# Image Classification on ImageNet-P (Corruption Robustness) Source: [Image Classification on ImageNet-P (Corruption Robustness)]() Collection index: [llms.txt]() Data updated: 2026-03-10T07:55:04.319Z Evaluation of Image Classification performance on the ImageNet-P dataset, reporting accuracy under various noise, blur, and geometric corruption conditions. A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. | Scope | Recorded value | | --- | --- | | Task | Image Classification | | Dataset (including any stated split) | ImageNet-P | | Metric | Accuracy (Gaussian Noise) | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 14 of 14 | | Coverage | All results currently returned for this benchmark | Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol. ## Metrics - [Accuracy (Gaussian Noise)](): Higher is better; unit: PERCENT. - [Top-5 Accuracy (Avg)](): Higher is better; unit: PERCENT. - [Accuracy (Shot Noise)](): Higher is better; unit: PERCENT. - [Accuracy (Speckle Noise)](): Higher is better; unit: PERCENT. - [Accuracy (Motion Blur)](): Higher is better; unit: PERCENT. - [Accuracy (Zoom Blur)](): Higher is better; unit: PERCENT. - [Accuracy (Gaussian Blur)](): Higher is better; unit: PERCENT. - [Accuracy (Snow)](): Higher is better; unit: PERCENT. - [Accuracy (Spatter)](): Higher is better; unit: PERCENT. - [Accuracy (Brightness)](): Higher is better; unit: PERCENT. - [Accuracy (Translate)](): Higher is better; unit: PERCENT. - [Accuracy (Rotate)](): Higher is better; unit: PERCENT. - [Accuracy (Tilt)](): Higher is better; unit: PERCENT. - [Accuracy (Scale)](): Higher is better; unit: PERCENT. - [Accuracy (Shear)](): Higher is better; unit: PERCENT. - [Top-1 Accuracy (Gaussian Noise)](): Higher is better; unit: PERCENT. - [Top-1 Accuracy (Shot Noise)](): Higher is better; unit: PERCENT. - [Top-1 Accuracy (Speckle Noise)](): Higher is better; unit: PERCENT. - [Top-1 Accuracy (Motion Blur)](): Higher is better; unit: PERCENT. - [Top-1 Accuracy (Zoom Blur)](): Higher is better; unit: PERCENT. - [Top-1 Accuracy (Gaussian Blur)](): Higher is better; unit: PERCENT. - [Top-1 Accuracy (Snow)](): Higher is better; unit: PERCENT. - [Top-1 Accuracy (Spatter)](): Higher is better; unit: PERCENT. - [Top-1 Accuracy (Brightness)](): Higher is better; unit: PERCENT. - [Top-1 Accuracy (Translate)](): Higher is better; unit: PERCENT. - [Top-1 Accuracy (Rotate)](): Higher is better; unit: PERCENT. - [Top-1 Accuracy (Tilt)](): Higher is better; unit: PERCENT. - [Top-1 Accuracy (Scale)](): Higher is better; unit: PERCENT. - [Top-1 Accuracy (Shear)](): Higher is better; unit: PERCENT. - [Top-1 Accuracy (Avg)](): Higher is better; unit: PERCENT. ## Recorded results | Method | Accuracy (Gaussian Noise) | Result date | Report paper | | --- | --- | --- | --- | | MoCo-v3 + SER | 89.96 | Not recorded | 69afa57f2d603e6de3e43a01 | | MoCo-v3 + E-SSL | 89.6 | Not recorded | 69afa57f2d603e6de3e43a01 | | MoCo-v3 + EquiMod | 88.78 | Not recorded | 69afa57f2d603e6de3e43a01 | | MoCo-v3 + SER | 88.62 | Not recorded | 69afa57f2d603e6de3e43a01 | | MoCo-v3 | 87.75 | Not recorded | 69afa57f2d603e6de3e43a01 | | MoCo-v3 + AugSelf | 87.58 | Not recorded | 69afa57f2d603e6de3e43a01 | | MoCo-v3 + STL | 86.67 | Not recorded | 69afa57f2d603e6de3e43a01 | ### MoCo-v3 + SER Accuracy (Gaussian Noise): 89.96. Result ID: 69afce3c6dbb7075b957ec4a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Algorithm":"MoCo-v3","variant":"‡"}. This record does not assert independent reproduction. ### MoCo-v3 + E-SSL Accuracy (Gaussian Noise): 89.6. Result ID: 69afce3c6dbb7075b957ec49. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Algorithm":"MoCo-v3","variant":"‡"}. This record does not assert independent reproduction. ### MoCo-v3 + EquiMod Accuracy (Gaussian Noise): 88.78. Result ID: 69afce3c6dbb7075b957ec48. [Reporting paper]() Recorded evaluation configuration: {"Base Algorithm":"MoCo-v3","variant":"†"}. This record does not assert independent reproduction. ### MoCo-v3 + SER Accuracy (Gaussian Noise): 88.62. Result ID: 69afce3c6dbb7075b957ec47. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Algorithm":"MoCo-v3"}. This record does not assert independent reproduction. ### MoCo-v3 Accuracy (Gaussian Noise): 87.75. Result ID: 69afce3c6dbb7075b957ec44. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Base Algorithm":"MoCo-v3"}. This record does not assert independent reproduction. ### MoCo-v3 + AugSelf Accuracy (Gaussian Noise): 87.58. Result ID: 69afce3c6dbb7075b957ec45. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Base Algorithm":"MoCo-v3"}. This record does not assert independent reproduction. ### MoCo-v3 + STL Accuracy (Gaussian Noise): 86.67. Result ID: 69afce3c6dbb7075b957ec46. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Algorithm":"MoCo-v3"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.