VersusDebias: Universal Zero-Shot Debiasing for Text-to-Image Models via SLM-Based Prompt Engineering and Generative Adversary.
Hanjun LuoZiye DengHaoyu HuangXuecheng LiuRuizhe ChenZuozhu LiuPublished in: CoRR (2024)
Keyphrases
- image data
- image features
- input image
- image content
- image retrieval
- image analysis
- image matching
- single image
- multiscale
- bayesian framework
- random fields
- text mining
- region of interest
- image representation
- edge detection
- low level
- parametric models
- probabilistic model
- statistical model
- high resolution
- information retrieval
- image pixels
- image segmentation
- web images
- keypoints
- generative model
- feature points
- image classification
- similarity measure
- hough transform
- bounding box
- discriminative learning
- image statistics
- visual effects