Add LLMDet zero-shot OD family #6248
Open
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Outdated suggestions cannot be applied.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.
Add 3 LLMDet zero-shot object detection models from iSEE-Laboratory:
llmdet-tiny-torch (692MB)
llmdet-base-torch (936MB)
llmdet-large-torch (1376MB)
All models:
FiftyOneZeroShotTransformerForObjectDetectionwrapperWhat changes are proposed in this pull request?
Add llmdet zero shot od family. (three models)
How is this patch tested? If it is not, please explain why.
Validated Transformers implementation on the
quickstartdataset: loaded each checkpoint, ran zero‑shot detection with COCO prompts, and confirmed reasonable detections and byte sizes. Spot‑checked zoo load.Release Notes
Is this a user-facing change that should be mentioned in the release notes?
notes for FiftyOne users.
Added three LLMDet open‑vocabulary zero‑shot object detectors (tiny/base/large) to the Torch model zoo.
What areas of FiftyOne does this PR affect?
fiftyonePython library changes