The first question after last year competition:
\nIs BirdNET available to use? And some external data on which BirdNET was trained? :)
\n(I hope not)
fyi: the last year winner
", "messages": [ { "id": 2172721, "postDate": "2023-03-07T18:18:10.923Z", "content": "The first question after last year competition:
\nIs BirdNET available to use? And some external data on which BirdNET was trained? :)
\n(I hope not)
fyi: the last year winner
", "rawMarkdown": "The first question after last year competition:\nIs BirdNET available to use? And some external data on which BirdNET was trained? :)\n(I hope not)\n\nfyi: [the last year winner](https://github.com/kahst/BirdNET-Analyzer)", "votes": 26 }, { "id": 2172839, "postDate": "2023-03-07T21:02:17.660Z", "content": "Yes, you can use it. HOWEVER: BirdNET does not feature all relevant species and was not trained with additional training data for the species of this competition. So, using vanilla BirdNET won't be enough to succeed in this competition. This year, we deliberately constrained BirdNET and the species it covers to avoid unfair advantages. Which is easier for us than restricting the use because it is open source and thus an eligible external source. Moreover, there is another model (\"Chirp\") available here.
", "rawMarkdown": "Yes, you can use it. HOWEVER: BirdNET does not feature all relevant species and was not trained with additional training data for the species of this competition. So, using vanilla BirdNET won't be enough to succeed in this competition. This year, we deliberately constrained BirdNET and the species it covers to avoid unfair advantages. Which is easier for us than restricting the use because it is open source and thus an eligible external source. Moreover, there is another model (\"Chirp\") available [here](https://www.kaggle.com/models/google/bird-vocalization-classifier).", "votes": 13, "replies": [ { "id": 2172856, "postDate": "2023-03-07T21:20:31.740Z", "content": "Are all the test species covered by the training data?
", "rawMarkdown": "Are all the test species covered by the training data?", "votes": 6, "replies": [ { "id": 2172877, "postDate": "2023-03-07T22:08:30.483Z", "content": "Yes, but not all of them are covered by BirdNET (I think only 40% or so are covered).
", "rawMarkdown": "Yes, but not all of them are covered by BirdNET (I think only 40% or so are covered).", "votes": 7 } ] }, { "id": 2172930, "postDate": "2023-03-07T23:57:53.827Z", "content": "Thanks for answer.
\nAnd the Macaulay data still restricted?
\n(I don't have access to it but maybe someone has)
Unfortunately, yes.
", "rawMarkdown": "Unfortunately, yes.", "votes": 2 }, { "id": 2214696, "postDate": "2023-04-08T17:12:18.013Z", "content": "Hi, I found there are many extra sound of the rare species on ebird site, but they are not allowed to use in this competition, right? (ebird site == Macaulay data? Are they the same?)
\nI think this is not a public dataset because not all the competitors have the api key(although you can just register and get one). I just want to have a double check.
Correct, ebird==macaulay, and is not allowed for the competition.
", "rawMarkdown": "Correct, ebird==macaulay, and is not allowed for the competition.", "votes": 2 }, { "id": 2214753, "postDate": "2023-04-08T18:09:29.733Z", "content": "Thanks, got it
", "rawMarkdown": "Thanks, got it" }, { "id": 2232799, "postDate": "2023-04-24T16:21:39.337Z", "rawMarkdown": "", "isDeleted": true }, { "id": 2235557, "postDate": "2023-04-26T06:21:05.933Z", "content": "Hi, I have another question. Does the data with CC BY-NC-ND license allowed in the competition?
\nI am reading the following url and according url1, in terms of training AI, data access to the copyright works was lawful at the point of input. Thus according to the graph in url2, the CC license is not triggered.
\nTo summarize, it seems that the data with CC BY-NC-ND is allowed to train AI, but I would like to have a double check to keep myself follow the competition rule.
", "rawMarkdown": "Hi, I have another question. Does the data with CC BY-NC-ND license allowed in the competition?\n\nI am reading the following url and according url1, in terms of training AI, data access to the copyright works was lawful at the point of input. Thus according to the graph in url2, the CC license is not triggered. \n\n1. https://creativecommons.org/2021/03/04/should-cc-licensed-content-be-used-to-train-ai-it-depends/#fn2\n2. https://creativecommons.org/faq/#artificial-intelligence-and-cc-licenses\n\nTo summarize, it seems that the data with CC BY-NC-ND is allowed to train AI, but I would like to have a double check to keep myself follow the competition rule.\n", "votes": 1 }, { "id": 2238767, "postDate": "2023-04-28T19:25:13.353Z", "content": "Training on CC-BY-NC-ND licenses is fine.
", "rawMarkdown": "Training on CC-BY-NC-ND licenses is fine.", "votes": 2 }, { "id": 2238956, "postDate": "2023-04-29T02:31:07.477Z", "content": "Thanks for your answer！
", "rawMarkdown": "Thanks for your answer！" } ] } ] }, { "id": 2236354, "postDate": "2023-04-26T19:27:41.720Z", "content": "So it sounds like it's OK to use BirdNet and Chirp, but they wouldn't be much good as classifiers. I was looking for ways just to localise the bird calls in the training data. Is this a crazy idea? They might still be pretty good at recognising new birds over background noise.
