ELIQ: A Label-Free Framework for Quality Assessment of Evolving AI-Generated Images

2026-02-03

AI Summary

The rapid advancement of generative AI image models has rendered traditional human-annotated quality assessment approaches inadequate. To address this challenge, this work proposes ELIQ, a novel framework that enables the first unsupervised, label-free evaluation focusing on visual quality and text-image alignment. Built upon pretrained multimodal models, ELIQ leverages instruction tuning, lightweight gated fusion, and a Quality Query Transformer to automatically construct positive-negative sample pairs and train a dual-dimension quality predictor. Extensive experiments demonstrate that ELIQ significantly outperforms existing label-free methods across multiple benchmarks, exhibits strong generalization across model generations, and seamlessly transfers to user-generated content scenarios, offering a scalable solution for evaluating dynamically evolving generative models.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Computer Vision: Diffusion Models for Vision
• Natural Language Processing: Generation

Application Category

• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Generative text-to-image models are advancing at an unprecedented pace, continuously shifting the perceptual quality ceiling and rendering previously collected labels unreliable for newer generations. To address this, we present ELIQ, a Label-free Framework for Quality Assessment of Evolving AI-generated Images. Specifically, ELIQ focuses on visual quality and prompt-image alignment, automatically constructs positive and aspect-specific negative pairs to cover both conventional distortions and AIGC-specific distortion modes, enabling transferable supervision without human annotations. Building on these pairs, ELIQ adapts a pre-trained multimodal model into a quality-aware critic via instruction tuning and predicts two-dimensional quality using lightweight gated fusion and a Quality Query Transformer. Experiments across multiple benchmarks demonstrate that ELIQ consistently outperforms existing label-free methods, generalizes from AI-generated content (AIGC) to user-generated content (UGC) scenarios without modification, and paves the way for scalable and label-free quality assessment under continuously evolving generative models. The code will be released upon publication.

Problem

• quality assessment
• AI-generated images
• label-free
• evolving generative models
• prompt-image alignment

Innovation

• label-free
• quality assessment
• AIGC
• instruction tuning
• multimodal fusion