Multi-Modal Language Models as Text-to-Image Model Evaluators

TL;DR

This paper introduces MT2IE, a novel evaluation framework using multi-modal large language models to assess text-to-image models more efficiently and accurately than traditional static benchmarks.

Contribution

The paper presents MT2IE, a dynamic, prompt-based evaluation method that correlates better with human judgment and requires significantly fewer prompts than existing benchmarks.

Findings

MT2IE's scores have higher correlation with human judgment.

MT2IE uses only 1/80th of prompts compared to static benchmarks.

MT2IE produces consistent relative rankings of T2I models.

Abstract

The steady improvements of text-to-image (T2I) generative models lead to slow deprecation of automatic evaluation benchmarks that rely on static datasets, motivating researchers to seek alternative ways to evaluate the T2I progress. In this paper, we explore the potential of multi-modal large language models (MLLMs) as evaluator agents that interact with a T2I model, with the objective of assessing prompt-generation consistency and image aesthetics. We present Multimodal Text-to-Image Eval (MT2IE), an evaluation framework that iteratively generates prompts for evaluation, scores generated images and matches T2I evaluation of existing benchmarks with a fraction of the prompts used in existing static benchmarks. Moreover, we show that MT2IE's prompt-generation consistency scores have higher correlation with human judgment than scores previously introduced in the literature. MT2IE generates prompts that are efficient at probing T2I model performance, producing the same relative T2I model rankings as existing benchmarks while using only 1/80th the number of prompts for evaluation.