I have been having problems with only a few images that end up crashing my notebooks when I compile with the entire dataset. One example. I debugged it back to how I am stacking the model in the code, but I do not know why it is having this issue. I can continue to by-pass these images, but this one is the smaller class. Any thoughts on why this 3072 size image does not stack correctly and the subsequent one stacks as expected.
", "messages": [ { "id": 1286350, "postDate": "2021-04-27T21:04:23.613Z", "content": "As I posted on the notebook , the intensities on the red and blue images have been scaled down for some reason. Multiplying them by 255 does the trick. Better multiply the green on by 2 as well , the protein didn't seem visible if you just used the green image as is.
\nhttps://imgur.com/a/WF7ENAE
I analyzed both of the blue nucleus images in my post. The second one has a pixel range of 0 to 1 with average of 0.1 which is ideal for modeling. No scaling is necessary for a model. However, the first has a maximum of 0.1. The real problem I am seeing is that particular cell the cell segmentation program finds no cells. The pred_cell module already scales to 255 but that first image only has a max value of 133 after segmentator.pred_cell. I’ll attempt to normalize to 1 or 255 as appropriate in the individual images and report back. Not sure it’s the same issue I’m having with seeing a stacked cell in matplotlib. If I skip it then everything works fine but that’s one of the mitotic spindle images so I would like to have my models train on it.
", "rawMarkdown": " I analyzed both of the blue nucleus images in my post. The second one has a pixel range of 0 to 1 with average of 0.1 which is ideal for modeling. No scaling is necessary for a model. However, the first has a maximum of 0.1. The real problem I am seeing is that particular cell the cell segmentation program finds no cells. The pred_cell module already scales to 255 but that first image only has a max value of 133 after segmentator.pred_cell. I’ll attempt to normalize to 1 or 255 as appropriate in the individual images and report back. Not sure it’s the same issue I’m having with seeing a stacked cell in matplotlib. If I skip it then everything works fine but that’s one of the mitotic spindle images so I would like to have my models train on it.", "votes": 1 } ] }, { "id": 1286268, "postDate": "2021-04-27T18:40:34.400Z", "content": "I have been having problems with only a few images that end up crashing my notebooks when I compile with the entire dataset. One example. I debugged it back to how I am stacking the model in the code, but I do not know why it is having this issue. I can continue to by-pass these images, but this one is the smaller class. Any thoughts on why this 3072 size image does not stack correctly and the subsequent one stacks as expected.
", "rawMarkdown": "I have been having problems with only a few images that end up crashing my notebooks when I compile with the entire dataset. [One example](https://www.kaggle.com/akimball002/cell-does-not-stack-help?scriptVersionId=61143864). I debugged it back to how I am stacking the model in the code, but I do not know why it is having this issue. I can continue to by-pass these images, but this one is the smaller class. Any thoughts on why this 3072 size image does not stack correctly and the subsequent one stacks as expected.\n", "votes": 2 } ], "comments": [ { "id": 1286350, "author_name": "Satwik", "author_url": "", "post_date": "2021-04-27T21:04:23.613000", "content": "As I posted on the notebook , the intensities on the red and blue images have been scaled down for some reason. Multiplying them by 255 does the trick. Better multiply the green on by 2 as well , the protein didn't seem visible if you just used the green image as is.
\nhttps://imgur.com/a/WF7ENAE
I analyzed both of the blue nucleus images in my post. The second one has a pixel range of 0 to 1 with average of 0.1 which is ideal for modeling. No scaling is necessary for a model. However, the first has a maximum of 0.1. The real problem I am seeing is that particular cell the cell segmentation program finds no cells. The pred_cell module already scales to 255 but that first image only has a max value of 133 after segmentator.pred_cell. I’ll attempt to normalize to 1 or 255 as appropriate in the individual images and report back. Not sure it’s the same issue I’m having with seeing a stacked cell in matplotlib. If I skip it then everything works fine but that’s one of the mitotic spindle images so I would like to have my models train on it.
", "votes": 1, "replies": [] } ] } ], "raw_markdown_by_id": { "1286350": "As I posted on the notebook , the intensities on the red and blue images have been scaled down for some reason. Multiplying them by 255 does the trick. Better multiply the green on by 2 as well , the protein didn't seem visible if you just used the green image as is. \nhttps://imgur.com/a/WF7ENAE", "1286268": "I have been having problems with only a few images that end up crashing my notebooks when I compile with the entire dataset. [One example](https://www.kaggle.com/akimball002/cell-does-not-stack-help?scriptVersionId=61143864). I debugged it back to how I am stacking the model in the code, but I do not know why it is having this issue. I can continue to by-pass these images, but this one is the smaller class. Any thoughts on why this 3072 size image does not stack correctly and the subsequent one stacks as expected.\n" } }