If anyone is facing difficulty in generating your own custom dataset, I can share my script (will take 1-2 days to run depending on your compute power).
\nOtherwise, I can share the generated data itself. Cz, I don't think someone without sufficient compute can generate it. Data has same size as current dataset but with all possible annotations you can think of - semantic segmentation, instance segmentation, classification, multilabel classification etc
", "messages": [ { "id": 1729733, "postDate": "2022-03-20T12:45:50.240Z", "content": "I don't think it's allowed. This is equivalent to hand labeling the training data.
", "rawMarkdown": "I don't think it's allowed. This is equivalent to hand labeling the training data.", "votes": 3, "replies": [ { "id": 1729988, "postDate": "2022-03-20T19:21:46.803Z", "content": "Can you please tell me where restrictions are mentioned? Weren't all restrictions removed?
", "rawMarkdown": "Can you please tell me where restrictions are mentioned? Weren't all restrictions removed?" }, { "id": 1729993, "postDate": "2022-03-20T19:33:33.100Z", "content": "https://www.kaggle.com/c/ultra-mnist/rules
\n\"Submissions may not use or incorporate information from hand labeling or human prediction of the training or test data records.\"
", "rawMarkdown": "https://www.kaggle.com/c/ultra-mnist/rules\n\n\"Submissions may not use or incorporate information from hand labeling or human prediction of the training or test data records.\"" }, { "id": 1729999, "postDate": "2022-03-20T19:44:08.053Z", "content": "It is not hand labeled data. This data is generated, so I think it can be used in this competition.
", "rawMarkdown": "It is not hand labeled data. This data is generated, so I think it can be used in this competition.", "votes": 1, "replies": [ { "id": 1730002, "postDate": "2022-03-20T19:48:51.247Z", "rawMarkdown": "", "isDeleted": true } ] }, { "id": 1741469, "postDate": "2022-03-31T20:13:27.913Z", "content": "Do you mean random image modifications such as translation, rotation etc. ?
\nOr do you mean something else?
@dkgupta90 can you please clear it out, whether this approach is acceptable or not for the GPU track
\nThank You
", "rawMarkdown": "@dkgupta90 can you please clear it out, whether this approach is acceptable or not for the GPU track\n\nThank You" }, { "id": 1743850, "postDate": "2022-04-03T11:13:39.160Z", "content": "As long as no hand-labelling is involved, it is good to go for the GPU track.
", "rawMarkdown": "As long as no hand-labelling is involved, it is good to go for the GPU track.", "votes": 1 } ] }, { "id": 1729294, "postDate": "2022-03-19T22:33:03.547Z", "content": "If anyone is facing difficulty in generating your own custom dataset, I can share my script (will take 1-2 days to run depending on your compute power).
\nOtherwise, I can share the generated data itself. Cz, I don't think someone without sufficient compute can generate it. Data has same size as current dataset but with all possible annotations you can think of - semantic segmentation, instance segmentation, classification, multilabel classification etc
", "rawMarkdown": "If anyone is facing difficulty in generating your own custom dataset, I can share my script (will take 1-2 days to run depending on your compute power).\n\nOtherwise, I can share the generated data itself. Cz, I don't think someone without sufficient compute can generate it. Data has same size as current dataset but with all possible annotations you can think of - semantic segmentation, instance segmentation, classification, multilabel classification etc", "votes": 3 }, { "id": 1741761, "postDate": "2022-04-01T05:57:55.277Z", "content": "Can you please share this dataset? It will be really helpful.
", "rawMarkdown": "Can you please share this dataset? It will be really helpful." }, { "id": 1736498, "postDate": "2022-03-27T11:48:21.497Z", "content": "Hi Rakesh, can you please share the custom data and script used to generate it? Also, have you confirmed whether we can use it for the competition or not?
", "rawMarkdown": "Hi Rakesh, can you please share the custom data and script used to generate it? Also, have you confirmed whether we can use it for the competition or not?" } ], "comments": [ { "id": 1729733, "author_name": "Max333", "author_url": "", "post_date": "2022-03-20T12:45:50.240000", "content": "I don't think it's allowed. This is equivalent to hand labeling the training data.
", "votes": 3, "replies": [ { "id": 1729988, "author_name": "Asapanna Rakesh", "author_url": "", "post_date": "2022-03-20T19:21:46.803000", "content": "Can you please tell me where restrictions are mentioned? Weren't all restrictions removed?
", "votes": 0, "replies": [] }, { "id": 1729993, "author_name": "Drum Droll", "author_url": "", "post_date": "2022-03-20T19:33:33.100000", "content": "https://www.kaggle.com/c/ultra-mnist/rules
\n\"Submissions may not use or incorporate information from hand labeling or human prediction of the training or test data records.\"
", "votes": 0, "replies": [] }, { "id": 1729999, "author_name": "dmitry_lessy", "author_url": "", "post_date": "2022-03-20T19:44:08.053000", "content": "It is not hand labeled data. This data is generated, so I think it can be used in this competition.
", "votes": 1, "replies": [ { "id": 1730002, "author_name": "", "author_url": "", "post_date": "2022-03-20T19:48:51.247000", "content": "", "votes": 0, "replies": [] } ] }, { "id": 1741469, "author_name": "faraday", "author_url": "", "post_date": "2022-03-31T20:13:27.913000", "content": "Do you mean random image modifications such as translation, rotation etc. ?
\nOr do you mean something else?
@dkgupta90 can you please clear it out, whether this approach is acceptable or not for the GPU track
\nThank You
", "votes": 0, "replies": [] }, { "id": 1743850, "author_name": "Deepak Gupta", "author_url": "", "post_date": "2022-04-03T11:13:39.160000", "content": "As long as no hand-labelling is involved, it is good to go for the GPU track.
", "votes": 1, "replies": [] } ] }, { "id": 1741761, "author_name": "Shankar Mahadevan", "author_url": "", "post_date": "2022-04-01T05:57:55.277000", "content": "Can you please share this dataset? It will be really helpful.
", "votes": 0, "replies": [] }, { "id": 1736498, "author_name": "Samip Timalsena", "author_url": "", "post_date": "2022-03-27T11:48:21.497000", "content": "Hi Rakesh, can you please share the custom data and script used to generate it? Also, have you confirmed whether we can use it for the competition or not?
", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "1729733": "I don't think it's allowed. This is equivalent to hand labeling the training data.", "1729294": "If anyone is facing difficulty in generating your own custom dataset, I can share my script (will take 1-2 days to run depending on your compute power).\n\nOtherwise, I can share the generated data itself. Cz, I don't think someone without sufficient compute can generate it. Data has same size as current dataset but with all possible annotations you can think of - semantic segmentation, instance segmentation, classification, multilabel classification etc", "1741761": "Can you please share this dataset? It will be really helpful.", "1736498": "Hi Rakesh, can you please share the custom data and script used to generate it? Also, have you confirmed whether we can use it for the competition or not?" } }