AIGCIQA2023: A Large-scale Image Quality Assessment Database for AI Generated Images: from the Perspectives of Quality, Authenticity and Correspondence

TL;DR

This paper introduces AIGCIQA2023, a large-scale database of AI-generated images evaluated for quality, authenticity, and correspondence, to better understand human preferences and benchmark existing IQA metrics.

Contribution

It creates a comprehensive database of over 2000 AI-generated images with human preference annotations across three perspectives, and benchmarks current IQA metrics on this dataset.

Findings

State-of-the-art IQA metrics show varying performance on AIGCIQA2023.

The database reveals insights into human preferences for AI-generated images.

Benchmark results highlight the need for improved IQA methods for AI-generated content.

Abstract

In this paper, in order to get a better understanding of the human visual preferences for AIGIs, a large-scale IQA database for AIGC is established, which is named as AIGCIQA2023. We first generate over 2000 images based on 6 state-of-the-art text-to-image generation models using 100 prompts. Based on these images, a well-organized subjective experiment is conducted to assess the human visual preferences for each image from three perspectives including quality, authenticity and correspondence. Finally, based on this large-scale database, we conduct a benchmark experiment to evaluate the performance of several state-of-the-art IQA metrics on our constructed database.