Hi everyone,
\nIn the data description, the contrails are expected to be 'visible' in 2 images; what does the term visible imply in terms of pixel values?
\nIn addition to this, in one of the cases, contrails are expected to appear suddenly. How does this happen in a real scenario without the contrails reaching the edge of the image?
You can read through the actual instructions that were given to the labellers at https://storage.googleapis.com/goes_contrails_dataset/20230419/Contrail_Detection_Dataset_Instruction.pdf. It provides examples for these phenomena. Pretty much any questions not answered in those instructions would have been left up to the subjective interpretation of the individual labellers.
", "rawMarkdown": "You can read through the actual instructions that were given to the labellers at https://storage.googleapis.com/goes_contrails_dataset/20230419/Contrail_Detection_Dataset_Instruction.pdf. It provides examples for these phenomena. Pretty much any questions not answered in those instructions would have been left up to the subjective interpretation of the individual labellers.", "replies": [ { "id": 2309048, "postDate": "2023-06-19T11:14:20.443Z", "content": "Thank you, this helps me understand the data better
", "rawMarkdown": "Thank you, this helps me understand the data better" } ] }, { "id": 2302872, "postDate": "2023-06-14T23:15:28.907Z", "content": "For your first question, refer to this post: https://www.kaggle.com/competitions/google-research-identify-contrails-reduce-global-warming/discussion/416597#2296989
\nFor your second, those are just indications for the labelers and not rules that respect the dataset 100% of the time. I think the formation of visible contrail is not an instant process after the plane has flew over the place, but it might be very fast to appear after some time.
Thank you, this helps me understand the process of how the data was labeled better.
", "rawMarkdown": "Thank you, this helps me understand the process of how the data was labeled better. " } ] }, { "id": 2301972, "postDate": "2023-06-14T09:01:29.067Z", "content": "Hi everyone,
\nIn the data description, the contrails are expected to be 'visible' in 2 images; what does the term visible imply in terms of pixel values?
\nIn addition to this, in one of the cases, contrails are expected to appear suddenly. How does this happen in a real scenario without the contrails reaching the edge of the image?
You can read through the actual instructions that were given to the labellers at https://storage.googleapis.com/goes_contrails_dataset/20230419/Contrail_Detection_Dataset_Instruction.pdf. It provides examples for these phenomena. Pretty much any questions not answered in those instructions would have been left up to the subjective interpretation of the individual labellers.
", "votes": 0, "replies": [ { "id": 2309048, "author_name": "Mad Hatter", "author_url": "", "post_date": "2023-06-19T11:14:20.443000", "content": "Thank you, this helps me understand the data better
", "votes": 0, "replies": [] } ] }, { "id": 2302872, "author_name": "JEANMPIA", "author_url": "", "post_date": "2023-06-14T23:15:28.907000", "content": "For your first question, refer to this post: https://www.kaggle.com/competitions/google-research-identify-contrails-reduce-global-warming/discussion/416597#2296989
\nFor your second, those are just indications for the labelers and not rules that respect the dataset 100% of the time. I think the formation of visible contrail is not an instant process after the plane has flew over the place, but it might be very fast to appear after some time.
Thank you, this helps me understand the process of how the data was labeled better.
", "votes": 0, "replies": [] } ] } ], "raw_markdown_by_id": { "2303832": "You can read through the actual instructions that were given to the labellers at https://storage.googleapis.com/goes_contrails_dataset/20230419/Contrail_Detection_Dataset_Instruction.pdf. It provides examples for these phenomena. Pretty much any questions not answered in those instructions would have been left up to the subjective interpretation of the individual labellers.", "2302872": "For your first question, refer to this post: https://www.kaggle.com/competitions/google-research-identify-contrails-reduce-global-warming/discussion/416597#2296989\nFor your second, those are just indications for the labelers and not rules that respect the dataset 100% of the time. I think the formation of visible contrail is not an instant process after the plane has flew over the place, but it might be very fast to appear after some time.", "2301972": "Hi everyone,\nIn the data description, the contrails are expected to be 'visible' in 2 images; what does the term visible imply in terms of pixel values? \nIn addition to this, in one of the cases, contrails are expected to appear suddenly. How does this happen in a real scenario without the contrails reaching the edge of the image?" } }