabled and some of keras pre trained models require internet access to get downloaded - i was wondering how we could go about that? thanks", "votes": 1, "isDeleted": true, "replies": [ { "id": 1772200, "postDate": "2022-04-30T01:15:28.503Z", "content": "

Great question!

\n

Most Kagglers upload them as a dataset and use them in their notebooks-you can checkout popular notebooks from older competitions with similar requirements.

", "rawMarkdown": "Great question! \n\nMost Kagglers upload them as a dataset and use them in their notebooks-you can checkout popular notebooks from older competitions with similar requirements. ", "votes": 1 } ] }, { "id": 1734340, "postDate": "2022-03-25T08:16:51.820Z", "rawMarkdown": "", "isDeleted": true }, { "id": 1727430, "postDate": "2022-03-18T01:11:52.543Z", "content": "

Hi, I have a quick question. What does the scored_birds.json actually mean here. There are ratings for all the birds in training metadata. How is scored different from that rating and why are there only few of the scored birds? thanks!

", "rawMarkdown": "Hi, I have a quick question. What does the scored_birds.json actually mean here. There are ratings for all the birds in training metadata. How is scored different from that rating and why are there only few of the scored birds? thanks!", "isDeleted": true, "replies": [ { "id": 1729629, "postDate": "2022-03-20T09:58:22.163Z", "content": "

Hey @rumman18, these are the Hawaiian species that should be in the submission file. Check the second cell of markdown/code in this public notebook for a more detailed description:
\nhttps://www.kaggle.com/code/stefankahl/how-to-submit-to-birdclef-2022/notebook

", "rawMarkdown": "Hey @rumman18, these are the Hawaiian species that should be in the submission file. Check the second cell of markdown/code in this public notebook for a more detailed description:\nhttps://www.kaggle.com/code/stefankahl/how-to-submit-to-birdclef-2022/notebook" } ] }, { "id": 1709595, "postDate": "2022-03-02T10:54:59.080Z", "rawMarkdown": "", "isDeleted": true }, { "id": 1774144, "postDate": "2022-05-01T21:52:49.863Z", "content": "

Thanks, everything is very clear now

", "rawMarkdown": "Thanks, everything is very clear now" } ], "comments": [ { "id": 1777943, "author_name": "Zhongkai Shangguan", "author_url": "", "post_date": "2022-05-05T00:00:45.037000", "content": "

Hi @stefankahl ,
\nmy recent submission encountered Notebook Threw Exception without any details. However, the notebook runs smoothly on the publicly available test dataset. Could you please help me check what is going on?

", "votes": 1, "replies": [] }, { "id": 1777381, "author_name": "atfujita", "author_url": "", "post_date": "2022-05-04T15:08:00.290000", "content": "

@stefankahl @tomdenton @amandanavine

\n

Thank you for organizing this interesting competition.
\nI have a question about the evaluation metric.

\n

Is species weighting, as well as positive and negative weighting, calculated on the entire test data?
\nOr is it calculated independently on public and private data?

\n

Thanks

", "votes": 1, "replies": [ { "id": 1777526, "author_name": "Sohier Dane", "author_url": "", "post_date": "2022-05-04T16:31:35.123000", "content": "

We generally don't disclose information related to the public vs private datasets while the competition is running.

", "votes": 1, "replies": [] } ] }, { "id": 1692315, "author_name": "Shiro", "author_url": "", "post_date": "2022-02-16T00:58:59.803000", "content": "

Thank you for organizing again a speech recognition competition!
\nJust a question about the data, can we use the data from previous competition ?

", "votes": 1, "replies": [ { "id": 1692427, "author_name": "Amanda Navine", "author_url": "", "post_date": "2022-02-16T03:15:57.517000", "content": "

Hello Shiro! You may use data other than the provided competition data (external data) to develop and test your submissions. However, you must ensure the external data is publicly available and equally accessible to use by all participants of the competition (See https://www.kaggle.com/c/birdclef-2022/rules for more details).

", "votes": 5, "replies": [] }, { "id": 1693779, "author_name": "Shiro", "author_url": "", "post_date": "2022-02-17T00:11:13.233000", "content": "

Thank for the answer @amandanavine
\nI have a quick question then, have you seen some overlap with previous competitions ? I suspect some of them, some sample have the same name in some other competition with same classes. Just the format of the file is different ogg vs mp3.

", "votes": 1, "replies": [] }, { "id": 1693817, "author_name": "Amanda Navine", "author_url": "", "post_date": "2022-02-17T01:08:16.203000", "content": "

The bulk of the training data consists of recordings from xenocanto.org, which have been used in previous competitions.

