T2I-ReasonBench: Benchmarking Reasoning-Informed Text-to-Image Generation
By: Kaiyue Sun , Rongyao Fang , Chengqi Duan and more
Potential Business Impact:
Tests if AI can draw pictures from tricky words.
We propose T2I-ReasonBench, a benchmark evaluating reasoning capabilities of text-to-image (T2I) models. It consists of four dimensions: Idiom Interpretation, Textual Image Design, Entity-Reasoning and Scientific-Reasoning. We propose a two-stage evaluation protocol to assess the reasoning accuracy and image quality. We benchmark various T2I generation models, and provide comprehensive analysis on their performances.
Similar Papers
Easier Painting Than Thinking: Can Text-to-Image Models Set the Stage, but Not Direct the Play?
CV and Pattern Recognition
Tests how well AI makes pictures from words.
OneIG-Bench: Omni-dimensional Nuanced Evaluation for Image Generation
CV and Pattern Recognition
Tests AI that makes pictures from words.
TiViBench: Benchmarking Think-in-Video Reasoning for Video Generative Models
CV and Pattern Recognition
Tests if AI can make videos that make sense.