Hi everyone,
\nWe are happy to announce our sixth annual challenge, focused on helping predict the presence of plant and animal species at locations using previous observations paired with aerial images and environmental features. The challenge is part of LifeCLEF 2022 lab at CLEF 2022 and of the Ninth Workshop on Fine-Grained Visual Categorization (FGVC9) at CVPR 2022.
\nIn our first challenges, GeoLifeCLEF 2017, GeoLifeCLEF 2018 and GeoLifeCLEF 2019, each observation was associated only with environmental features given as vectors or patches extracted around the observation. Each year, the number of observations and species to predict increased gradually. Last two years, in GeoLifeCLEF 2020 and GeoLifeCLEF 2021, aerial images were added to each observation, furthermore, the number of species to detect was drastically increased to cover 33K species using data from, both, Pl@ntNet and iNaturalist.
\nThis year's challenge is based on a cleaned-up version of last year's data. The number of species was reduced to 17K mainly by removing species observed less than 3 times. In total, the number of observations was reduced from 1.9M to 1.6M. For a complete description of the changes, please check the changelog on the Data page.
\nLooking forward to the interesting and creative solutions that you all come up with. Feel free to ask questions here, or on the GitHub page!
", "messages": [ { "id": 1717922, "postDate": "2022-03-10T10:45:01.170Z", "content": "Hi everyone,
\nWe are happy to announce our sixth annual challenge, focused on helping predict the presence of plant and animal species at locations using previous observations paired with aerial images and environmental features. The challenge is part of LifeCLEF 2022 lab at CLEF 2022 and of the Ninth Workshop on Fine-Grained Visual Categorization (FGVC9) at CVPR 2022.
\nIn our first challenges, GeoLifeCLEF 2017, GeoLifeCLEF 2018 and GeoLifeCLEF 2019, each observation was associated only with environmental features given as vectors or patches extracted around the observation. Each year, the number of observations and species to predict increased gradually. Last two years, in GeoLifeCLEF 2020 and GeoLifeCLEF 2021, aerial images were added to each observation, furthermore, the number of species to detect was drastically increased to cover 33K species using data from, both, Pl@ntNet and iNaturalist.
\nThis year's challenge is based on a cleaned-up version of last year's data. The number of species was reduced to 17K mainly by removing species observed less than 3 times. In total, the number of observations was reduced from 1.9M to 1.6M. For a complete description of the changes, please check the changelog on the Data page.
\nLooking forward to the interesting and creative solutions that you all come up with. Feel free to ask questions here, or on the GitHub page!
", "rawMarkdown": "Hi everyone,\n\nWe are happy to announce our sixth annual challenge, focused on helping predict the presence of plant and animal species at locations using previous observations paired with aerial images and environmental features. The challenge is part of [LifeCLEF 2022](https://www.imageclef.org/LifeCLEF2022) lab at [CLEF 2022](http://clef2022.clef-initiative.eu/) and of the [Ninth Workshop on Fine-Grained Visual Categorization (FGVC9)](https://sites.google.com/view/fgvc9) at [CVPR 2022](http://cvpr2022.thecvf.com/).\n\nIn our first challenges, [GeoLifeCLEF 2017](https://www.imageclef.org/lifeclef/2017/GeoLifeCLEF), [GeoLifeCLEF 2018](https://www.imageclef.org/node/229) and [GeoLifeCLEF 2019](https://www.imageclef.org/GeoLifeCLEF2019), each observation was associated only with environmental features given as vectors or patches extracted around the observation. Each year, the number of observations and species to predict increased gradually. Last two years, in [GeoLifeCLEF 2020](https://www.imageclef.org/GeoLifeCLEF2020) and [GeoLifeCLEF 2021](https://www.imageclef.org/GeoLifeCLEF2021), aerial images were added to each observation, furthermore, the number of species to detect was drastically increased to cover 33K species using data from, both, Pl@ntNet and iNaturalist.\n\nThis year's challenge is based on a cleaned-up version of last year's data. The number of species was reduced to 17K mainly by removing species observed less than 3 times. In total, the number of observations was reduced from 1.9M to 1.6M. For a complete description of the changes, please check the changelog on the [Data](https://www.kaggle.com/c/geolifeclef-2022-lifeclef-2022-fgvc9/data) page.\n\nLooking forward to the interesting and creative solutions that you all come up with. Feel free to ask questions here, or on the [GitHub](https://github.com/maximiliense/GLC) page!", "votes": 6 }, { "id": 1718076, "postDate": "2022-03-10T13:38:57Z", "content": "Thanks for the competition. It looks really interesting. Unfortunately I am not so much into computer vision myself. However the data seems to match a potential usecase of mine. Given some lat long coordinate I'd like to describe the surroundings. Would you mind to elaborate a bit on data collection process ? You mention the sources, but do you use standard channels ? do you know of demos (notebooks) on how to use data sources ? some ways to make request 'en masse' ?
