.png){kind=icon}
.png){kind=icon}
Return
Describe-to-Score: A text-guided framework for image complexity assessment
.png){kind=icon}
PRE
AI
S
L
D
Z
DOI：10.1016/j.patcog.2026.114549.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
• We propose D2S, a text-guided framework for image complexity assessment with training-time semantic supervision and no test-time cost. • We develop feature alignment and entropy distribution alignment to regularize the visual encoder with caption-derived semantic structure. • D2S achieves state-of-the-art results on IC9600 while maintaining strong efficiency and competitive transferability across related benchmarks.
Keywords:
Image complexity assessment
Multi-modal fusion
Visual–text alignment
Journal
IF:
7.6
Papers:
1.3W
Citations:
4.5W
Organization
Cited Papers