SneakyPrompt: Evaluating Robustness of Text-to-image Generative Models' Safety Filters.
Yuchen YangBo HuiHaolin YuanNeil Zhenqiang GongYinzhi CaoPublished in: CoRR (2023)
Keyphrases
- generative model
- bayesian framework
- input image
- image content
- image features
- mixture model
- test images
- probabilistic model
- em algorithm
- image classification
- visual patterns
- image retrieval
- image segmentation
- information retrieval
- discriminative learning
- prior knowledge
- conditional random fields
- object parts
- generative and discriminative models
- bounding box
- keypoints
- segmentation method
- action recognition
- segmentation algorithm
- image representation
- natural images
- semi supervised