Project-MONAI / Project-MONAI/model-zoo

using the original labels, instead of inverted transformed labels for evaluation

Open
#181 11 comments 2 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Python
Stars
338
Forks
106
PR merge metrics
No merged PRs in 30d

Description

Currently quite a few bundles use
the inverted version of preprocessed original labels as the reference during evaluation:

https://github.com/Project-MONAI/model-zoo/blob/987f49fe9e0f20f047a2252c1ace38bd760cfb33/models/spleen_ct_segmentation/configs/evaluate.json#L12-L21

which is different from the approach taken in the tutorial using the original label directly during evaluation (https://github.com/Project-MONAI/tutorials/blob/main/3d_segmentation/spleen_segmentation_3d.ipynb)

would be great to consider the standard method used in the tutorial for bundle evaluate.json

cc @holgerroth @dongyang0122

(this needs an update as well) https://github.com/Project-MONAI/tutorials/tree/main/bundle/spleen_segmentation)

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Compare the model-zoo spleen_ct_segmentation/configs/evaluate.json reference at the linked revision with the 3d_segmentation/spleen_segmentation_3d.ipynb tutorial. Review the related tutorials/bundle/spleen_segmentation directory and identify bundles using inverted preprocessed labels. Done means the affected evaluate.json files use the tutorial's original-label evaluation approach and the linked bundle tutorial is updated as needed.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
machine-learning
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.