We are pleased to pre-announce that we plan to release supplementary annotation of all 357 pasture images in the training dataset of this competition in early January 2026. The annotated dataset will include pixel-level semantic segmentation for each image, with masks for the following four classes:
\nPlease watch this topic in early January 2026.
\n-----------------------------------------Follow up-----------------------------------------------
\nAfter carefully considering your feedback, we have decided not to release the additional data annotations until the competition concludes. These annotations will be made available after the competition ends and may potentially be used for a follow-up, non-monetary competition, with the goal of further advancing model utility. All data and annotations will remain unchanged until the competition concludes. Thank you for all your valuable input and suggestions.
\n@addisonhoward @forcewithme @mushfirat @pcjimmmy @ravi20076 @thalro @sweetyheehee
", "messages": [ { "id": "3377475", "postDate": "12/16/2025 11:59:44", "content": "We are pleased to pre-announce that we plan to release supplementary annotation of all 357 pasture images in the training dataset of this competition in early January 2026. The annotated dataset will include pixel-level semantic segmentation for each image, with masks for the following four classes:
\nPlease watch this topic in early January 2026.
\n-----------------------------------------Follow up-----------------------------------------------
\nAfter carefully considering your feedback, we have decided not to release the additional data annotations until the competition concludes. These annotations will be made available after the competition ends and may potentially be used for a follow-up, non-monetary competition, with the goal of further advancing model utility. All data and annotations will remain unchanged until the competition concludes. Thank you for all your valuable input and suggestions.
\n@addisonhoward @forcewithme @mushfirat @pcjimmmy @ravi20076 @thalro @sweetyheehee
", "rawMarkdown": "We are pleased to pre-announce that we plan to release supplementary annotation of all 357 pasture images in the training dataset of this competition in early January 2026. The annotated dataset will include **pixel-level semantic segmentation** for each image, with masks for the following four classes:\n\n1. Background\n2. Clover\n3. Green matter\n4. Dead matter\n\nPlease watch this topic in early January 2026.\n\n-----------------------------------------Follow up-----------------------------------------------\n\nAfter carefully considering your feedback, we have decided not to release the additional data annotations until the competition concludes. These annotations will be made available after the competition ends and may potentially be used for a follow-up, non-monetary competition, with the goal of further advancing model utility. **All data and annotations will remain unchanged until the competition concludes.** Thank you for all your valuable input and suggestions.\n\n@addisonhoward @forcewithme @mushfirat @pcjimmmy @ravi20076 @thalro @sweetyheehee", "votes": null }, { "id": "3377482", "postDate": "12/16/2025 12:23:06", "content": "Hi @qiyuliaocsiro ,
\nI must strongly protest this decision, as it will fundamentally change the competition into a completely different task. With or without semantic segmentation labels, our approaches and the models we have trained are entirely different. The competition ends at the end of January. Releasing this dataset in early January will turn all our efforts over the past two months into a joke.
These semantic annotation labels could have been released before the competition started, or after it ends—but they absolutely must not be released during the competition, especially not in the final month.
I understand the intention to obtain the best possible solutions, but this is a matter of fairness in running the competition. Kaggle is an authoritative platform, and part of its authority comes from its commitment to fairness. Releasing mask labels in the final month— would completely overturn the competition landscape — severely compromises that fairness.
Hi , @addisonhoward , Please check this issues. 🥹
\nHi @qiyuliaocsiro , Thank you for your reply. I truly appreciate it! I look forward to further exploring this challenge with additional information after the competition!
", "rawMarkdown": "Hi @qiyuliaocsiro ,\n- **I must strongly protest this decision**, as it will fundamentally **change the competition into a completely different task**. With or without semantic segmentation labels, our approaches and the models we have trained are entirely different. The competition ends at the end of January. Releasing this dataset in early January will **turn all our efforts over the past two months into a joke**.\n\n- These semantic annotation labels could have been released before the competition started, or after it ends—but **they absolutely must not be released during the competition**, especially not in the final month.\n\n- I understand the intention to obtain the best possible solutions, but this is a matter of **fairness in running the competition**. Kaggle is an authoritative platform, and part of its authority comes from its commitment to fairness. Releasing mask labels in the final month— **would completely overturn the competition landscape** — severely **compromises that fairness**.\n\nHi , @addisonhoward , Please check this issues. 🥹\n\n-----------------------------------------------------------------------------------\nHi @qiyuliaocsiro , Thank you for your reply. I truly appreciate it! I look forward to further exploring this challenge with additional information after the competition!", "votes": null }, { "id": "3377521", "postDate": "12/16/2025 14:03:05", "content": "\n\nPlease watch this topic in early January 2026.
\n
I would have definitely supported this decision if it had been made early in the competition, but even now it’s a bit late, let alone waiting until next January. There’s hardly any time left.
