Hi everyone,
\n\nWe are happy to announce this year's iNaturalist species classification competition. The competition will be held in conjunction with the Sixth Workshop on Fine-Grained Visual Categorization at CVPR 2019 in Long Beach, CA.
\n\nThis year, to make the competition more accessible, and at the same time more challenging, we have reduced the overall size of the dataset by focusing on highly similar categories. In total, the 2019 dataset contains 1,010 species, with a combined training and validation set of 268,243 images that have been collected and verified by multiple users from the citizen science website iNaturalist.
\n\nLooking forward to the interesting and creative solutions that you all come up with. Feel free to ask questions here.
", "messages": [ { "id": 505168, "postDate": "2019-04-01T16:37:00.267Z", "content": "Hi everyone,
\n\nWe are happy to announce this year's iNaturalist species classification competition. The competition will be held in conjunction with the Sixth Workshop on Fine-Grained Visual Categorization at CVPR 2019 in Long Beach, CA.
\n\nThis year, to make the competition more accessible, and at the same time more challenging, we have reduced the overall size of the dataset by focusing on highly similar categories. In total, the 2019 dataset contains 1,010 species, with a combined training and validation set of 268,243 images that have been collected and verified by multiple users from the citizen science website iNaturalist.
\n\nLooking forward to the interesting and creative solutions that you all come up with. Feel free to ask questions here.
", "rawMarkdown": "Hi everyone,\n\nWe are happy to announce this year's iNaturalist species classification competition. The competition will be held in conjunction with the [Sixth Workshop on Fine-Grained Visual Categorization](https://sites.google.com/view/fgvc6) at CVPR 2019 in Long Beach, CA.\n\nThis year, to make the competition more accessible, and at the same time more challenging, we have reduced the overall size of the dataset by focusing on highly similar categories. In total, the 2019 dataset contains 1,010 species, with a combined training and validation set of 268,243 images that have been collected and verified by multiple users from the citizen science website [iNaturalist](https://www.inaturalist.org).\n\nLooking forward to the interesting and creative solutions that you all come up with. Feel free to ask questions here. ", "votes": 9 }, { "id": 530483, "postDate": "2019-05-13T01:48:38.873Z", "content": "Happy to join Kaggle. Its my first trail. Greetings everyone!
", "rawMarkdown": "Happy to join Kaggle. Its my first trail. Greetings everyone!", "votes": 1 }, { "id": 506941, "postDate": "2019-04-04T04:01:10.997Z", "content": "Thanks for coordinating.\nI have a simple question. Why the category of this year is only about 1000 while that of last year was 8000?
", "rawMarkdown": "Thanks for coordinating.\nI have a simple question. Why the category of this year is only about 1000 while that of last year was 8000?", "votes": 1, "replies": [ { "id": 506962, "postDate": "2019-04-04T04:50:07.563Z", "content": "Thanks for your interest. This year we opted to reduce the overall size of the dataset by focusing efforts on a smaller number of highly similar categories.
", "rawMarkdown": "Thanks for your interest. This year we opted to reduce the overall size of the dataset by focusing efforts on a smaller number of highly similar categories. ", "votes": 2 } ] }, { "id": 505347, "postDate": "2019-04-01T22:41:08.123Z", "content": "Thank you for organizing this, is that possible to provide some detail of the baseline \"Inception Benchmark\".\nFor example, the resolution, framework, etc, thank you!
", "rawMarkdown": "Thank you for organizing this, is that possible to provide some detail of the baseline \"Inception Benchmark\".\nFor example, the resolution, framework, etc, thank you!\n", "votes": 1, "replies": [ { "id": 506967, "postDate": "2019-04-04T04:53:47.400Z", "content": "Resolution is 299x299, trained with TensorFlow.
", "rawMarkdown": "Resolution is 299x299, trained with TensorFlow.", "votes": 2 }, { "id": 544237, "postDate": "2019-06-05T09:47:39.947Z", "content": "Does Resolution have to be 229x299 ?
", "rawMarkdown": "Does Resolution have to be 229x299 ?" }, { "id": 546506, "postDate": "2019-06-06T16:16:34.127Z", "content": "You are free to train with any resolution you want.
", "rawMarkdown": "You are free to train with any resolution you want. " } ] }, { "id": 507107, "postDate": "2019-04-04T09:54:30.520Z", "content": "Just for curiosity in the val2019.json file and for others as well, in the 'annotations' field, there are two subfield of 'id' and 'image_id' which are essentially the same. So is there is some reason that there is duplication here. Or am I missing something?
