MedShapeNet and example #9823
Open
MedShapeNet and example #9823
Conversation
This file contains 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
|
“MedShapeNet is not divided by train and test by default. Therefore, train data is downloaded from the first n required number of samples, and the test data is downloaded from the last n required number of samples.” i would recommend the data be split randomly if train/val/test is not predefined |
note that CI is also failing with this issue. however, I am having trouble installing medshapenet through pip to verify if your code actually works. Is that how you install it? could you provide your setup commands? based on this i will also advise wether those instructions should be included in your code somewhere (i will decide based on the complexity of the set up instructions) |
jdhenaos
force-pushed
the
medshapenet
branch
from
0825c61 to
e2f7faf
Compare
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## master #9823 +/- ##
==========================================
- Coverage 87.07% 86.31% -0.76%
==========================================
Files 490 490
Lines 32285 32280 -5
==========================================
- Hits 28112 27863 -249
- Misses 4173 4417 +244 ☔ View full report in Codecov by Sentry. |
jdhenaos
force-pushed
the
medshapenet
branch
from
237bd37 to
e2f7faf
Compare
jdhenaos
force-pushed
the
medshapenet
branch
from
a20df74 to
8a1e2d4
Compare
jdhenaos
force-pushed
the
medshapenet
branch
from
85fda59 to
1da4e60
Compare
jdhenaos
force-pushed
the
medshapenet
branch
from
9a25a82 to
028ff15
Compare
jdhenaos
force-pushed
the
medshapenet
branch
from
7feffdb to
73e96d3
Compare
jdhenaos
force-pushed
the
medshapenet
branch
from
75d9ce2 to
033987e
Compare
|
Should it be |
|
Indeed, The problem is that MedShapeNet uses a down-grade version of numpy. By default, it downloads version I'm not sure how to proceed with this conflict. Any hint you can give will be welcome. |
|
@xnuohz and @puririshi98 I have talked with the MedShapeNet. Looks like they already solved the problem, I was able to run the Dataset on Python 3.10.12. Please corroborate and let me know if there is some pending issue to solve |
xnuohz
suggested changes
Feb 18, 2025
Comment on lines
+51
to
+57
| train_dataset = MedShapeNet(root=root, size=50, split="train", | ||
| pre_transform=pre_transform, | ||
| transform=transform, force_reload=False) | ||
| print('Loading test data') | ||
| test_dataset = MedShapeNet(root=root, size=50, split="test", | ||
| pre_transform=pre_transform, | ||
| transform=transform, force_reload=False) |
There was a problem hiding this comment.
can the class initialization parameter signature be consistent with other datasets? in this way it can be much simpler here
also please share the training log of this new added dataset
Comment on lines
+120
to
+121
| train_data, val_data, test_data = random_split( | ||
| stl_files, [train_size, val_size, test_size]) |
There was a problem hiding this comment.
can we move random splits out of the dataset? imo, you need to split it in training stage.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
I have added MedShapeNet dataset, adding eight classes.
The next folders were ignored from MedShapeNet Dataset due to a lack of samples:
ASOCA (n=43)
AVT (n=45)
AutoImplantCraniotomy (n=14)
FaceVR (n=14)
This Dataset requires a size number as a parameter to download the same number of samples per shape to ensure class balance.
MedShapeNet is not divided by train and test by default. Therefore, train data is downloaded from the first n required number of samples, and the test data is downloaded from the last n required number of samples.
dgcnn classification example was modified to use both ModelNet and MedShapeNet.