", "rawMarkdown": ">Please watch this topic in early January 2026.\n\nI would have definitely supported this decision if it had been made early in the competition, but even now it’s a bit late, let alone waiting until next January. There’s hardly any time left.", "votes": null }, { "id": "3377536", "postDate": "12/16/2025 14:25:05", "content": "Hey all,
\nWe've been in discussion with the host on this - we'll report back with any decisions or changes made
", "rawMarkdown": "Hey all,\n\nWe've been in discussion with the host on this - we'll report back with any decisions or changes made", "votes": null }, { "id": "3377612", "postDate": "12/16/2025 16:27:15", "content": "We're 6 weeks into this competition, and 6 more weeks left.
\nIf you release those supplementary data just 4 weeks before the end, then it doesn't make any sense to put more effort into our current solutions because they'll become obsolete just 2 weeks later.
\nI don't support this decision but if you must release this supplementary data, please do it at least within the next few days so that we have enough time to incorporate this new data to our existing strategy. Otherwise, if you delay and release it just 3/4 weeks before the end, we won't have enough time to utilize the data to its maximum potential and the best solution won't be the best possible solution. That's surely not the outcome anybody would want.
", "rawMarkdown": "We're 6 weeks into this competition, and 6 more weeks left.\n\nIf you release those supplementary data just 4 weeks before the end, then it doesn't make any sense to put more effort into our current solutions because they'll become obsolete just 2 weeks later. \n\nI don't support this decision but if you **must** release this supplementary data, please do it at least within the next few days so that we have enough time to incorporate this new data to our existing strategy. Otherwise, if you delay and release it just 3/4 weeks before the end, we won't have enough time to utilize the data to its maximum potential and the best solution won't be the best *possible* solution. That's surely not the outcome anybody would want.", "votes": null }, { "id": "3377617", "postDate": "12/16/2025 16:30:26", "content": "I think it is already too late for additional data @mushfirat \nPerhaps they could do a similar competition later next year and release all the data together.
", "rawMarkdown": "I think it is already too late for additional data @mushfirat \nPerhaps they could do a similar competition later next year and release all the data together.", "votes": null }, { "id": "3377621", "postDate": "12/16/2025 16:35:49", "content": "I agree @ravi20076
\nReleasing this data TODAY would indeed be too late.
\nThis decision will be unfair to all participants who worked tirelessly for the last 6 weeks towards the best possible solution based on such a small dataset.
\nAnd it would be even more unfair if this data is to be released 2 weeks later, instead of today.
", "rawMarkdown": "I agree @ravi20076 \n\nReleasing this data TODAY would indeed be too late.\n\nThis decision will be unfair to all participants who worked tirelessly for the last 6 weeks towards the best *possible* solution based on such a small dataset.\n\nAnd it would be **even more unfair** if this data is to be released 2 weeks later, instead of today.", "votes": null }, { "id": "3377640", "postDate": "12/16/2025 17:46:29", "content": "Must admit I am ready to quit today. A huge sharp turn in the competition that makes such little sense close to the end.
\nIts been a real chore trying to make hay out of this small dataset with its apparent errors and issues. But chores can be fun.
\nStart a follow-up competition soon with this added data please, but don't mess with this one.
", "rawMarkdown": "Must admit I am ready to quit today. A huge sharp turn in the competition that makes such little sense close to the end.\n\nIts been a real chore trying to make hay out of this small dataset with its apparent errors and issues. But chores can be fun. \n\nStart a follow-up competition soon with this added data please, but don't mess with this one.", "votes": null }, { "id": "3377698", "postDate": "12/16/2025 19:59:10", "content": "Hi, I would also like to voice the concern that this would invalidate much of the work competitors have put in so far. I am not on track for a gold medal, but having invested a significant amount of effort into finding a good solution to the current problem, I would feel cheated if the game were fundamentally changed just as we are entering the final stretch.
", "rawMarkdown": "Hi, I would also like to voice the concern that this would invalidate much of the work competitors have put in so far. I am not on track for a gold medal, but having invested a significant amount of effort into finding a good solution to the current problem, I would feel cheated if the game were fundamentally changed just as we are entering the final stretch.", "votes": null }, { "id": "3377705", "postDate": "12/16/2025 20:05:16", "content": "@thalro in such a small dataset, nothing can be said
", "rawMarkdown": "thalro in such a small dataset, nothing can be said", "votes": null }, { "id": "3377710", "postDate": "12/16/2025 20:19:28", "content": "Yes, true. I am expecting a lot of movement when the private leaderboard is released. I decided to take part in this competition because it is relatively simple in terms of setup and more about controlling overfitting than fancy techniques. Changing the nature of the problem so close to the end would pretty much make everything I did so far worthless.
