Research map / Detection and segmentation / Open-set and few-shot detection
Examining the Impact of Optical Aberrations to Image Classification and Object Detection Models
Patrick Müller, Alexander Braun, Margret Keuper
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add OpticsAugment training | ResNeXt50 · ImageNet-100 OpticsBench | accuracy: 28.42 -> 50.12 | Improved |
| add OpticsAugment training | DenseNet161 · ImageNet-100 OpticsBench | accuracy: 32.56 -> 52.32 | Improved |
| add OpticsAugment training | ResNet101 · ImageNet-100 OpticsBench | accuracy: 41.79 -> 57.16 | Improved |
| add OpticsAugment training | MobileNet_v3_large · ImageNet-100 OpticsBench | accuracy: 28.90 -> 39.04 | Improved |
| add OpticsAugment training | EfficientNet_b0 · ImageNet-100 OpticsBench | accuracy: 32.27 -> 42.42 | Improved |
| add OpticsAugment training instead of AugMix | ResNeXt50 · ImageNet-100 OpticsBench | accuracy: +11.6% points over AugMix | Improved |
| add OpticsAugment training | ResNet50 · ImageNet-1k OpticsBench | average accuracy: 33.2 -> 41.7 | Improved |
| add OpticsAugment training | ResNet50 · ImageNet-1k val | accuracy: 76.1 -> 74.2 | Hurt |
| add OpticsAugment training | ResNeXt50 · ImageNet-100 2D common corruptions | accuracy | Mixed |
| cascade AugMix with OpticsAugment | EfficientNet · ImageNet-100 2D common corruptions | accuracy: +3.4% points | Improved |
| cascade AugMix with OpticsAugment | ImageNet-100 models · ImageNet-100 OpticsBench | accuracy | Hurt |
| add OpticsAugment training | ResNeXt50 · ImageNet-100 adversarial robustness (APGD-DLR) | robust accuracy: 1.2 -> 3.1 | No clear effect |
Closest papers by meaning
- Deep Optics for Monocular Depth Estimation and 3D Object Detection (ICCV 2019 · similarity 0.88)
- Fast Two-Step Blind Optical Aberration Correction (ECCV 2022 · similarity 0.87)
- A Physics-informed Low-rank Deep Neural Network for Blind and Universal Lens Aberration Correction (CVPR 2024 · similarity 0.86)
- Towards Universal Computational Aberration Correction in Photographic Cameras: A Comprehensive Benchmark Analysis (CVPR 2026 · similarity 0.86)
- LAION-C: An Out-of-Distribution Benchmark for Web-Scale Vision Models (ICML 2025 · similarity 0.84)
- Blurry-Edges: Photon-Limited Depth Estimation from Defocused Boundaries (CVPR 2025 · similarity 0.83)
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.
