MarMaster/

corruption-gaussian_blur

	Corruption Dataset: Gaussian_Blur




	
	Dataset Description

This dataset contains corrupted versions of ImageNet-1K images using gaussian_blur corruption. It is part of the ImageNet-C benchmark for evaluating model robustness to common image corruptions.

Train: 1,281,167 corrupted images Validation: 50,000 corrupted images
Classes: 1000 ImageNet-1K classes Format: Arrow (Hugging Face Datasets)

Applies… See the full description on the dataset page: https://huggingface.co/datasets/MarMaster/corruption-gaussian_blur.