A lot of the data is rotated or scaled differently and most opensource algorithms are not nearly as sophisticated as certain patented ones. The ones that are freely available use simple feature generators and these are not invariant to rotation and scale and probably won't deliver as good results.
\n\nMany patents exclude commercial use, but explicitly allow academic and research use. Are we allowed to use such patented algorithms for this competition, because the current submission rules explicitly prohibit them.
", "messages": [ { "id": "117717", "postDate": "04/30/2016 14:25:58", "content": "A lot of the data is rotated or scaled differently and most opensource algorithms are not nearly as sophisticated as certain patented ones. The ones that are freely available use simple feature generators and these are not invariant to rotation and scale and probably won't deliver as good results.
\n\nMany patents exclude commercial use, but explicitly allow academic and research use. Are we allowed to use such patented algorithms for this competition, because the current submission rules explicitly prohibit them.
", "rawMarkdown": "A lot of the data is rotated or scaled differently and most opensource algorithms are not nearly as sophisticated as certain patented ones. The ones that are freely available use simple feature generators and these are not invariant to rotation and scale and probably won't deliver as good results.\r\n\r\nMany patents exclude commercial use, but explicitly allow academic and research use. Are we allowed to use such patented algorithms for this competition, because the current submission rules explicitly prohibit them.", "votes": null }, { "id": "117773", "postDate": "04/30/2016 20:45:54", "content": "Dunno if they'd count as closed source, but if so it wouldn't be prize eligible.
\n\nOTOH patents are probably broken all the time without knowing it, sigh.
", "rawMarkdown": "Dunno if they'd count as closed source, but if so it wouldn't be prize eligible.\r\n\r\nOTOH patents are probably broken all the time without knowing it, sigh.", "votes": null }, { "id": "117778", "postDate": "04/30/2016 21:22:44", "content": "The rule of note is:
\n\nFor one of these algorithms I have downloaded code and it's composed of three stages in total. One of those stages is patented. The patent excludes commercial use only, but allows academic and research use.
\n\nTo clarify further, I'm not related to any of this patent and have no links, dependencies or relationships with any of the legal persons involved in that patent.
", "rawMarkdown": "The rule of note is:\r\n\r\n* does not contain confidential information or trade secrets and is not the subject of a registered patent or pending patent application;\r\n\r\nFor one of these algorithms I have downloaded code and it's composed of three stages in total. One of those stages is patented. The patent excludes commercial use only, but allows academic and research use.\r\n\r\nTo clarify further, I'm not related to any of this patent and have no links, dependencies or relationships with any of the legal persons involved in that patent.", "votes": null }, { "id": "119444", "postDate": "05/10/2016 10:49:03", "content": "I think people should be careful not to rule themselves out of prizes due to this. SIFT would be potentially very useful and there are a few open source implementations but the algorithm itself is still under patent.
", "rawMarkdown": "I think people should be careful not to rule themselves out of prizes due to this. SIFT would be potentially very useful and there are a few open source implementations but the algorithm itself is still under patent.", "votes": null }, { "id": "119468", "postDate": "05/10/2016 13:16:42", "content": "[quote=hairykrishna;119444]\nI think people should be careful not to rule themselves out of prizes due to this. SIFT would be potentially very useful and there are a few open source implementations but the algorithm itself is still under patent.\n[/quote]
\n\nIt turned out not to be a problem at all, change one line of code and you can use AKAZE instead of SIFT, and AKAZE is not patented:\nhttps://www.kaggle.com/nigelcarpenter/draper-satellite-image-chronology/akaze-stitching/code
", "rawMarkdown": "[quote=hairykrishna;119444]\r\nI think people should be careful not to rule themselves out of prizes due to this. SIFT would be potentially very useful and there are a few open source implementations but the algorithm itself is still under patent.\r\n[/quote]\r\n\r\nIt turned out not to be a problem at all, change one line of code and you can use AKAZE instead of SIFT, and AKAZE is not patented:\r\nhttps://www.kaggle.com/nigelcarpenter/draper-satellite-image-chronology/akaze-stitching/code", "votes": null }, { "id": "119483", "postDate": "05/10/2016 16:58:49", "content": "In addition to the above, I have two questions: \n1) Can we use MATLAB for prototype coding ?\n2) Is using OpenCV - given that its opensource, allowed?
", "rawMarkdown": "In addition to the above, I have two questions: \r\n1) Can we use MATLAB for prototype coding ?\r\n2) Is using OpenCV - given that its opensource, allowed?", "votes": null }, { "id": "119686", "postDate": "05/12/2016 08:44:15", "content": "As far as i know..SIFT is also a patented algorithm.. Can we use SIFT in this competition
", "rawMarkdown": "As far as i know..SIFT is also a patented algorithm.. Can we use SIFT in this competition", "votes": null }, { "id": "119721", "postDate": "05/12/2016 13:34:48", "content": "^ : Pretty sure not, but there are other similar keypoint detectors available: https://www.kaggle.com/nigelcarpenter/draper-satellite-image-chronology/akaze-stitching/code
", "rawMarkdown": "^ : Pretty sure not, but there are other similar keypoint detectors available: https://www.kaggle.com/nigelcarpenter/draper-satellite-image-chronology/akaze-stitching/code", "votes": null } ], "comments": [ { "id": 117773, "author_name": "happycube", "author_url": "", "post_date": "04/30/2016 20:45:54", "content": "Dunno if they'd count as closed source, but if so it wouldn't be prize eligible.
