facebookresearch / facebookresearch/jepa
Question about adapting V-JEPA to 3D imaging
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 4.1k
- Forks
- 424
- PR merge metrics
- No merged PRs in 30d
Description
Hello everyone,
Thanks for the source code !
Is it possible to adapt V-JEPA to 3D imaging (like medical imaging)?
I don't know which aproach would work best between converting imaging to pseudo-short movies of 2d slices or to adapt Vision transformer model to accept volumes instead of 2d imaging.
Thanks a lot :)
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
The issue names the V-JEPA source code but no files, tests, or model entry points. Start by reviewing the repository's model input assumptions and compare the two proposed approaches: pseudo-movies of 2D slices or volumetric Vision Transformer input. Done would require a defined adaptation scope and evaluation plan, neither of which the issue currently provides.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python, pytorch
- Domain
- computer-vision, machine-learning
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 20/100