", "rawMarkdown": "Yes, true. I am expecting a lot of movement when the private leaderboard is released. I decided to take part in this competition because it is relatively simple in terms of setup and more about controlling overfitting than fancy techniques. Changing the nature of the problem so close to the end would pretty much make everything I did so far worthless.", "votes": null }, { "id": "3387707", "postDate": "01/07/2026 14:16:41", "content": "Hi may I ask which annotation tool you use :)
", "rawMarkdown": "Hi may I ask which annotation tool you use :)", "votes": null } ], "comments": [ { "id": 3377482, "author_name": "forcewithme", "author_url": "", "post_date": "12/16/2025 12:23:06", "content": "Hi @qiyuliaocsiro ,
\nI must strongly protest this decision, as it will fundamentally change the competition into a completely different task. With or without semantic segmentation labels, our approaches and the models we have trained are entirely different. The competition ends at the end of January. Releasing this dataset in early January will turn all our efforts over the past two months into a joke.
These semantic annotation labels could have been released before the competition started, or after it ends—but they absolutely must not be released during the competition, especially not in the final month.
I understand the intention to obtain the best possible solutions, but this is a matter of fairness in running the competition. Kaggle is an authoritative platform, and part of its authority comes from its commitment to fairness. Releasing mask labels in the final month— would completely overturn the competition landscape — severely compromises that fairness.
Hi , @addisonhoward , Please check this issues. 🥹
\nHi @qiyuliaocsiro , Thank you for your reply. I truly appreciate it! I look forward to further exploring this challenge with additional information after the competition!
", "votes": null, "replies": [ { "id": 3377521, "author_name": "sweetyheehee", "author_url": "", "post_date": "12/16/2025 14:03:05", "content": "\n\nPlease watch this topic in early January 2026.
\n
I would have definitely supported this decision if it had been made early in the competition, but even now it’s a bit late, let alone waiting until next January. There’s hardly any time left.
", "votes": null, "replies": [] }, { "id": 3377536, "author_name": "addisonhoward", "author_url": "", "post_date": "12/16/2025 14:25:05", "content": "Hey all,
\nWe've been in discussion with the host on this - we'll report back with any decisions or changes made
", "votes": null, "replies": [ { "id": 3377698, "author_name": "thalro", "author_url": "", "post_date": "12/16/2025 19:59:10", "content": "Hi, I would also like to voice the concern that this would invalidate much of the work competitors have put in so far. I am not on track for a gold medal, but having invested a significant amount of effort into finding a good solution to the current problem, I would feel cheated if the game were fundamentally changed just as we are entering the final stretch.
", "votes": null, "replies": [ { "id": 3377705, "author_name": "ravi20076", "author_url": "", "post_date": "12/16/2025 20:05:16", "content": "@thalro in such a small dataset, nothing can be said
", "votes": null, "replies": [ { "id": 3377710, "author_name": "thalro", "author_url": "", "post_date": "12/16/2025 20:19:28", "content": "Yes, true. I am expecting a lot of movement when the private leaderboard is released. I decided to take part in this competition because it is relatively simple in terms of setup and more about controlling overfitting than fancy techniques. Changing the nature of the problem so close to the end would pretty much make everything I did so far worthless.
", "votes": null, "replies": [] } ] } ] } ] }, { "id": 3387707, "author_name": "aliozanmemetoglu", "author_url": "", "post_date": "01/07/2026 14:16:41", "content": "Hi may I ask which annotation tool you use :)
", "votes": null, "replies": [] } ] }, { "id": 3377612, "author_name": "mushfirat", "author_url": "", "post_date": "12/16/2025 16:27:15", "content": "We're 6 weeks into this competition, and 6 more weeks left.
\nIf you release those supplementary data just 4 weeks before the end, then it doesn't make any sense to put more effort into our current solutions because they'll become obsolete just 2 weeks later.
\nI don't support this decision but if you must release this supplementary data, please do it at least within the next few days so that we have enough time to incorporate this new data to our existing strategy. Otherwise, if you delay and release it just 3/4 weeks before the end, we won't have enough time to utilize the data to its maximum potential and the best solution won't be the best possible solution. That's surely not the outcome anybody would want.
", "votes": null, "replies": [ { "id": 3377617, "author_name": "ravi20076", "author_url": "", "post_date": "12/16/2025 16:30:26", "content": "I think it is already too late for additional data @mushfirat \nPerhaps they could do a similar competition later next year and release all the data together.
", "votes": null, "replies": [ { "id": 3377621, "author_name": "mushfirat", "author_url": "", "post_date": "12/16/2025 16:35:49", "content": "I agree @ravi20076
\nReleasing this data TODAY would indeed be too late.
\nThis decision will be unfair to all participants who worked tirelessly for the last 6 weeks towards the best possible solution based on such a small dataset.
