![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F3924325%2Ff1e1bc3fdf8282a2cf31cf1523f93f6f%2Fdog1.JPG?generation=1596613010999841&alt=media\")
Ok that was a shakeup for the Kaggle history books 😳
\n\nBut for you who moved 300-600 places because you selected an approved model instead of your top-50-Yolo-model, have patience while the Kaggle team does its review.
\n\nI had a Yolo model that had a ranking top ~ 35, but I know like many of you that if Yolo had been ok to use, many of the top teams that worked with other models would have moved up the Yolo model's score significantly higher, so I will not focus and compare to much with my Yolo score.
\n\nSome of you have already mentioned previous competitions that have had similar rules, it can be a good benchmark because Kaggle themselves do not openly discuss how they work to find cheaters, which is a good way to get people to focus their thinking on the competition and not to find ways to cheat.
\n\nSo if you think nothing will happen now and the train has stoped, take a look at this facts:\nIn the \"Generative Dog Images\" Competition it took ~14 days to finalize the leaderboard and they reviewed more than 700 Kernels.
\n\n
There are certainly many reasons why so many chose to use their Yolo model as their final submission, but enough time has been spent on that analysis. I feel most sorry for the Kaggle team and those who have to sit for several weeks and manually review codes.
", "messages": [ { "id": 958884, "postDate": "2020-08-05T07:43:34.950Z", "content": "Ok that was a shakeup for the Kaggle history books 😳
\n\nBut for you who moved 300-600 places because you selected an approved model instead of your top-50-Yolo-model, have patience while the Kaggle team does its review.
\n\nI had a Yolo model that had a ranking top ~ 35, but I know like many of you that if Yolo had been ok to use, many of the top teams that worked with other models would have moved up the Yolo model's score significantly higher, so I will not focus and compare to much with my Yolo score.
\n\nSome of you have already mentioned previous competitions that have had similar rules, it can be a good benchmark because Kaggle themselves do not openly discuss how they work to find cheaters, which is a good way to get people to focus their thinking on the competition and not to find ways to cheat.
\n\nSo if you think nothing will happen now and the train has stoped, take a look at this facts:\nIn the \"Generative Dog Images\" Competition it took ~14 days to finalize the leaderboard and they reviewed more than 700 Kernels.
\n\n
There are certainly many reasons why so many chose to use their Yolo model as their final submission, but enough time has been spent on that analysis. I feel most sorry for the Kaggle team and those who have to sit for several weeks and manually review codes.
", "rawMarkdown": "Ok that was a shakeup for the Kaggle history books 😳 \n\nBut for you who moved 300-600 places because you selected an approved model instead of your top-50-Yolo-model, have patience while the Kaggle team does its review.\n\nI had a Yolo model that had a ranking top ~ 35, but I know like many of you that if Yolo had been ok to use, many of the top teams that worked with other models would have moved up the Yolo model's score significantly higher, so I will not focus and compare to much with my Yolo score.\n\nSome of you have already mentioned previous competitions that have had similar rules, it can be a good benchmark because Kaggle themselves do not openly discuss how they work to find cheaters, which is a good way to get people to focus their thinking on the competition and not to find ways to cheat.\n\nSo if you think nothing will happen now and the train has stoped, take a look at this facts:\nIn the \"Generative Dog Images\" Competition it took ~14 days to finalize the leaderboard and they reviewed more than 700 Kernels.\n\n\n\nThere are certainly many reasons why so many chose to use their Yolo model as their final submission, but enough time has been spent on that analysis. I feel most sorry for the Kaggle team and those who have to sit for several weeks and manually review codes.", "votes": 7 }, { "id": 958892, "postDate": "2020-08-05T07:47:59.603Z", "content": "HI! How many people used yolov5 you thing?
", "rawMarkdown": "HI! How many people used yolov5 you thing?", "replies": [ { "id": 958899, "postDate": "2020-08-05T07:54:35.903Z", "content": "Don't know but this analysis by matthewmasters \"Leaderboard of Compliant Submissions\"
\ncan may be indicative https://www.kaggle.com/c/global-wheat-detection/discussion/166793
Maybe they didnot use yolov5 and didnot participate Mathew's Board.
", "rawMarkdown": "Maybe they didnot use yolov5 and didnot participate Mathew's Board." } ] }, { "id": 958946, "postDate": "2020-08-05T08:28:35.407Z", "rawMarkdown": "", "isDeleted": true } ], "comments": [ { "id": 958892, "author_name": "Zekun", "author_url": "", "post_date": "2020-08-05T07:47:59.603000", "content": "HI! How many people used yolov5 you thing?
", "votes": 0, "replies": [ { "id": 958899, "author_name": "Kirderf", "author_url": "", "post_date": "2020-08-05T07:54:35.903000", "content": "Don't know but this analysis by matthewmasters \"Leaderboard of Compliant Submissions\"
\ncan may be indicative https://www.kaggle.com/c/global-wheat-detection/discussion/166793
Maybe they didnot use yolov5 and didnot participate Mathew's Board.
", "votes": 0, "replies": [] } ] }, { "id": 958946, "author_name": "", "author_url": "", "post_date": "2020-08-05T08:28:35.407000", "content": "", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "958884": "Ok that was a shakeup for the Kaggle history books 😳 \n\nBut for you who moved 300-600 places because you selected an approved model instead of your top-50-Yolo-model, have patience while the Kaggle team does its review.\n\nI had a Yolo model that had a ranking top ~ 35, but I know like many of you that if Yolo had been ok to use, many of the top teams that worked with other models would have moved up the Yolo model's score significantly higher, so I will not focus and compare to much with my Yolo score.\n\nSome of you have already mentioned previous competitions that have had similar rules, it can be a good benchmark because Kaggle themselves do not openly discuss how they work to find cheaters, which is a good way to get people to focus their thinking on the competition and not to find ways to cheat.\n\nSo if you think nothing will happen now and the train has stoped, take a look at this facts:\nIn the \"Generative Dog Images\" Competition it took ~14 days to finalize the leaderboard and they reviewed more than 700 Kernels.\n\n\n\nThere are certainly many reasons why so many chose to use their Yolo model as their final submission, but enough time has been spent on that analysis. I feel most sorry for the Kaggle team and those who have to sit for several weeks and manually review codes.", "958892": "HI! How many people used yolov5 you thing?", "958946": "" } }