Probing Commonsense Reasoning Capability of Text-to-Image Generative Models via Non-visual Description.
Mianzhi PanJianfei LiMingyue YuZheng MaKanzhi ChengJianbing ZhangJiajun ChenPublished in: CoRR (2023)
Keyphrases
- generative model
- commonsense reasoning
- visual patterns
- image classification
- input image
- image features
- web images
- low level
- probabilistic model
- image segmentation
- image retrieval
- spatial information
- discriminative learning
- high resolution
- visual information
- prior knowledge
- visual features
- high level
- semi supervised
- learning algorithm
- image representation
- discriminative models