Can Vision Transformers Learn without Natural Images?
Kodai NakashimaHirokatsu KataokaAsato MatsumotoKenji IwataNakamasa InoueYutaka SatohPublished in: AAAI (2022)
Keyphrases
- natural images
- natural scenes
- higher order
- multiscale
- image denoising
- object recognition
- input image
- sparse coding
- image patches
- scale invariant
- denoising
- statistics of natural images
- receptive fields
- saliency map
- natural image patches
- power spectrum
- natural image statistics
- computer vision
- vision system
- visual scene
- learning algorithm
- markov random field
- visual cortex
- visual patterns
- image statistics
- prior models
- image processing
- single image
- color constancy
- image features
- feature extraction
- deep learning
- natural textures