AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

Lost in Translation: Modern Neural Networks Still Struggle With Small Realistic Image Transformations

Ofir Shifman, Yair Weiss

ECCV 2024 · Label-efficient and robust learning › Recognition and long-tail · approach: Loss design and classifier theory · 2 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
apply RICS crop selection (random-filter scoring) to DINOv2 at inferenceDINOv2 · ImageNet subset, 1-pixel realistic shiftAdversarial robustness (1-NN): 62.42 -> 94.73Improved
apply RICS to DINOv2 on accuracyDINOv2 · ImageNet subsetaccuracy (%): 84.11 -> 82.66Hurt
use Mexican-hat scoring instead of random filter scoring (ablation)RICS · ImageNet subset, DINOv2Adversarial robustness (1-NN): 94.73 -> 95.14No clear effect
increase crop size k (ablation)RICS · ImageNetaccuracy vs robustnessMixed
scale up pretraining data (open-CLIP on LAION-2B, DINOv2) without translation-specific designopen-CLIP, DINOv2 · ImageNet subset, 1-pixel shiftAdversarial robustness (1-NN): 62.42 (DINOv2) vs 46.33 (ResNet50)No clear effect
use cyclic-shift-invariant architecture (AFC) on realistic non-cyclic shiftsAFC · ImageNet subset, 1-pixel realistic shiftAdversarial robustness (1-NN): 100% cyclic vs 89.77 realisticMixed
apply RICS with cyclic crops to DINOv2DINOv2 · ImageNet val cyclic shiftsaccuracy (%): 84.1 -> 83.93Hurt
apply RICS to DINOv2 on Places365DINOv2 · Places365 subsetAdversarial robustness (1-NN) 1px: 74.94 -> 96.89Improved

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