", "rawMarkdown": "Just for curiosity in the val2019.json file and for others as well, in the 'annotations' field, there are two subfield of 'id' and 'image_id' which are essentially the same. So is there is some reason that there is duplication here. Or am I missing something?", "votes": 2, "replies": [ { "id": 507398, "postDate": "2019-04-04T16:34:00.547Z", "content": "Hey Kushajveer,
\n\nWe aimed to keep our annotation format consistent with the COCO dataset. If you take a look at the COCO description you will see that they also have an 'id' and 'image_id' for each annotation. For the annotations list, 'id' is the id of the annotation and 'image_id' is the image it refers to. For iNat2019 you will notice that they are the same.
", "rawMarkdown": "Hey Kushajveer, \n\nWe aimed to keep our annotation format consistent with the COCO dataset. If you take a look at the [COCO description](http://cocodataset.org/#format-data) you will see that they also have an 'id' and 'image_id' for each annotation. For the annotations list, 'id' is the id of the annotation and 'image_id' is the image it refers to. For iNat2019 you will notice that they are the same. \n", "votes": 3 }, { "id": 507489, "postDate": "2019-04-04T18:43:01.087Z", "content": "Thanks for the clarification.
", "rawMarkdown": "Thanks for the clarification.", "votes": 1 } ] }, { "id": 523341, "postDate": "2019-04-26T03:22:14.260Z", "content": "Thanks for your great effort to organize the competition.\nAlthogh the max team size is 5, can I have more than 5 members in the team(such as 6) when I get to top-2 and go to CVPR for presentation?
", "rawMarkdown": "Thanks for your great effort to organize the competition.\nAlthogh the max team size is 5, can I have more than 5 members in the team(such as 6) when I get to top-2 and go to CVPR for presentation?", "replies": [ { "id": 523627, "postDate": "2019-04-26T16:43:47.503Z", "content": "Hi NoName, thanks for participating.
\n\nAccording to the rules the maximum team size is five. To be fair to the other competitors we will keep this fixed. We can consider revising this for next year if there is enough interest from participants.
\n\nTo clarify, the top entries are not guaranteed a presentation slot at the workshop. This will be decided based on the novelty of the method and taking time restrictions into account. The workshop will be open to anyone at CVPR we welcome participation from many people.
\n\nBest of luck! I hope this answers your question.
", "rawMarkdown": "Hi NoName, thanks for participating. \n\nAccording to the rules the maximum team size is five. To be fair to the other competitors we will keep this fixed. We can consider revising this for next year if there is enough interest from participants.\n\nTo clarify, the top entries are not guaranteed a presentation slot at the workshop. This will be decided based on the novelty of the method and taking time restrictions into account. The workshop will be open to anyone at CVPR we welcome participation from many people. \n\nBest of luck! I hope this answers your question. " } ] }, { "id": 520161, "postDate": "2019-04-20T09:53:08.903Z", "rawMarkdown": "", "isDeleted": true }, { "id": 505260, "postDate": "2019-04-01T19:28:53.343Z", "rawMarkdown": "", "isDeleted": true } ], "comments": [ { "id": 530483, "author_name": "iamctR", "author_url": "", "post_date": "2019-05-13T01:48:38.873000", "content": "Happy to join Kaggle. Its my first trail. Greetings everyone!
", "votes": 1, "replies": [] }, { "id": 506941, "author_name": "NADATER_YCYLG", "author_url": "", "post_date": "2019-04-04T04:01:10.997000", "content": "Thanks for coordinating.\nI have a simple question. Why the category of this year is only about 1000 while that of last year was 8000?
", "votes": 1, "replies": [ { "id": 506962, "author_name": "macaodha", "author_url": "", "post_date": "2019-04-04T04:50:07.563000", "content": "Thanks for your interest. This year we opted to reduce the overall size of the dataset by focusing efforts on a smaller number of highly similar categories.
", "votes": 2, "replies": [] } ] }, { "id": 505347, "author_name": "zz", "author_url": "", "post_date": "2019-04-01T22:41:08.123000", "content": "Thank you for organizing this, is that possible to provide some detail of the baseline \"Inception Benchmark\".\nFor example, the resolution, framework, etc, thank you!
", "votes": 1, "replies": [ { "id": 506967, "author_name": "macaodha", "author_url": "", "post_date": "2019-04-04T04:53:47.400000", "content": "Resolution is 299x299, trained with TensorFlow.
