Add classification template

Project-MONAI / model-zoo

MONAI Model Zoo that hosts models in the MONAI Bundle format.

Apache License 2.0

186 stars 68 forks source link

Add classification template #533

Closed KumoLiu closed 10 months ago

KumoLiu commented 10 months ago

Part of tutorial#1456

Description

Add a classification template

Status

Ready

Please ensure all the checkboxes:

[x] Codeformat tests passed locally by running ./runtests.sh --codeformat.
[ ] In-line docstrings updated.
[ ] Update version and changelog in metadata.json if changing an existing bundle.
[ ] Please ensure the naming rules in config files meet our requirements (please refer to: CONTRIBUTING.md).
[ ] Ensure versions of packages such as monai, pytorch and numpy are correct in metadata.json.
[ ] Descriptions should be consistent with the content, such as eval_metrics of the provided weights and TorchScript modules.
[ ] Files larger than 25MB are excluded and replaced by providing download links in large_file.yml.
[ ] Avoid using path that contains personal information within config files (such as use /home/your_name/ for "bundle_root").

KumoLiu commented 10 months ago

/build

ericspod commented 10 months ago

Hi @KumoLiu looks good! I had a few points about how transforms are being applied but it works for you so it's just a best practice thing. One other suggestion is to have a common.yaml file with all the common definitions in it (eg. network_def), then include that when doing things like training. I hadn't done that for the segmentation template because I wanted to maintain compatibility with the current version of MONAILabel. Here we don't need to worry about that so we can change things now to have a common file or leave it for a future refactor.

KumoLiu commented 10 months ago

Hi @ericspod, I didn't get the point of why we need the common.yaml. I see you created an issue here and also mentioned it in the dev meeting. But seems we don't need to specify the common field in multiple configs, we only need to add what we want to override in the evaluate.yaml or inference.yaml or other configs.

KumoLiu commented 10 months ago

After reading again the configs, I think I get the points. It makes sense if we add a common.yaml but every time we may need to specify multiple configs due to the existence of common.yaml. Perhaps we could refactor it later when we reach a consensus. What do you think?

ericspod commented 10 months ago

I'm good to refactor later, it's something we will likely do for both template bundles. I had recommended having a common config file in this bundle tutorial but it's not necessary.