Feature/ultralytics all models for Classification #281
Conversation
|
Hi, @mayankagarwals 👋🏻! Could you create a small Google Colab to test that change? |
supervision/classification/core.py
Outdated
| @@ -40,6 +42,9 @@ def __post_init__(self) -> None: | |||
| _validate_confidence(self.confidence, n) | |||
|
|
|||
| @classmethod | |||
| @deprecated( | |||
| "This method is deprecated and removed in 0.15.0 release. Use sv.Classifications.from_ultralytics() instead." | |||
There was a problem hiding this comment.
Let's make that 0.16.0. We should give people at least 2 full release cycles to migrate. That change will be released with 0.14.0.
supervision/classification/core.py
Outdated
| >>> import supervision as sv | ||
|
|
||
| >>> image = cv2.imread(SOURCE_IMAGE_PATH) | ||
| >>> model = YOLO('yolov8s.pt') |
There was a problem hiding this comment.
Let's update the example to use only detection and segmentation models here.
There was a problem hiding this comment.
Didn't quite get you. yolov8s.pt is a detection model itself. We can add a classification model here if that's what you mean
There was a problem hiding this comment.
That is my point exactly. We are documenting sv.Classifications.from_ultralytics but we are showing examples of usage with detection models. We should show classification model examples.
There was a problem hiding this comment.
Done! Although I can see only yolo models for classification on ultralytics https://docs.ultralytics.com/tasks/classify/
|
Hi @artyaltanzaya 👋🏻 I did an initial round of reviews on this PR. Please take it from here 🙏🏻 and cooperate with @mayankagarwals to make sure we will be able to release that change with |
|
Hi @SkalskiP Have added a testing colab in the description. |
|
@mayankagarwals 🙏🏻 please also make sure to resolve |
|
|
…classificationUltralytics
|
@onuralpszr Couple things with the pre-commit-ci
|
For CLA I need to ask because I did not added. For flake it I only saw long line error so it is normal. :) Also you can squash those changes for make it look clean :) |
|
@SkalskiP this require your attention for CLA-bot: https://github.com/cla-assistant/cla-assistant#can-i-allow-bot-user-contributions |
|
@mayankagarwals CLA part has been solved, thanks to @capjamesg. |
Yes yes, just FYI that there is a minor behavior change. If it's intentional, no issues
Amazing, thanks! @artyaltanzaya All issues are addressed :) |
|
@mayankagarwals I found a small problem during the final tests. You can take a look here. Committed a fix into your branch. It looks like we are ready to merge. |
|
@capjamesg thanks for solving issues with bot users 🙏🏻 |
@SkalskiP Thanks for the fix That's quite a crucial fix. Apologies for the lapse on my part, it's my first commit to the classification part of codebase. Looks like I got the semantics of the Classifications feature wrong. For some reason I thought it was class of bounding boxes classified. That's my bad. Your fix does solve that. My only worry is Again, thanks for the support! |
No worries! That's why we have multiple people looking at the code and testing. ;) |
Successor of #222
Testing colab: https://colab.research.google.com/drive/1pxIqThTJwoP8Hwu95KZ3CUUio69w3_Ig?usp=sharing