![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F10704200%2F18109d387a489550ae94ed30704d1692%2Fdiagnostic.png?generation=1735168919369498&alt=media\")
Thought I would mention a tool I created that you can use to generate diagnostic output from your model like that shown below. True positives, false positives, and false negatives are green, yellow, and red respectively.
\nNotebook link is here:
\nhttps://www.kaggle.com/code/davidlist/diagnostic-visualization-tool
Enjoy!
\n
Thought I would mention a tool I created that you can use to generate diagnostic output from your model like that shown below. True positives, false positives, and false negatives are green, yellow, and red respectively.
\nNotebook link is here:
\nhttps://www.kaggle.com/code/davidlist/diagnostic-visualization-tool
Enjoy!
\n
It might be useful to note that the top and bottom z-edges of the tomogram are typically the most affected by artifacts during the imaging process and contain little meaningful information. If I recall correctly (others can confirm), this is because these regions are directly exposed to the environment. It is typical practice to ignore the 10-20 slices at the top and the bottom.
\nAnother note is that the circular line at the right edge is called a \"carbon edge\" and is artifact due to the carbon film substrate. The carbon edge tends to give lots of false positives in automated particle picking algorithms.
\n@uermel or @rezaparaan are a lot more knowledgeable about this, and might be able to add more info.
", "rawMarkdown": "It might be useful to note that the top and bottom z-edges of the tomogram are typically the most affected by artifacts during the imaging process and contain little meaningful information. If I recall correctly (others can confirm), this is because these regions are directly exposed to the environment. It is typical practice to ignore the 10-20 slices at the top and the bottom.\n\nAnother note is that the circular line at the right edge is called a \"carbon edge\" and is artifact due to the carbon film substrate. The carbon edge tends to give lots of false positives in automated particle picking algorithms. \n\n@uermel or @rezaparaan are a lot more knowledgeable about this, and might be able to add more info.", "votes": 8, "isDeleted": true, "replies": [ { "id": 3046178, "postDate": "2024-11-15T07:32:29.043Z", "content": "Thank you @saugatkandel! That's very useful info!
", "rawMarkdown": "Thank you @saugatkandel! That's very useful info!" } ] } ], "comments": [ { "id": 3046123, "author_name": "", "author_url": "", "post_date": "2024-11-15T06:23:25", "content": "It might be useful to note that the top and bottom z-edges of the tomogram are typically the most affected by artifacts during the imaging process and contain little meaningful information. If I recall correctly (others can confirm), this is because these regions are directly exposed to the environment. It is typical practice to ignore the 10-20 slices at the top and the bottom.
\nAnother note is that the circular line at the right edge is called a \"carbon edge\" and is artifact due to the carbon film substrate. The carbon edge tends to give lots of false positives in automated particle picking algorithms.
\n@uermel or @rezaparaan are a lot more knowledgeable about this, and might be able to add more info.
", "votes": 8, "replies": [ { "id": 3046178, "author_name": "David List", "author_url": "", "post_date": "2024-11-15T07:32:29.043000", "content": "Thank you @saugatkandel! That's very useful info!
", "votes": 0, "replies": [] } ] } ], "raw_markdown_by_id": { "3046081": "Thought I would mention a tool I created that you can use to generate diagnostic output from your model like that shown below. True positives, false positives, and false negatives are green, yellow, and red respectively.\n\nNotebook link is here:\n[https://www.kaggle.com/code/davidlist/diagnostic-visualization-tool](https://www.kaggle.com/code/davidlist/diagnostic-visualization-tool)\n\nEnjoy!\n\n", "3046123": "It might be useful to note that the top and bottom z-edges of the tomogram are typically the most affected by artifacts during the imaging process and contain little meaningful information. If I recall correctly (others can confirm), this is because these regions are directly exposed to the environment. It is typical practice to ignore the 10-20 slices at the top and the bottom.\n\nAnother note is that the circular line at the right edge is called a \"carbon edge\" and is artifact due to the carbon film substrate. The carbon edge tends to give lots of false positives in automated particle picking algorithms. \n\n@uermel or @rezaparaan are a lot more knowledgeable about this, and might be able to add more info." } }