AI Paper Atlas

Research map / Label-efficient and robust learning / Test-time adaptation

Processing and acquisition traces in visual encoders: What does CLIP know about your camera?

Ryan Ramos, Vladan Stojnić, Giorgos Kordopatis-Zilos, Yuta Nakashima, Giorgos Tolias, Noa Garcia

ICCV 2025 · Label-efficient and robust learning › Test-time adaptation · approach: Test-time adaptation of vision-language models · 3 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use contrastive vision-language training instead of supervised or self-supervisedvisual encoder embeddings · FlickrExif model (smart vs non-smart)top-k neighbor same-camera rate: 59.0% to 70.0% vs at most 58.3%Improved
use contrastive vision-language trainingvisual encoder embeddings · FlickrExif make prediction (95% masking)linear probe accuracy: exceed 20% vs baseline of 11.11%Improved
train CLIP with DINOv2-style color augmentations (ablation)OpenCLIP ViT-B/32 · processing-label prediction on ImageNet/iNaturalistlinear probe accuracyImproved
train CLIP with DINOv2-style color augmentations (ablation)OpenCLIP ViT-B/32 · acquisition-label prediction on FlickrExiflinear probe accuracyMixed
train CLIP with color augmentations (ablation)OpenCLIP ViT-B/32 · semantic kNN under coupled metadatasemantic accuracyImproved
make metadata labels coincide with semantic labels (pos-same, neg-same)CVL and ConvNeXt encoders · ImageNet/iNaturalist kNNsemantic accuracyHurt
apply different processing between train and test (all-diff, uniform)visual encoders · ImageNet/iNaturalist kNNsemantic accuracyNo clear effect
make retrieval negatives share query camera typeCVL encoders · PairCamsrecall@1: near-perfect recall@1 to below 0.85Hurt
increase center masking to 95%visual encoders · ImageNet validationsemantic accuracyHurt
mask image content (95% vs none) when predicting acquisition labelsvisual encoders · FlickrExifacquisition accuracyHurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy