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

arXiv 2508.10637 ↗First posted 14 Aug 2025cs.CVcited by 3

Not on the map. Why?

Abstract (arXiv)

Prior work has analyzed the robustness of visual encoders to image transformations and corruptions, particularly in cases where such alterations are not seen during training. When this occurs, they introduce a form of distribution shift at test time, often leading to performance degradation. The primary focus has been on severe corruptions that, when applied aggressively, distort useful signals necessary for accurate semantic predictions. We take a different perspective by analyzing parameters of the image acquisition process and transformations that may be subtle or even imperceptible to the human eye. We find that such parameters are systematically encoded in the learned visual representations and can be easily recovered. More strikingly, their presence can have a profound impact, either positively or negatively, on semantic predictions. This effect depends on whether there is a strong correlation or anti-correlation between semantic labels and these acquisition-based or processing-based labels. Our code and data are available at: https://github.com/ryan-caesar-ramos/visual-encoder-traces

Key facts. Cited by 3 papers in the map (data through 14 Oct 2026). Builds on 63 of its 81 references that are in the map. First posted on arXiv 14 Aug 2025.

Builds on

63 of 81 references.

References · 63

2015Deep Residual Learning for Image Recognitioncited by 52,096
2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,744
2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 30,186
2014Very Deep Convolutional Networks for Large-Scale Image Recognitioncited by 15,364
2020A Simple Framework for Contrastive Learning of Visual Representationscited by 12,049
2015Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networkscited by 9,639
2014ImageNet Large Scale Visual Recognition Challengecited by 8,498
2023Segment Anythingcited by 8,234
2023DINOv2: Learning Robust Visual Features without Supervisioncited by 7,974
2014Explaining and Harnessing Adversarial Examplescited by 7,364
2017Mask R-CNNcited by 7,277
2021Emerging Properties in Self-Supervised Vision Transformerscited by 6,968

38 more

2015You Only Look Once: Unified, Real-Time Object Detectioncited by 4,852
2022A ConvNet for the 2020scited by 3,579
2023Sigmoid Loss for Language Image Pre-Trainingcited by 3,109
2019Benchmarking Neural Network Robustness to Common Corruptions and Perturbationscited by 2,877
2020The Many Faces of Robustness: A Critical Analysis of Out-of-Distribution Generalizationcited by 1,850
2016Adversarial examples in the physical worldcited by 1,769
2021An Empirical Study of Training Self-Supervised Vision Transformerscited by 1,616
2018ImageNet-trained CNNs are biased towards texture; increasing shape bias improves accuracy and robustnesscited by 1,432
2019Natural Adversarial Examplescited by 1,408
2019Distributionally Robust Neural Networks for Group Shifts: On the Importance of Regularization for Worst-Case Generalizationcited by 1,247
2025SigLIP 2: Multilingual Vision-Language Encoders with Improved Semantic Understanding, Localization, and Dense Featurescited by 1,188
2022Reproducible scaling laws for contrastive language-image learningcited by 1,173
2017One pixel attack for fooling deep neural networkscited by 582
2017Fine-tuning CNN Image Retrieval with No Human Annotationcited by 499
2023Towards Universal Fake Image Detectors that Generalize Across Generative Modelscited by 490
2021Intriguing Properties of Vision Transformerscited by 446
2020Improving robustness against common corruptions by covariate shift adaptationcited by 431
2019Making Convolutional Networks Shift-Invariant Againcited by 391
2018Learning Not to Learn: Training Deep Neural Networks with Biased Datacited by 255
2019Towards Fairness in Visual Recognition: Effective Strategies for Bias Mitigationcited by 252
2023What does CLIP know about a red circle? Visual prompt engineering for VLMscited by 218
2019Learning De-biased Representations with Biased Representationscited by 199
2020How Much Position Information Do Convolutional Neural Networks Encode?cited by 172
2022DE-FAKE: Detection and Attribution of Fake Images Generated by Text-to-Image Generation Modelscited by 168
2016Improving the Robustness of Deep Neural Networks via Stability Trainingcited by 156
2021MultiBench: Multiscale Benchmarks for Multimodal Representation Learningcited by 146
2020On Translation Invariance in CNNs: Convolutional Layers can Exploit Absolute Spatial Locationcited by 142
2024Robust CLIP: Unsupervised Adversarial Fine-Tuning of Vision Embeddings for Robust Large Vision-Language Modelscited by 120
2022Understanding Zero-Shot Adversarial Robustness for Large-Scale Modelscited by 117
2024OpenBias: Open-set Bias Detection in Text-to-Image Generative Modelscited by 61
2023Discovering and Mitigating Visual Biases through Keyword Explanationcited by 58
2024One Prompt Word is Enough to Boost Adversarial Robustness for Pre-trained Vision-Language Modelscited by 48
2024A Decade's Battle on Dataset Bias: Are We There Yet?cited by 47
2024Fake or JPEG? Revealing Common Biases in Generated Image Detection Datasetscited by 43
2022Unbiased Supervised Contrastive Learningcited by 40
2024WATT: Weight Average Test-Time Adaptation of CLIPcited by 33
2023ImageNet-E: Benchmarking Neural Network Robustness via Attribute Editingcited by 30
2025ILIAS: Instance-Level Image retrieval At Scalecited by 30

50 most cited of 63 All 63 (JSON)

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 3.

3 papers.