", "votes": 1, "replies": [] }, { "id": 1694737, "author_name": "Tom Denton", "author_url": "", "post_date": "2022-02-17T17:17:17.503000", "content": "

Hi, Shiro;
\nThe species which overlap with the previous competition(s) will have repeated data. The Xeno-Canto identifiers are unique, so xc123456.ogg and xc123456.mp3 will contain the same audio with a different encoding.

\n

(We have re-encoded the audio to 32khz to reduce the overall dataset filesize. Thanks to the Nyquist theorem a 32khz sample rate is plenty to capture just about any bird vocalization, which top out around the 13kHz frequency range… And even for the really high pitched birds like the golden crowned kinglet and brown creeper, most of the vocal activity is still under 11kHz.)

", "votes": 3, "replies": [] }, { "id": 1698637, "author_name": "Shiro", "author_url": "", "post_date": "2022-02-20T14:52:28.897000", "content": "

@amandanavine @tomdenton Thank you for your reply. I checked few samples and they seems the same. However, I saw that the primary/secondary labels have different typologie name between the Cornell Competition and Birdclef One. I think it is a more \"scientific name\" such as Limnodromus scolopaceus_Long-billed Dowitcher . I suppose you have a mapping name which translate \"Limnodromus scolopaceus_Long-billed Dowitcher\" into \"lobdow\"? I would like to convert the primary/secondary labels from cornell competition into the Birdclef format. Is it possible to get the same mapping ?

", "votes": 0, "replies": [] }, { "id": 1702838, "author_name": "Shiro", "author_url": "", "post_date": "2022-02-24T01:36:04.550000", "content": "

Just make the list of missing code name, is there a way to get their code name? :

\n
{'amazilia beryllina',\n  'amazilia cyanocephala',\n  'amazilia violiceps',\n  'amphispiza quinquestriata',\n  'atthis heloisa',\n  'campylopterus curvipennis',\n  'caracara cheriway',\n  'chlorostilbon ricordii',\n  'corvus caurinus',\n  'euphonia elegantissima',\n  'hylocharis leucotis',\n  'hylocharis xantusii',\n  'phalacrocorax brasilianus',\n  'sylvia cantillans',\n  'sylvia communis',\n  'sylvia curruca',\n  'sylvia melanocephala',\n  'sylvia mystacea',\n  'tetrao tetrix'}\n
", "votes": 0, "replies": [] }, { "id": 1704657, "author_name": "Tom Denton", "author_url": "", "post_date": "2022-02-25T17:34:03.793000", "content": "

Here's what I've got for those species:

\n
berhum,     amazilia beryllina\nazchum1,     amazilia cyanocephala\nvichum,     amazilia violiceps\nfisspa,     amphispiza quinquestriata\nbumhum,     atthis heloisa\nwetsab1,     campylopterus curvipennis\ncrecar1,     caracara cheriway\ncubeme1,     chlorostilbon ricordii\nnorcro,     corvus caurinus\neleeup1,     euphonia elegantissima\nwhehum,     hylocharis leucotis\nxanhum,     hylocharis xantusii\nneocor,     phalacrocorax brasilianus\nsubwar1,     sylvia cantillans\ngrewhi1,     sylvia communis\nleswhi1,     sylvia curruca\nsarwar1,     sylvia melanocephala\nmenwar1,     sylvia mystacea\nblagro1,     tetrao tetrix\n
\n

These are the e-bird codes, so you can generally get the code by googling the species, finding its page at ebird.org and looking at the code in the address bar. Like this one:
\nhttps://ebird.org/species/azchum1

", "votes": 3, "replies": [] } ] }, { "id": 1697085, "author_name": "Mithil Salunkhe", "author_url": "", "post_date": "2022-02-19T11:07:51.620000", "content": "

Is scraping the https://xeno-canto.org/ website allowed ?

", "votes": -1, "replies": [ { "id": 1697089, "author_name": "Sanyam Bhutani", "author_url": "", "post_date": "2022-02-19T11:12:04.053000", "content": "

@stefankahl had actually posted this last year.

\n

I think the website might have been updated since then, from the same thread, GM Rohan Rao had posted:

\n
\n

For those interested, the complete extended datasets (23,620 additional recordings) are publicly available for all 264 species (split in two by first alphabet due to Kaggle's 20GB limitation):

\n
\n

https://www.kaggle.com/rohanrao/xeno-canto-bird-recordings-extended-a-m
\nhttps://www.kaggle.com/rohanrao/xeno-canto-bird-recordings-extended-n-z

", "votes": 6, "replies": [] } ] }, { "id": 1774236, "author_name": "tc0000", "author_url": "", "post_date": "2022-05-02T01:56:35.187000", "content": "

Hi @tomdenton
\nyou display the leaderboard to the third decimal place?
\nLooking at the crowded leaderboards, it seems that the display up to the second decimal place is not enough. Consider Thank you.

