Can Vision Transformers Learn without Natural Images?
Kodai NakashimaHirokatsu KataokaAsato MatsumotoKenji IwataNakamasa InouePublished in: CoRR (2021)
Keyphrases
- natural images
- natural scenes
- image denoising
- multiscale
- higher order
- object recognition
- input image
- sparse coding
- image patches
- computer vision
- natural image statistics
- denoising
- receptive fields
- visual cortex
- statistics of natural images
- test images
- human vision
- vision system
- scale invariant
- visual patterns
- power spectrum
- machine learning
- visual information
- intrinsic images
- image segmentation
- learning algorithm