", "votes": 2, "replies": [] }, { "id": 544237, "author_name": "shuwang001", "author_url": "", "post_date": "2019-06-05T09:47:39.947000", "content": "Does Resolution have to be 229x299 ?
", "votes": 0, "replies": [] }, { "id": 546506, "author_name": "macaodha", "author_url": "", "post_date": "2019-06-06T16:16:34.127000", "content": "You are free to train with any resolution you want.
", "votes": 0, "replies": [] } ] }, { "id": 507107, "author_name": "Kushajveer Singh", "author_url": "", "post_date": "2019-04-04T09:54:30.520000", "content": "Just for curiosity in the val2019.json file and for others as well, in the 'annotations' field, there are two subfield of 'id' and 'image_id' which are essentially the same. So is there is some reason that there is duplication here. Or am I missing something?
", "votes": 2, "replies": [ { "id": 507398, "author_name": "macaodha", "author_url": "", "post_date": "2019-04-04T16:34:00.547000", "content": "Hey Kushajveer,
\n\nWe aimed to keep our annotation format consistent with the COCO dataset. If you take a look at the COCO description you will see that they also have an 'id' and 'image_id' for each annotation. For the annotations list, 'id' is the id of the annotation and 'image_id' is the image it refers to. For iNat2019 you will notice that they are the same.
", "votes": 3, "replies": [] }, { "id": 507489, "author_name": "Kushajveer Singh", "author_url": "", "post_date": "2019-04-04T18:43:01.087000", "content": "Thanks for the clarification.
", "votes": 1, "replies": [] } ] }, { "id": 523341, "author_name": "Megvii Research Nanjing", "author_url": "", "post_date": "2019-04-26T03:22:14.260000", "content": "Thanks for your great effort to organize the competition.\nAlthogh the max team size is 5, can I have more than 5 members in the team(such as 6) when I get to top-2 and go to CVPR for presentation?
", "votes": 0, "replies": [ { "id": 523627, "author_name": "macaodha", "author_url": "", "post_date": "2019-04-26T16:43:47.503000", "content": "Hi NoName, thanks for participating.
\n\nAccording to the rules the maximum team size is five. To be fair to the other competitors we will keep this fixed. We can consider revising this for next year if there is enough interest from participants.
\n\nTo clarify, the top entries are not guaranteed a presentation slot at the workshop. This will be decided based on the novelty of the method and taking time restrictions into account. The workshop will be open to anyone at CVPR we welcome participation from many people.
\n\nBest of luck! I hope this answers your question.
", "votes": 0, "replies": [] } ] }, { "id": 520161, "author_name": "", "author_url": "", "post_date": "2019-04-20T09:53:08.903000", "content": "", "votes": 0, "replies": [] }, { "id": 505260, "author_name": "", "author_url": "", "post_date": "2019-04-01T19:28:53.343000", "content": "", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "505168": "Hi everyone,\n\nWe are happy to announce this year's iNaturalist species classification competition. The competition will be held in conjunction with the [Sixth Workshop on Fine-Grained Visual Categorization](https://sites.google.com/view/fgvc6) at CVPR 2019 in Long Beach, CA.\n\nThis year, to make the competition more accessible, and at the same time more challenging, we have reduced the overall size of the dataset by focusing on highly similar categories. In total, the 2019 dataset contains 1,010 species, with a combined training and validation set of 268,243 images that have been collected and verified by multiple users from the citizen science website [iNaturalist](https://www.inaturalist.org).\n\nLooking forward to the interesting and creative solutions that you all come up with. Feel free to ask questions here. ", "530483": "Happy to join Kaggle. Its my first trail. Greetings everyone!", "506941": "Thanks for coordinating.\nI have a simple question. Why the category of this year is only about 1000 while that of last year was 8000?", "505347": "Thank you for organizing this, is that possible to provide some detail of the baseline \"Inception Benchmark\".\nFor example, the resolution, framework, etc, thank you!\n", "507107": "Just for curiosity in the val2019.json file and for others as well, in the 'annotations' field, there are two subfield of 'id' and 'image_id' which are essentially the same. So is there is some reason that there is duplication here. Or am I missing something?", "523341": "Thanks for your great effort to organize the competition.\nAlthogh the max team size is 5, can I have more than 5 members in the team(such as 6) when I get to top-2 and go to CVPR for presentation?", "520161": "", "505260": "" } }