\n\nOTOH patents are probably broken all the time without knowing it, sigh.
", "votes": null, "replies": [] }, { "id": 117778, "author_name": "remap1", "author_url": "", "post_date": "04/30/2016 21:22:44", "content": "The rule of note is:
\n\nFor one of these algorithms I have downloaded code and it's composed of three stages in total. One of those stages is patented. The patent excludes commercial use only, but allows academic and research use.
\n\nTo clarify further, I'm not related to any of this patent and have no links, dependencies or relationships with any of the legal persons involved in that patent.
", "votes": null, "replies": [] }, { "id": 119444, "author_name": "hairykrishna", "author_url": "", "post_date": "05/10/2016 10:49:03", "content": "I think people should be careful not to rule themselves out of prizes due to this. SIFT would be potentially very useful and there are a few open source implementations but the algorithm itself is still under patent.
", "votes": null, "replies": [] }, { "id": 119468, "author_name": "itodorovic", "author_url": "", "post_date": "05/10/2016 13:16:42", "content": "[quote=hairykrishna;119444]\nI think people should be careful not to rule themselves out of prizes due to this. SIFT would be potentially very useful and there are a few open source implementations but the algorithm itself is still under patent.\n[/quote]
\n\nIt turned out not to be a problem at all, change one line of code and you can use AKAZE instead of SIFT, and AKAZE is not patented:\nhttps://www.kaggle.com/nigelcarpenter/draper-satellite-image-chronology/akaze-stitching/code
", "votes": null, "replies": [] }, { "id": 119483, "author_name": "aneesh3108", "author_url": "", "post_date": "05/10/2016 16:58:49", "content": "In addition to the above, I have two questions: \n1) Can we use MATLAB for prototype coding ?\n2) Is using OpenCV - given that its opensource, allowed?
", "votes": null, "replies": [] }, { "id": 119686, "author_name": "shahnawazakhtar", "author_url": "", "post_date": "05/12/2016 08:44:15", "content": "As far as i know..SIFT is also a patented algorithm.. Can we use SIFT in this competition
", "votes": null, "replies": [] }, { "id": 119721, "author_name": "aneesh3108", "author_url": "", "post_date": "05/12/2016 13:34:48", "content": "^ : Pretty sure not, but there are other similar keypoint detectors available: https://www.kaggle.com/nigelcarpenter/draper-satellite-image-chronology/akaze-stitching/code
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "117717": "A lot of the data is rotated or scaled differently and most opensource algorithms are not nearly as sophisticated as certain patented ones. The ones that are freely available use simple feature generators and these are not invariant to rotation and scale and probably won't deliver as good results.\r\n\r\nMany patents exclude commercial use, but explicitly allow academic and research use. Are we allowed to use such patented algorithms for this competition, because the current submission rules explicitly prohibit them.", "117773": "Dunno if they'd count as closed source, but if so it wouldn't be prize eligible.\r\n\r\nOTOH patents are probably broken all the time without knowing it, sigh.", "117778": "The rule of note is:\r\n\r\n* does not contain confidential information or trade secrets and is not the subject of a registered patent or pending patent application;\r\n\r\nFor one of these algorithms I have downloaded code and it's composed of three stages in total. One of those stages is patented. The patent excludes commercial use only, but allows academic and research use.\r\n\r\nTo clarify further, I'm not related to any of this patent and have no links, dependencies or relationships with any of the legal persons involved in that patent.", "119444": "I think people should be careful not to rule themselves out of prizes due to this. SIFT would be potentially very useful and there are a few open source implementations but the algorithm itself is still under patent.", "119468": "[quote=hairykrishna;119444]\r\nI think people should be careful not to rule themselves out of prizes due to this. SIFT would be potentially very useful and there are a few open source implementations but the algorithm itself is still under patent.\r\n[/quote]\r\n\r\nIt turned out not to be a problem at all, change one line of code and you can use AKAZE instead of SIFT, and AKAZE is not patented:\r\nhttps://www.kaggle.com/nigelcarpenter/draper-satellite-image-chronology/akaze-stitching/code", "119483": "In addition to the above, I have two questions: \r\n1) Can we use MATLAB for prototype coding ?\r\n2) Is using OpenCV - given that its opensource, allowed?", "119686": "As far as i know..SIFT is also a patented algorithm.. Can we use SIFT in this competition", "119721": "^ : Pretty sure not, but there are other similar keypoint detectors available: https://www.kaggle.com/nigelcarpenter/draper-satellite-image-chronology/akaze-stitching/code" }, "source": "meta" }