Getting it Right: Improving Spatial Consistency in Text-to-Image Models.
Agneet ChatterjeeGabriela Ben Melech StanEstelle AflaloSayak PaulDhruba GhoshTejas GokhaleLudwig SchmidtHannaneh HajishirziVasudev LalChitta BaralYezhou YangPublished in: CoRR (2024)
Keyphrases
- spatial information
- bayesian framework
- image content
- spatial constraints
- image data
- single image
- spatial relationships
- input image
- spatial distribution
- image classification
- image analysis
- web images
- statistical model
- spatial frequency
- high resolution
- image segmentation
- parametric models
- random fields
- color distribution
- spatial relations
- spatial features
- visual effects
- multiscale
- low level
- probabilistic model
- image features
- spatial and temporal
- spatio temporal
- template matching
- segmentation method
- image representation
- image regions
- visual features
- image processing
- spatial arrangement
- object models
- d objects
- text mining
- image pixels
- image collections
- segmentation algorithm
- region of interest
- test images
- keypoints
- spatial data
- hough transform