Segmenting MRI images using 2D Unet#142
Open
NicMarchant wants to merge 24 commits intoshakes76:topic-recognitionfrom
Open
Segmenting MRI images using 2D Unet#142NicMarchant wants to merge 24 commits intoshakes76:topic-recognitionfrom
NicMarchant wants to merge 24 commits intoshakes76:topic-recognitionfrom
Conversation
…t was just the unsegmented data, added import calls to train.py to access each dataset
The encoder computes conv's and max pooling and saves the skip connections The bottleneck computes two convolutions with 1024 features The decoder apply upconv's, adds the skip connections and does convolutions The unet function groups each of these together and makes a model
Modified the input shape in modules.py, the model sucessfully runs epochs. Going to add the valdation and testing to modules.py as well as dice similarity
… the training, validation and testing of the model to the train.py. Added imports and plots to the predict.py so that the results of the model could be visualised. Going to add prints of the true mask vs the predicted mask to show in the readme Havent successfully tested the model yet
…RANGPUR. Going to impliment a custom dice function to calculate the dice score. After the results from Rangpur are found I will tweak the model accordingly
true segmentation and predicted segmentations Current predictions are very poor, however only 1/10 of the data is being used to train as my local machine is running the code. going to modify the train.py to include working dice
Added dice to train.py referenced it aswell Increased the learning rate to 0.0001 cleaned up unuesd imports
Removed mid function comments from modules to make it less busy Tried multiple different loss functions however I cannot get better dice than around 0.4 If dice loss is used then the preducted segmentation is black
I added data sorting as per someones ed post and it made the model significantly better, also changedd the folder layout to suit the spec
Changed to dice and binary cross entropy, getting 65 dice off 3 epochs of all the data
Accidently removed files from github thinking they were duplicates.
going to run with the full dataset now to get the performance and then will work on readme
Collaborator
|
This is an initial inspection, no action is required at this point
Note |
Collaborator
|
Nicholas Marchant |
Collaborator
Marking
Marked as per the due date and changes after which aren't necessarily allowed to contribute to grade for fairness. |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Collaborator
|
Nicholas Marchant |
Owner
|
Cant merge because of conflicting changes to main repo files. Please update for merge, doesn't affect grade. |
Author
|
Good morning,
Sorry I’m a little confused as to which files need to be updated and which are conflicting.
Kind regards,
Nicholas Marchant
…________________________________
From: Shakes ***@***.***>
Sent: Tuesday, November 19, 2024 2:46:23 PM
To: shakes76/PatternAnalysis-2024 ***@***.***>
Cc: NicMarchant ***@***.***>; Author ***@***.***>
Subject: Re: [shakes76/PatternAnalysis-2024] Segmenting MRI images using 2D Unet (PR #142)
Cant merge because of conflicting changes to main repo files. Please update for merge, doesn't affect grade.
—
Reply to this email directly, view it on GitHub<#142 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/BHWFOR6JDUFD4MQ2FWJ3SST2BLGC7AVCNFSM6AAAAABQUOYEYWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDIOBUG42TQNJYG4>.
You are receiving this because you authored the thread.Message ID: ***@***.***>
|
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
6 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
This pull request implements a 2D Unet to create binary class segmentation of hip MRI images.