", "votes": 0, "replies": [] }, { "id": 1733514, "author_name": "Hao", "author_url": "", "post_date": "2022-03-24T11:02:36.247000", "content": "
\n

In this competition, we focus on birds that are endemic to the chain of islands of Hawaii. We collected soundscape recordings and expert labels for 10 Hawaiian bird species,

\n
\n

So it`s clear that the test_soundscape recordings are all recorded on islands of Hawaii, but how about the train_audio recordings, were all of them recorded on islands of Hawaii as well?

", "votes": 0, "replies": [ { "id": 1733616, "author_name": "Bilzard", "author_url": "", "post_date": "2022-03-24T12:52:46.513000", "content": "

Since training data are downloaded from xeno-canto, it's not generally true, but I think almost all of the endemic species are only recorded in Hawaii.

\n

You can check by yourself by doing EDA of longitude and latitude of the recordings like myself[1].

\n", "votes": 1, "replies": [] } ] }, { "id": 1730092, "author_name": "Bilzard", "author_url": "", "post_date": "2022-03-20T23:09:29.977000", "content": "

I have a question.

\n
\n

We collected soundscape recordings and expert labels for 10 Hawaiian bird species

\n
\n

Why only 10 species? The number of scored species are 21. Is that mean that the rest of 11 species are processed in a different way?

", "votes": 0, "replies": [ { "id": 1730837, "author_name": "Sohier Dane", "author_url": "", "post_date": "2022-03-21T16:43:25.537000", "content": "

If I recall correctly, the other 11 were recorded incidentally during attempts to get audio for the 10 that are of primary interest.

", "votes": 2, "replies": [] }, { "id": 1733790, "author_name": "Bilzard", "author_url": "", "post_date": "2022-03-24T16:13:36.673000", "content": "

By the way, as far as the host can answer, do you differentiate the scores between the 10 primary concerned (endangered) species and the other 11 non-endangered species? For example, the 10 primary concerned species are weighted twice as much as the rest.

", "votes": 0, "replies": [] }, { "id": 1733864, "author_name": "Sohier Dane", "author_url": "", "post_date": "2022-03-24T17:08:05.350000", "content": "

Each species was assigned an equal weight.

", "votes": 1, "replies": [] } ] }, { "id": 1702410, "author_name": "MorillaBiomedical", "author_url": "", "post_date": "2022-02-23T15:28:11.990000", "content": "

Hello! We are a team that enters a competition for the first time, we have a question regarding the limitation of 9 hours, are they hours of training?

", "votes": 0, "replies": [ { "id": 1702557, "author_name": "Shiro", "author_url": "", "post_date": "2022-02-23T17:38:20.880000", "content": "

hours of inference

", "votes": 1, "replies": [] } ] }, { "id": 1774159, "author_name": "", "author_url": "", "post_date": "2022-05-01T22:55:07.963000", "content": "

Hi all, i am getting 'Notebook Exceeded Allowed Compute'. As far as I know I am doing prediction one at a time and not saving any of the files I am generating to make the prediction. This is my submission nb. Can anyone help me understand what is going on? thanks

\n

https://www.kaggle.com/code/rumman18/submission-notebook

", "votes": 0, "replies": [] }, { "id": 1772195, "author_name": "", "author_url": "", "post_date": "2022-04-30T01:03:56.177000", "content": "

Hi all, i have a question about pre trained models. Since internet will be disabled and some of keras pre trained models require internet access to get downloaded - i was wondering how we could go about that? thanks

", "votes": 1, "replies": [ { "id": 1772200, "author_name": "Sanyam Bhutani", "author_url": "", "post_date": "2022-04-30T01:15:28.503000", "content": "

Great question!

\n

Most Kagglers upload them as a dataset and use them in their notebooks-you can checkout popular notebooks from older competitions with similar requirements.

", "votes": 1, "replies": [] } ] }, { "id": 1734340, "author_name": "", "author_url": "", "post_date": "2022-03-25T08:16:51.820000", "content": "", "votes": 0, "replies": [] }, { "id": 1727430, "author_name": "", "author_url": "", "post_date": "2022-03-18T01:11:52.543000", "content": "

Hi, I have a quick question. What does the scored_birds.json actually mean here. There are ratings for all the birds in training metadata. How is scored different from that rating and why are there only few of the scored birds? thanks!