", "rawMarkdown": "So it sounds like it's OK to use BirdNet and Chirp, but they wouldn't be much good as classifiers. I was looking for ways just to localise the bird calls in the training data. Is this a crazy idea? They might still be pretty good at recognising new birds over background noise.", "votes": 1, "replies": [ { "id": 2238455, "postDate": "2023-04-28T14:41:32.690Z", "content": "Yes, sure, feel free to utilize these models for different tasks. You are right, though, they might not be of much help for the final submission.
", "rawMarkdown": "Yes, sure, feel free to utilize these models for different tasks. You are right, though, they might not be of much help for the final submission." } ] } ], "comments": [ { "id": 2172839, "author_name": "Stefan Kahl", "author_url": "", "post_date": "2023-03-07T21:02:17.660000", "content": "Yes, you can use it. HOWEVER: BirdNET does not feature all relevant species and was not trained with additional training data for the species of this competition. So, using vanilla BirdNET won't be enough to succeed in this competition. This year, we deliberately constrained BirdNET and the species it covers to avoid unfair advantages. Which is easier for us than restricting the use because it is open source and thus an eligible external source. Moreover, there is another model (\"Chirp\") available here.
", "votes": 13, "replies": [ { "id": 2172856, "author_name": "beluga", "author_url": "", "post_date": "2023-03-07T21:20:31.740000", "content": "Are all the test species covered by the training data?
", "votes": 6, "replies": [ { "id": 2172877, "author_name": "Stefan Kahl", "author_url": "", "post_date": "2023-03-07T22:08:30.483000", "content": "Yes, but not all of them are covered by BirdNET (I think only 40% or so are covered).
", "votes": 7, "replies": [] } ] }, { "id": 2172930, "author_name": "Kolya Forrat", "author_url": "", "post_date": "2023-03-07T23:57:53.827000", "content": "Thanks for answer.
\nAnd the Macaulay data still restricted?
\n(I don't have access to it but maybe someone has)
Unfortunately, yes.
", "votes": 2, "replies": [] }, { "id": 2214696, "author_name": "RihanPiggy", "author_url": "", "post_date": "2023-04-08T17:12:18.013000", "content": "Hi, I found there are many extra sound of the rare species on ebird site, but they are not allowed to use in this competition, right? (ebird site == Macaulay data? Are they the same?)
\nI think this is not a public dataset because not all the competitors have the api key(although you can just register and get one). I just want to have a double check.
Correct, ebird==macaulay, and is not allowed for the competition.
", "votes": 2, "replies": [] }, { "id": 2214753, "author_name": "RihanPiggy", "author_url": "", "post_date": "2023-04-08T18:09:29.733000", "content": "Thanks, got it
", "votes": 0, "replies": [] }, { "id": 2232799, "author_name": "", "author_url": "", "post_date": "2023-04-24T16:21:39.337000", "content": "", "votes": 0, "replies": [] }, { "id": 2235557, "author_name": "RihanPiggy", "author_url": "", "post_date": "2023-04-26T06:21:05.933000", "content": "Hi, I have another question. Does the data with CC BY-NC-ND license allowed in the competition?
\nI am reading the following url and according url1, in terms of training AI, data access to the copyright works was lawful at the point of input. Thus according to the graph in url2, the CC license is not triggered.
\nTo summarize, it seems that the data with CC BY-NC-ND is allowed to train AI, but I would like to have a double check to keep myself follow the competition rule.
", "votes": 1, "replies": [] }, { "id": 2238767, "author_name": "Tom Denton", "author_url": "", "post_date": "2023-04-28T19:25:13.353000", "content": "Training on CC-BY-NC-ND licenses is fine.
", "votes": 2, "replies": [] }, { "id": 2238956, "author_name": "RihanPiggy", "author_url": "", "post_date": "2023-04-29T02:31:07.477000", "content": "Thanks for your answer！
", "votes": 0, "replies": [] } ] } ] }, { "id": 2236354, "author_name": "Olly Powell", "author_url": "", "post_date": "2023-04-26T19:27:41.720000", "content": "So it sounds like it's OK to use BirdNet and Chirp, but they wouldn't be much good as classifiers. I was looking for ways just to localise the bird calls in the training data. Is this a crazy idea? They might still be pretty good at recognising new birds over background noise.
", "votes": 1, "replies": [ { "id": 2238455, "author_name": "Stefan Kahl", "author_url": "", "post_date": "2023-04-28T14:41:32.690000", "content": "Yes, sure, feel free to utilize these models for different tasks. You are right, though, they might not be of much help for the final submission.
", "votes": 0, "replies": [] } ] } ], "raw_markdown_by_id": { "2172721": "The first question after last year competition:\nIs BirdNET available to use? And some external data on which BirdNET was trained? :)\n(I hope not)\n\nfyi: [the last year winner](https://github.com/kahst/BirdNET-Analyzer)", "2172839": "Yes, you can use it. HOWEVER: BirdNET does not feature all relevant species and was not trained with additional training data for the species of this competition. So, using vanilla BirdNET won't be enough to succeed in this competition. This year, we deliberately constrained BirdNET and the species it covers to avoid unfair advantages. Which is easier for us than restricting the use because it is open source and thus an eligible external source. Moreover, there is another model (\"Chirp\") available [here](https://www.kaggle.com/models/google/bird-vocalization-classifier).", "2236354": "So it sounds like it's OK to use BirdNet and Chirp, but they wouldn't be much good as classifiers. I was looking for ways just to localise the bird calls in the training data. Is this a crazy idea? They might still be pretty good at recognising new birds over background noise." } }