", "rawMarkdown": "Thanks for the competition. It looks really interesting. Unfortunately I am not so much into computer vision myself. However the data seems to match a potential usecase of mine. Given some lat long coordinate I'd like to describe the surroundings. Would you mind to elaborate a bit on data collection process ? You mention the sources, but do you use standard channels ? do you know of demos (notebooks) on how to use data sources ? some ways to make request 'en masse' ?", "replies": [ { "id": 1718129, "postDate": "2022-03-10T14:24:01.873Z", "content": "Hi Lucas, thanks for your interest! The data sources were collected manually before being extracted using custom scripts. Maybe Google Earth Engine API would be of interest for your use case.
", "rawMarkdown": "Hi Lucas, thanks for your interest! The data sources were collected manually before being extracted using custom scripts. Maybe Google Earth Engine API would be of interest for your use case.", "votes": 1 } ] }, { "id": 1738686, "postDate": "2022-03-29T13:03:00.377Z", "rawMarkdown": "", "isDeleted": true } ], "comments": [ { "id": 1718076, "author_name": "Lucas Morin", "author_url": "", "post_date": "2022-03-10T13:38:57", "content": "Thanks for the competition. It looks really interesting. Unfortunately I am not so much into computer vision myself. However the data seems to match a potential usecase of mine. Given some lat long coordinate I'd like to describe the surroundings. Would you mind to elaborate a bit on data collection process ? You mention the sources, but do you use standard channels ? do you know of demos (notebooks) on how to use data sources ? some ways to make request 'en masse' ?
", "votes": 0, "replies": [ { "id": 1718129, "author_name": "Titouan Lorieul", "author_url": "", "post_date": "2022-03-10T14:24:01.873000", "content": "Hi Lucas, thanks for your interest! The data sources were collected manually before being extracted using custom scripts. Maybe Google Earth Engine API would be of interest for your use case.
", "votes": 1, "replies": [] } ] }, { "id": 1738686, "author_name": "", "author_url": "", "post_date": "2022-03-29T13:03:00.377000", "content": "", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "1717922": "Hi everyone,\n\nWe are happy to announce our sixth annual challenge, focused on helping predict the presence of plant and animal species at locations using previous observations paired with aerial images and environmental features. The challenge is part of [LifeCLEF 2022](https://www.imageclef.org/LifeCLEF2022) lab at [CLEF 2022](http://clef2022.clef-initiative.eu/) and of the [Ninth Workshop on Fine-Grained Visual Categorization (FGVC9)](https://sites.google.com/view/fgvc9) at [CVPR 2022](http://cvpr2022.thecvf.com/).\n\nIn our first challenges, [GeoLifeCLEF 2017](https://www.imageclef.org/lifeclef/2017/GeoLifeCLEF), [GeoLifeCLEF 2018](https://www.imageclef.org/node/229) and [GeoLifeCLEF 2019](https://www.imageclef.org/GeoLifeCLEF2019), each observation was associated only with environmental features given as vectors or patches extracted around the observation. Each year, the number of observations and species to predict increased gradually. Last two years, in [GeoLifeCLEF 2020](https://www.imageclef.org/GeoLifeCLEF2020) and [GeoLifeCLEF 2021](https://www.imageclef.org/GeoLifeCLEF2021), aerial images were added to each observation, furthermore, the number of species to detect was drastically increased to cover 33K species using data from, both, Pl@ntNet and iNaturalist.\n\nThis year's challenge is based on a cleaned-up version of last year's data. The number of species was reduced to 17K mainly by removing species observed less than 3 times. In total, the number of observations was reduced from 1.9M to 1.6M. For a complete description of the changes, please check the changelog on the [Data](https://www.kaggle.com/c/geolifeclef-2022-lifeclef-2022-fgvc9/data) page.\n\nLooking forward to the interesting and creative solutions that you all come up with. Feel free to ask questions here, or on the [GitHub](https://github.com/maximiliense/GLC) page!", "1718076": "Thanks for the competition. It looks really interesting. Unfortunately I am not so much into computer vision myself. However the data seems to match a potential usecase of mine. Given some lat long coordinate I'd like to describe the surroundings. Would you mind to elaborate a bit on data collection process ? You mention the sources, but do you use standard channels ? do you know of demos (notebooks) on how to use data sources ? some ways to make request 'en masse' ?", "1738686": "" } }