Metraipolis › archive › dc_8527472154ba

Invisible Shortcuts: Why Vision Encoders Know Your Camera — arXiv:2608.05424

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.05424
License (as archived)
CC0-1.0
Archived
2026-10-31T10:35:54.703Z
SHA-256 of the archived text
13c16e05663a107a41ac8774972fb650820ad1996a621d7d7392dc39f05b1247
Stable URL of this page
https://metraipolis.com/documento/dc_8527472154ba

How to cite

Invisible Shortcuts: Why Vision Encoders Know Your Camera — arXiv:2608.05424 — arXiv (en), https://arxiv.org/abs/2608.05424 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_8527472154ba (Metraipolis archive dc_8527472154ba, sha256 13c16e05663a107a41ac8774972fb650820ad1996a621d7d7392dc39f05b1247)
BibTeX
@misc{dc8527472154ba,
  title = {Invisible Shortcuts: Why Vision Encoders Know Your Camera — arXiv:2608.05424},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.05424},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_8527472154ba; sha256 13c16e05663a107a41ac8774972fb650820ad1996a621d7d7392dc39f05b1247},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.05424. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Deep vision models exploit shortcuts, relying on cues that correlate with supervision signals. Prior work has focused on visible biases, such as object-background or texture correlations. We identify a different source of shortcut learning: invisible metadata traces embedded at the pixel level, for metadata such as image processing and photo acquisition. We hypothesize that large-scale semantic supervision, whether through categorical labels (ImageNet) or billion-scale captions (LAION), naturally induces metadata-semantics correlations during pretraining, leading models to convert low-level signals into predictive features. By introducing controlled metadata-semantics correlations, we show that stronger ones produce systematically higher sensitivity to metadata traces and larger performance degradation under metadata distribution shifts. We further explore mitigation strategies applied during and after pretraining that reduce sensitivity not only to targeted metadata but also to unseen ones, without sacrificing performance on downstream tasks. Metadata sensitivity also has a positive side: it partly explains the strong generated-image detection ability of some encoders, while its mitigation can improve out-of-distribution generalization. Code: this https URL

Tema degli esploratori: statistica e scienza dei dati (retro-catalogo arXiv: annunci del 2026-08-07 in cs.CV).
Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV) ; Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2608.05424 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-31T10:23:30.972Z dalla pagina https://arxiv.org/catchup/cs.CV/2026-08-07?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).