In case people haven't seen this thread from the organisers, the private leaderboard will be rerun with the updated private test set. According to the thread, the results should be out in a couple of days.
\n\nAlso if you did any re-labelling of training data the organisers would appreciate it if you are willing to share it with them. Finally, the thread also contains details of how and when to share paper submissions for the upcoming Computer Vision Problems in Plant Phenotyping workshop.
", "messages": [ { "id": 959637, "postDate": "2020-08-05T19:08:13.237Z", "content": "In case people haven't seen this thread from the organisers, the private leaderboard will be rerun with the updated private test set. According to the thread, the results should be out in a couple of days.
\n\nAlso if you did any re-labelling of training data the organisers would appreciate it if you are willing to share it with them. Finally, the thread also contains details of how and when to share paper submissions for the upcoming Computer Vision Problems in Plant Phenotyping workshop.
", "rawMarkdown": "In case people haven't seen this [thread](https://www.kaggle.com/c/global-wheat-detection/discussion/172589) from the organisers, the private leaderboard will be rerun with the updated private test set. According to the thread, the results should be out in a couple of days.\n\nAlso if you did any re-labelling of training data the organisers would appreciate it if you are willing to share it with them. Finally, the thread also contains details of how and when to share paper submissions for the upcoming Computer Vision Problems in Plant Phenotyping workshop.", "votes": 5 }, { "id": 959768, "postDate": "2020-08-05T21:54:16.823Z", "content": "My god... Why not just use the actual private test set first? Oh well, maybe the new test set can help us gain some of the 319 positions we lost :p
", "rawMarkdown": "My god... Why not just use the actual private test set first? Oh well, maybe the new test set can help us gain some of the 319 positions we lost :p", "votes": 1, "replies": [ { "id": 960029, "postDate": "2020-08-06T05:06:37.033Z", "content": "What do you mean by '319 positions we lost' ?
", "rawMarkdown": "What do you mean by '319 positions we lost' ?", "votes": 1 }, { "id": 960102, "postDate": "2020-08-06T06:21:55.160Z", "content": "Went from top 100 to about 430th...
", "rawMarkdown": "Went from top 100 to about 430th...", "votes": 1 }, { "id": 962560, "postDate": "2020-08-08T08:32:01.220Z", "content": "Normally how the score reacts in a shakeup is a good opportunity to learn, but, in this case YoloV5 entries are clouding the issue. Like you I am hoping for a good resolution to this. The fact the hosts are spending time working on it is a good sign but all we can do is be patient I guess.
\n\nIn our team we chose 2 final subs that were very different to each other (neither was YoloV5). One of them was a lot more successful in the shakeup than the other. One team mate describes what we did here.
", "rawMarkdown": "Normally how the score reacts in a shakeup is a good opportunity to learn, but, in this case YoloV5 entries are clouding the issue. Like you I am hoping for a good resolution to this. The fact the hosts are spending time working on it is a good sign but all we can do is be patient I guess.\n\nIn our team we chose 2 final subs that were very different to each other (neither was YoloV5). One of them was a lot more successful in the shakeup than the other. One team mate describes what we did [here](https://www.kaggle.com/c/global-wheat-detection/discussion/172416)." } ] }, { "id": 962556, "postDate": "2020-08-08T08:26:20.957Z", "content": "The thread has been updated and the hosts are working on the finalized leaderboard but it will take a few more days.
", "rawMarkdown": "The [thread](https://www.kaggle.com/c/global-wheat-detection/discussion/172589#962179) has been updated and the hosts are working on the finalized leaderboard but it will take a few more days." }, { "id": 959715, "postDate": "2020-08-05T20:54:10.037Z", "rawMarkdown": "", "votes": -1, "isDeleted": true, "replies": [ { "id": 959754, "postDate": "2020-08-05T21:35:19.790Z", "content": "My interpretation of this thread is that yes we should expect a new shakeup because running the models against updated labels should create different scores. How different the updated labels are to the originals I do not know so I am not in a position to predict the scale of this second shake up. If someone else has more insight into this please share.
", "rawMarkdown": "My interpretation of this thread is that yes we should expect a new shakeup because running the models against updated labels should create different scores. How different the updated labels are to the originals I do not know so I am not in a position to predict the scale of this second shake up. If someone else has more insight into this please share.", "votes": 1 } ] } ], "comments": [ { "id": 959768, "author_name": "Stanley Zheng", "author_url": "", "post_date": "2020-08-05T21:54:16.823000", "content": "My god... Why not just use the actual private test set first? Oh well, maybe the new test set can help us gain some of the 319 positions we lost :p
", "votes": 1, "replies": [ { "id": 960029, "author_name": "bendang", "author_url": "", "post_date": "2020-08-06T05:06:37.033000", "content": "What do you mean by '319 positions we lost' ?
", "votes": 1, "replies": [] }, { "id": 960102, "author_name": "Stanley Zheng", "author_url": "", "post_date": "2020-08-06T06:21:55.160000", "content": "Went from top 100 to about 430th...
", "votes": 1, "replies": [] }, { "id": 962560, "author_name": "MichaelP", "author_url": "", "post_date": "2020-08-08T08:32:01.220000", "content": "Normally how the score reacts in a shakeup is a good opportunity to learn, but, in this case YoloV5 entries are clouding the issue. Like you I am hoping for a good resolution to this. The fact the hosts are spending time working on it is a good sign but all we can do is be patient I guess.
\n\nIn our team we chose 2 final subs that were very different to each other (neither was YoloV5). One of them was a lot more successful in the shakeup than the other. One team mate describes what we did here.
", "votes": 0, "replies": [] } ] }, { "id": 962556, "author_name": "MichaelP", "author_url": "", "post_date": "2020-08-08T08:26:20.957000", "content": "The thread has been updated and the hosts are working on the finalized leaderboard but it will take a few more days.
", "votes": 0, "replies": [] }, { "id": 959715, "author_name": "", "author_url": "", "post_date": "2020-08-05T20:54:10.037000", "content": "", "votes": -1, "replies": [ { "id": 959754, "author_name": "MichaelP", "author_url": "", "post_date": "2020-08-05T21:35:19.790000", "content": "My interpretation of this thread is that yes we should expect a new shakeup because running the models against updated labels should create different scores. How different the updated labels are to the originals I do not know so I am not in a position to predict the scale of this second shake up. If someone else has more insight into this please share.
", "votes": 1, "replies": [] } ] } ], "raw_markdown_by_id": { "959637": "In case people haven't seen this [thread](https://www.kaggle.com/c/global-wheat-detection/discussion/172589) from the organisers, the private leaderboard will be rerun with the updated private test set. According to the thread, the results should be out in a couple of days.\n\nAlso if you did any re-labelling of training data the organisers would appreciate it if you are willing to share it with them. Finally, the thread also contains details of how and when to share paper submissions for the upcoming Computer Vision Problems in Plant Phenotyping workshop.", "959768": "My god... Why not just use the actual private test set first? Oh well, maybe the new test set can help us gain some of the 319 positions we lost :p", "962556": "The [thread](https://www.kaggle.com/c/global-wheat-detection/discussion/172589#962179) has been updated and the hosts are working on the finalized leaderboard but it will take a few more days.", "959715": "" } }