\nAnd it would be even more unfair if this data is to be released 2 weeks later, instead of today.
", "votes": null, "replies": [] } ] } ] }, { "id": 3377640, "author_name": "pcjimmmy", "author_url": "", "post_date": "12/16/2025 17:46:29", "content": "Must admit I am ready to quit today. A huge sharp turn in the competition that makes such little sense close to the end.
\nIts been a real chore trying to make hay out of this small dataset with its apparent errors and issues. But chores can be fun.
\nStart a follow-up competition soon with this added data please, but don't mess with this one.
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "3377475": "We are pleased to pre-announce that we plan to release supplementary annotation of all 357 pasture images in the training dataset of this competition in early January 2026. The annotated dataset will include **pixel-level semantic segmentation** for each image, with masks for the following four classes:\n\n1. Background\n2. Clover\n3. Green matter\n4. Dead matter\n\nPlease watch this topic in early January 2026.\n\n-----------------------------------------Follow up-----------------------------------------------\n\nAfter carefully considering your feedback, we have decided not to release the additional data annotations until the competition concludes. These annotations will be made available after the competition ends and may potentially be used for a follow-up, non-monetary competition, with the goal of further advancing model utility. **All data and annotations will remain unchanged until the competition concludes.** Thank you for all your valuable input and suggestions.\n\n@addisonhoward @forcewithme @mushfirat @pcjimmmy @ravi20076 @thalro @sweetyheehee", "3377482": "Hi @qiyuliaocsiro ,\n- **I must strongly protest this decision**, as it will fundamentally **change the competition into a completely different task**. With or without semantic segmentation labels, our approaches and the models we have trained are entirely different. The competition ends at the end of January. Releasing this dataset in early January will **turn all our efforts over the past two months into a joke**.\n\n- These semantic annotation labels could have been released before the competition started, or after it ends—but **they absolutely must not be released during the competition**, especially not in the final month.\n\n- I understand the intention to obtain the best possible solutions, but this is a matter of **fairness in running the competition**. Kaggle is an authoritative platform, and part of its authority comes from its commitment to fairness. Releasing mask labels in the final month— **would completely overturn the competition landscape** — severely **compromises that fairness**.\n\nHi , @addisonhoward , Please check this issues. 🥹\n\n-----------------------------------------------------------------------------------\nHi @qiyuliaocsiro , Thank you for your reply. I truly appreciate it! I look forward to further exploring this challenge with additional information after the competition!", "3377521": ">Please watch this topic in early January 2026.\n\nI would have definitely supported this decision if it had been made early in the competition, but even now it’s a bit late, let alone waiting until next January. There’s hardly any time left.", "3377536": "Hey all,\n\nWe've been in discussion with the host on this - we'll report back with any decisions or changes made", "3377612": "We're 6 weeks into this competition, and 6 more weeks left.\n\nIf you release those supplementary data just 4 weeks before the end, then it doesn't make any sense to put more effort into our current solutions because they'll become obsolete just 2 weeks later. \n\nI don't support this decision but if you **must** release this supplementary data, please do it at least within the next few days so that we have enough time to incorporate this new data to our existing strategy. Otherwise, if you delay and release it just 3/4 weeks before the end, we won't have enough time to utilize the data to its maximum potential and the best solution won't be the best *possible* solution. That's surely not the outcome anybody would want.", "3377617": "I think it is already too late for additional data @mushfirat \nPerhaps they could do a similar competition later next year and release all the data together.", "3377621": "I agree @ravi20076 \n\nReleasing this data TODAY would indeed be too late.\n\nThis decision will be unfair to all participants who worked tirelessly for the last 6 weeks towards the best *possible* solution based on such a small dataset.\n\nAnd it would be **even more unfair** if this data is to be released 2 weeks later, instead of today.", "3377640": "Must admit I am ready to quit today. A huge sharp turn in the competition that makes such little sense close to the end.\n\nIts been a real chore trying to make hay out of this small dataset with its apparent errors and issues. But chores can be fun. \n\nStart a follow-up competition soon with this added data please, but don't mess with this one.", "3377698": "Hi, I would also like to voice the concern that this would invalidate much of the work competitors have put in so far. I am not on track for a gold medal, but having invested a significant amount of effort into finding a good solution to the current problem, I would feel cheated if the game were fundamentally changed just as we are entering the final stretch.", "3377705": "thalro in such a small dataset, nothing can be said", "3377710": "Yes, true. I am expecting a lot of movement when the private leaderboard is released. I decided to take part in this competition because it is relatively simple in terms of setup and more about controlling overfitting than fancy techniques. Changing the nature of the problem so close to the end would pretty much make everything I did so far worthless.", "3387707": "Hi may I ask which annotation tool you use :)" }, "source": "meta" }