", "votes": 0, "replies": [ { "id": 1729629, "author_name": "Luis F. F. Furtado", "author_url": "", "post_date": "2022-03-20T09:58:22.163000", "content": "

Hey @rumman18, these are the Hawaiian species that should be in the submission file. Check the second cell of markdown/code in this public notebook for a more detailed description:
\nhttps://www.kaggle.com/code/stefankahl/how-to-submit-to-birdclef-2022/notebook

", "votes": 0, "replies": [] } ] }, { "id": 1709595, "author_name": "", "author_url": "", "post_date": "2022-03-02T10:54:59.080000", "content": "", "votes": 0, "replies": [] }, { "id": 1774144, "author_name": "john_lock", "author_url": "", "post_date": "2022-05-01T21:52:49.863000", "content": "

Thanks, everything is very clear now

", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "1691604": "Thanks for participating in this year’s BirdCLEF competition. Just like previous editions, this year’s challenge focuses on the science of recognizing birds by sound. There has been significant progress in this area of research in recent years, but we are not there yet. This is especially true for rare and endangered bird species. In many cases, only very few training recordings exists for those birds, yet, they're significantly more important to monitor due to their elevated conservation status.\n\n**Species portraits**\n\nIn this competition, we focus on birds that are endemic to the chain of islands of Hawaii. We collected soundscape recordings and expert labels for 10 Hawaiian bird species, and our host [Amanda](https://www.kaggle.com/amandanavine) prepared a [notebook that introduces each of the birds](https://www.kaggle.com/amandanavine/hawaiian-bird-species), so you get a better understanding of what we're looking for.\n\n**Sample submissions**\n\nAs in previous editions, this year's test data will be hidden and submitting a solution to this competition can be tricky. That's why we prepared another [notebook that briefly introduces the submission process](https://www.kaggle.com/stefankahl/how-to-submit-to-birdclef-2022).\n\n**Other resources**\n\nPlease make sure to check out previous editions either here on Kaggle ([2020](https://www.kaggle.com/c/birdsong-recognition), [2021](https://www.kaggle.com/c/birdclef-2021/overview/description)) or by browsing the archive of BirdCLEF [working notes and overview papers](https://scholar.google.com/scholar?q=birdclef).", "1777943": "Hi @stefankahl ,\nmy recent submission encountered Notebook Threw Exception without any details. However, the notebook runs smoothly on the publicly available test dataset. Could you please help me check what is going on?", "1777381": "@stefankahl @tomdenton @amandanavine \n\nThank you for organizing this interesting competition.\nI have a question about the evaluation metric.\n\nIs species weighting, as well as positive and negative weighting, calculated on the entire test data?\nOr is it calculated independently on public and private data?\n\nThanks", "1692315": "Thank you for organizing again a speech recognition competition!\nJust a question about the data, can we use the data from previous competition ?", "1697085": "Is scraping the https://xeno-canto.org/ website allowed ?", "1774236": "Hi @tomdenton \nyou display the leaderboard to the third decimal place?\nLooking at the crowded leaderboards, it seems that the display up to the second decimal place is not enough. Consider Thank you.", "1733514": "> In this competition, we focus on birds that are endemic to the chain of islands of Hawaii. We collected soundscape recordings and expert labels for 10 Hawaiian bird species,\n\nSo it`s clear that the test_soundscape recordings are all recorded on islands of Hawaii, but how about the train_audio recordings, were all of them recorded on islands of Hawaii as well?", "1730092": "I have a question.\n\n> We collected soundscape recordings and expert labels for 10 Hawaiian bird species\n\nWhy only 10 species? The number of scored species are 21. Is that mean that the rest of 11 species are processed in a different way?", "1702410": "Hello! We are a team that enters a competition for the first time, we have a question regarding the limitation of 9 hours, are they hours of training?", "1774159": "Hi all, i am getting 'Notebook Exceeded Allowed Compute'. As far as I know I am doing prediction one at a time and not saving any of the files I am generating to make the prediction. This is my submission nb. Can anyone help me understand what is going on? thanks\n\nhttps://www.kaggle.com/code/rumman18/submission-notebook", "1772195": "Hi all, i have a question about pre trained models. Since internet will be disabled and some of keras pre trained models require internet access to get downloaded - i was wondering how we could go about that? thanks", "1734340": "", "1727430": "Hi, I have a quick question. What does the scored_birds.json actually mean here. There are ratings for all the birds in training metadata. How is scored different from that rating and why are there only few of the scored birds? thanks!", "1709595": "", "1774144": "Thanks, everything is very clear now" } }