lightly-ai/lightly

Use InterpolationMode.BICUBIC instead of magic interpolation values in transforms

Closed

#2,009 opened on Jul 30, 2026

View on GitHubContributor guide
 (0 comments) (0 reactions) (0 assignees)Python (337 forks)auto 404
good first issuehelp wantedrefactor

Repository metrics

Stars
 (3,775 stars)
PR merge metrics
 (PR metrics pending)

Description

lightly/transforms/ has two inconsistent ways of specifying bicubic interpolation for RandomResizedCrop:

  • interpolation=3 — mae_transform.py, ijepa_transform.py, wmse_transform.py, capi_transform.py
  • interpolation=PIL.Image.BICUBIC with a # type: ignore[attr-defined] — dino_transform.py, ibot_transform.py

Both should become:

T.RandomResizedCrop(input_size, scale=(min_scale, 1.0), interpolation=T.InterpolationMode.BICUBIC)

T is torchvision_transforms from lightly.transforms.torchvision_v2_compatibility, already imported in every affected file, and T.InterpolationMode.BICUBIC resolves under both the v1 and v2 compatibility paths. No new import needed, and it drops the type: ignore in the DINO/iBOT files.

Files to update:

  • lightly/transforms/mae_transform.py
  • lightly/transforms/ijepa_transform.py
  • lightly/transforms/wmse_transform.py
  • lightly/transforms/capi_transform.py
  • lightly/transforms/dino_transform.py
  • lightly/transforms/ibot_transform.py

Contributor guide

Research direction
Inspect the six transform files under lightly/transforms, locate the interpolation arguments in RandomResizedCrop, and replace both the numeric 3 and PIL.Image.BICUBIC with T.InterpolationMode.BICUBIC. Remove the type: ignore comment in the DINO and iBOT files. Run the relevant tests to confirm nothing breaks.
Tech stack
pythonpytorch
Domain
machine learning
Issue type
Refactor
DifficultyEstimated implementation difficulty for a new contributor, from 1 for very small changes to 5 for expert-level work.
2
Estimated timeA rough time range for an experienced contributor to investigate, implement, test, and prepare a pull request.
1-3 hours
Activity statusHow available the issue appears right now: fresh, active, stale, blocked, or waiting on maintainer input.
Fresh
ClarityHow clearly the issue explains the expected change, acceptance criteria, and next step.
Clear
Prerequisites
GitPythonPyTorch
Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.
85