RAW-Adapter: Adapting Pre-trained Visual Model to Camera RAW Images and A Benchmark
Not on the map. Why?
Abstract (arXiv)
In the computer vision community, the preference for pre-training visual models has largely shifted toward sRGB images due to their ease of acquisition and compact storage. However, camera RAW images preserve abundant physical details across diverse real-world scenarios. Despite this, most existing visual perception methods that utilize RAW data directly integrate image signal processing (ISP) stages with subsequent network modules, often overlooking potential synergies at the model level. Building on recent advances in adapter-based methodologies in both NLP and computer vision, we propose RAW-Adapter, a novel framework that incorporates learnable ISP modules as input-level adapters to adjust RAW inputs. At the same time, it employs model-level adapters to seamlessly bridge ISP processing with high-level downstream architectures. Moreover, RAW-Adapter serves as a general framework applicable to various computer vision frameworks. Furthermore, we introduce RAW-Bench, which incorporates 17 types of RAW-based common corruptions, including lightness degradations, weather effects, blurriness, camera imaging degradations, and variations in camera color response. Using this benchmark, we systematically compare the performance of RAW-Adapter with state-of-the-art (SOTA) ISP methods and other RAW-based high-level vision algorithms. Additionally, we propose a RAW-based data augmentation strategy to further enhance RAW-Adapter's performance and improve its out-of-domain (OOD) generalization ability. Extensive experiments substantiate the effectiveness and efficiency of RAW-Adapter, highlighting its robust performance across diverse scenarios.
Key facts. Cited by 2 papers in the map (data through 7 Oct 2026). Builds on 78 of its 123 references that are in the map. First posted on arXiv 21 Mar 2025.
Builds on
78 of 123 references.
References · 78
- 2017Attention Is All You Needcited by 59,932
- 2015Deep Residual Learning for Image Recognitioncited by 51,809
- 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 29,915
- 2015U-Net: Convolutional Networks for Biomedical Image Segmentationcited by 19,604
- 2014Microsoft COCO: Common Objects in Contextcited by 17,498
- 2021Swin Transformer: Hierarchical Vision Transformer using Shifted Windowscited by 12,082
- 2023Segment Anythingcited by 8,158
- 2020End-to-End Object Detection with Transformerscited by 7,384
- 2019Parameter-Efficient Transfer Learning for NLPcited by 4,195
- 2015Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classificationcited by 3,590
- 2021SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformerscited by 2,923
- 2019Benchmarking Neural Network Robustness to Common Corruptions and Perturbationscited by 2,857
38 more
- 2020Implicit Neural Representations with Periodic Activation Functionscited by 2,214
- 2022Visual Prompt Tuningcited by 1,674
- 2019MMDetection: Open MMLab Detection Toolbox and Benchmarkcited by 1,379
- 2024Depth Anything: Unleashing the Power of Large-Scale Unlabeled Datacited by 1,346
- 2021Per-Pixel Classification is Not All You Need for Semantic Segmentationcited by 1,148
- 2022Vision Transformer Adapter for Dense Predictionscited by 564
- 2017Learning multiple visual domains with residual adapterscited by 553
- 2020Sparse R-CNN: End-to-End Object Detection with Learnable Proposalscited by 494
- 2018Learning to See in the Darkcited by 459
- 2023Segment Anything in High Qualitycited by 373
- 2022Tip-Adapter: Training-free Adaption of CLIP for Few-shot Classificationcited by 355
- 2021NeRF in the Dark: High Dynamic Range View Synthesis from Noisy Raw Imagescited by 336
- 2021VL-Adapter: Parameter-Efficient Transfer Learning for Vision-and-Language Taskscited by 319
- 2018Unprocessing Images for Learned Raw Denoisingcited by 250
- 2021TransWeather: Transformer-based Restoration of Images Degraded by Adverse Weather Conditionscited by 243
- 2020CycleISP: Real Image Restoration via Improved Data Synthesiscited by 212
- 2022Improving Out-of-Distribution Robustness via Selective Augmentationcited by 202
- 2021Data Augmentation Can Improve Robustnesscited by 193
- 2020A Physics-based Noise Formation Model for Extreme Low-light Raw Denoisingcited by 138
- 2017On the Robustness of Semantic Segmentation Models to Adversarial Attackscited by 126
- 20223D Common Corruptions and Data Augmentationcited by 120
- 2019Benchmarking the Robustness of Semantic Segmentation Modelscited by 111
- 2020Replacing Mobile Camera ISP with a Single Deep Learning Modelcited by 108
- 2020Memory-Efficient Incremental Learning Through Feature Adaptationcited by 107
- 2021Invertible Image Signal Processingcited by 94
- 2018DeepISP: Towards Learning an End-to-End Image Processing Pipelinecited by 88
- 2022You Only Need 90K Parameters to Adapt Light: A Light Weight Transformer for Image Enhancement and Exposure Correctioncited by 81
- 2021Mobile Computational Photography: A Tourcited by 62
- 2022Multitask AET with Orthogonal Tangent Regularity for Dark Object Detectioncited by 59
- 2024Dynamic Adapter Meets Prompt Tuning: Parameter-Efficient Transfer Learning for Point Cloud Analysiscited by 52
- 2020How to Train Neural Networks for Flare Removalcited by 50
- 2022Flare7K: A Phenomenological Nighttime Flare Removal Datasetcited by 49
- 2022Dancing under the stars: video denoising in starlightcited by 47
- 2023Instance Segmentation in the Darkcited by 45
- 2021Learning RAW-to-sRGB Mappings with Inaccurately Aligned Supervisioncited by 43
- 2022Model-Based Image Signal Processors via Learnable Dictionariescited by 43
- 2017Dirty Pixels: Towards End-to-End Image Processing and Perceptioncited by 42
- 2019CameraNet: A Two-Stage Framework for Effective Camera ISP Learningcited by 38
50 most cited of 78 All 78 (JSON)
Built on by
Citing papers per year: 2026: 2.
2 papers.