Will it be possible if I create my own masks as my model training labels, even if I hand create label masks for that, will it be fine by the organizers or is it against the rules, just wanna be sure.
\n\nThanks!
", "messages": [ { "id": 837334, "postDate": "2020-05-07T17:40:06.983Z", "content": "For clarity you can use this approach only for training.
\n\nSection 5 of the rules under Competition Entry: \"Submissions may not use or incorporate information from hand labeling or human prediction of the validation dataset or test data records.\"
", "rawMarkdown": "For clarity you can use this approach only for training. \n\nSection 5 of the rules under Competition Entry: \"Submissions may not use or incorporate information from hand labeling or human prediction of the validation dataset or test data records.\" \n\n", "votes": 1, "replies": [ { "id": 837619, "postDate": "2020-05-07T22:41:39.280Z", "content": "Yeah, I will create the labels masks ONLY for training, thank you for the confirmation!
", "rawMarkdown": "Yeah, I will create the labels masks ONLY for training, thank you for the confirmation!" } ] }, { "id": 837700, "postDate": "2020-05-08T01:26:59.797Z", "content": "I don't understand your question. We are given the train data, we can do whatever we want with it. By the way, we don't have access to test data, so any hand label for test is impossible.
", "rawMarkdown": "I don't understand your question. We are given the train data, we can do whatever we want with it. By the way, we don't have access to test data, so any hand label for test is impossible.", "replies": [ { "id": 837812, "postDate": "2020-05-08T04:10:46.140Z", "content": "I was trying to say to make my own labels for the training data, like my own masks by hand of the training data, then train with them. Anyways, I dont have any plans to continue in this competition as per now.
", "rawMarkdown": "I was trying to say to make my own labels for the training data, like my own masks by hand of the training data, then train with them. Anyways, I dont have any plans to continue in this competition as per now." }, { "id": 837818, "postDate": "2020-05-08T04:15:06.670Z", "content": "Sad to hear that. Still impressed by your kernels in Deepfake.
", "rawMarkdown": "Sad to hear that. Still impressed by your kernels in Deepfake.", "votes": 1 }, { "id": 837939, "postDate": "2020-05-08T06:56:05.223Z", "content": "Thanks!, well, if I return, I will contribute some amazing kernels and a lot of sneak peaks to high scores in discussions here too :)
", "rawMarkdown": "Thanks!, well, if I return, I will contribute some amazing kernels and a lot of sneak peaks to high scores in discussions here too :)" } ] }, { "id": 830914, "postDate": "2020-05-03T01:07:30.940Z", "content": "Will it be possible if I create my own masks as my model training labels, even if I hand create label masks for that, will it be fine by the organizers or is it against the rules, just wanna be sure.
\n\nThanks!
", "rawMarkdown": "Will it be possible if I create my own masks as my model training labels, even if I hand create label masks for that, will it be fine by the organizers or is it against the rules, just wanna be sure.\n\nThanks!" } ], "comments": [ { "id": 837334, "author_name": "Maggie", "author_url": "", "post_date": "2020-05-07T17:40:06.983000", "content": "For clarity you can use this approach only for training.
\n\nSection 5 of the rules under Competition Entry: \"Submissions may not use or incorporate information from hand labeling or human prediction of the validation dataset or test data records.\"
", "votes": 1, "replies": [ { "id": 837619, "author_name": "Harshit Sheoran", "author_url": "", "post_date": "2020-05-07T22:41:39.280000", "content": "Yeah, I will create the labels masks ONLY for training, thank you for the confirmation!
", "votes": 0, "replies": [] } ] }, { "id": 837700, "author_name": "Kha Vo", "author_url": "", "post_date": "2020-05-08T01:26:59.797000", "content": "I don't understand your question. We are given the train data, we can do whatever we want with it. By the way, we don't have access to test data, so any hand label for test is impossible.
", "votes": 0, "replies": [ { "id": 837812, "author_name": "Harshit Sheoran", "author_url": "", "post_date": "2020-05-08T04:10:46.140000", "content": "I was trying to say to make my own labels for the training data, like my own masks by hand of the training data, then train with them. Anyways, I dont have any plans to continue in this competition as per now.
", "votes": 0, "replies": [] }, { "id": 837818, "author_name": "Kha Vo", "author_url": "", "post_date": "2020-05-08T04:15:06.670000", "content": "Sad to hear that. Still impressed by your kernels in Deepfake.
", "votes": 1, "replies": [] }, { "id": 837939, "author_name": "Harshit Sheoran", "author_url": "", "post_date": "2020-05-08T06:56:05.223000", "content": "Thanks!, well, if I return, I will contribute some amazing kernels and a lot of sneak peaks to high scores in discussions here too :)
", "votes": 0, "replies": [] } ] } ], "raw_markdown_by_id": { "837334": "For clarity you can use this approach only for training. \n\nSection 5 of the rules under Competition Entry: \"Submissions may not use or incorporate information from hand labeling or human prediction of the validation dataset or test data records.\" \n\n", "837700": "I don't understand your question. We are given the train data, we can do whatever we want with it. By the way, we don't have access to test data, so any hand label for test is impossible.", "830914": "Will it be possible if I create my own masks as my model training labels, even if I hand create label masks for that, will it be fine by the organizers or is it against the rules, just wanna be sure.\n\nThanks!" } }