TempViz: On the Evaluation of Temporal Knowledge in Text-to-Image Models
This addresses a gap in evaluating how text-to-image models handle temporal changes, which is crucial for generating contextually relevant images, though it is incremental as it focuses on benchmarking rather than improving models.
The authors tackled the problem of evaluating temporal knowledge in text-to-image models by introducing TempViz, a dataset of 7.9k prompts and 600 reference images, and found that models perform poorly, with no model exceeding 75% accuracy across categories.
Time alters the visual appearance of entities in our world, like objects, places, and animals. Thus, for accurately generating contextually-relevant images, knowledge and reasoning about time can be crucial (e.g., for generating a landscape in spring vs. in winter). Yet, although substantial work exists on understanding and improving temporal knowledge in natural language processing, research on how temporal phenomena appear and are handled in text-to-image (T2I) models remains scarce. We address this gap with TempViz, the first data set to holistically evaluate temporal knowledge in image generation, consisting of 7.9k prompts and more than 600 reference images. Using TempViz, we study the capabilities of five T2I models across five temporal knowledge categories. Human evaluation shows that temporal competence is generally weak, with no model exceeding 75% accuracy across categories. Towards larger-scale studies, we also examine automated evaluation methods, comparing several established approaches against human judgments. However, none of these approaches provides a reliable assessment of temporal cues - further indicating the pressing need for future research on temporal knowledge in T2I.