Hi everyone!
\nWe are happy to announce our 5th annual camera trap challenge, focused on helping to automate data processing and analysis in challenging passive-monitoring cameras placed in the wild and used by ecologists to study animal populations and behavior. The challenge is part of the Ninth Workshop on Fine-Grained Visual Categorization (FGVC9) at CVPR 2022.
\nIn our first challenge, iWildCam 2018, we focused on binary empty/animal classification. In the second year of our challenge, iWildCam 2019, we focused on generalization to an out-of-sample part of the world with a non-identical set of species. In iWildCam 2020 we focused on multimodal data fusion as a way to improve generalization to new camera locations across the globe by providing matched multispectral remote sensing data and iNaturalist data. Finally, in iWildCam 2021, we extended our competition to an even more challenging task: counting how many of each species is visible across a sequence of images.
\nWe learned from last year how difficult it is to solve and interpret the results on two hard tasks at the same time, so this year in iWildCam 2022 we will be focusing entirely on the counting task: how many animals do you see in each sequence of images, irrespective of the species? We hope that the simpler evaluation metric (mean absolute error) and the fact that we are now providing count annotations on a portion of the train sequences will help participants in building better models.
\nWe're looking forward to your creative solutions! Feel free to ask questions in the Kaggle forum or by simply opening an issue on the competition's GitHub page.
", "messages": [ { "id": "1732592", "postDate": "03/23/2022 14:40:50", "content": "Hi everyone!
\nWe are happy to announce our 5th annual camera trap challenge, focused on helping to automate data processing and analysis in challenging passive-monitoring cameras placed in the wild and used by ecologists to study animal populations and behavior. The challenge is part of the Ninth Workshop on Fine-Grained Visual Categorization (FGVC9) at CVPR 2022.
\nIn our first challenge, iWildCam 2018, we focused on binary empty/animal classification. In the second year of our challenge, iWildCam 2019, we focused on generalization to an out-of-sample part of the world with a non-identical set of species. In iWildCam 2020 we focused on multimodal data fusion as a way to improve generalization to new camera locations across the globe by providing matched multispectral remote sensing data and iNaturalist data. Finally, in iWildCam 2021, we extended our competition to an even more challenging task: counting how many of each species is visible across a sequence of images.
\nWe learned from last year how difficult it is to solve and interpret the results on two hard tasks at the same time, so this year in iWildCam 2022 we will be focusing entirely on the counting task: how many animals do you see in each sequence of images, irrespective of the species? We hope that the simpler evaluation metric (mean absolute error) and the fact that we are now providing count annotations on a portion of the train sequences will help participants in building better models.
\nWe're looking forward to your creative solutions! Feel free to ask questions in the Kaggle forum or by simply opening an issue on the competition's GitHub page.
", "rawMarkdown": "Hi everyone!\n\nWe are happy to announce our 5th annual camera trap challenge, focused on helping to automate data processing and analysis in challenging passive-monitoring cameras placed in the wild and used by ecologists to study animal populations and behavior. The challenge is part of the [Ninth Workshop on Fine-Grained Visual Categorization (FGVC9)](https://sites.google.com/view/fgvc9) at [CVPR 2022](https://cvpr2022.thecvf.com/).\n\nIn our first challenge, [iWildCam 2018](https://www.kaggle.com/c/iwildcam2018), we focused on binary empty/animal classification. In the second year of our challenge, [iWildCam 2019](https://www.kaggle.com/c/iwildcam-2019-fgvc6), we focused on generalization to an out-of-sample part of the world with a non-identical set of species. In [iWildCam 2020](https://www.kaggle.com/c/iwildcam-2020-fgvc7) we focused on multimodal data fusion as a way to improve generalization to new camera locations across the globe by providing matched multispectral remote sensing data and iNaturalist data. Finally, in [iWildCam 2021](https://www.kaggle.com/c/iwildcam2021-fgvc8), we extended our competition to an even more challenging task: counting how many of each species is visible across a sequence of images.\n\nWe learned from last year how difficult it is to solve and interpret the results on two hard tasks at the same time, so this year in [iWildCam 2022](https://www.kaggle.com/c/iwildcam2022-fgvc9) we will be focusing entirely on the counting task: **how many animals do you see in each sequence of images, irrespective of the species?** We hope that the simpler evaluation metric (mean absolute error) and the fact that we are now providing count annotations on a portion of the train sequences will help participants in building better models.\n\nWe're looking forward to your creative solutions! Feel free to ask questions in the [Kaggle forum](https://www.kaggle.com/c/iwildcam2022-fgvc9/discussion) or by simply opening an issue on the competition's [GitHub page](https://github.com/visipedia/iwildcam_comp).", "votes": null }, { "id": "3267477", "postDate": "08/11/2025 09:42:40", "content": "It seems very intreresting
", "rawMarkdown": "It seems very intreresting", "votes": null } ], "comments": [ { "id": 3267477, "author_name": "justmariakorol", "author_url": "", "post_date": "08/11/2025 09:42:40", "content": "It seems very intreresting
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "1732592": "Hi everyone!\n\nWe are happy to announce our 5th annual camera trap challenge, focused on helping to automate data processing and analysis in challenging passive-monitoring cameras placed in the wild and used by ecologists to study animal populations and behavior. The challenge is part of the [Ninth Workshop on Fine-Grained Visual Categorization (FGVC9)](https://sites.google.com/view/fgvc9) at [CVPR 2022](https://cvpr2022.thecvf.com/).\n\nIn our first challenge, [iWildCam 2018](https://www.kaggle.com/c/iwildcam2018), we focused on binary empty/animal classification. In the second year of our challenge, [iWildCam 2019](https://www.kaggle.com/c/iwildcam-2019-fgvc6), we focused on generalization to an out-of-sample part of the world with a non-identical set of species. In [iWildCam 2020](https://www.kaggle.com/c/iwildcam-2020-fgvc7) we focused on multimodal data fusion as a way to improve generalization to new camera locations across the globe by providing matched multispectral remote sensing data and iNaturalist data. Finally, in [iWildCam 2021](https://www.kaggle.com/c/iwildcam2021-fgvc8), we extended our competition to an even more challenging task: counting how many of each species is visible across a sequence of images.\n\nWe learned from last year how difficult it is to solve and interpret the results on two hard tasks at the same time, so this year in [iWildCam 2022](https://www.kaggle.com/c/iwildcam2022-fgvc9) we will be focusing entirely on the counting task: **how many animals do you see in each sequence of images, irrespective of the species?** We hope that the simpler evaluation metric (mean absolute error) and the fact that we are now providing count annotations on a portion of the train sequences will help participants in building better models.\n\nWe're looking forward to your creative solutions! Feel free to ask questions in the [Kaggle forum](https://www.kaggle.com/c/iwildcam2022-fgvc9/discussion) or by simply opening an issue on the competition's [GitHub page](https://github.com/visipedia/iwildcam_comp).", "3267477": "It seems very intreresting" }, "source": "meta" }