HPSv3: Towards Wide-Spectrum Human Preference Score

TL;DR

HPSv3 introduces a comprehensive human preference dataset and a preference model to evaluate and enhance text-to-image generation, enabling more human-aligned and high-quality image synthesis.

Contribution

The paper presents the first wide-spectrum human preference dataset and a novel preference model, along with an iterative image refinement method that improves quality without additional data.

Findings

HPSv3 is a robust metric for wide-spectrum image evaluation.

CoHP effectively enhances image quality using human preference data.

The dataset and model outperform existing evaluation metrics.

Abstract

Evaluating text-to-image generation models requires alignment with human perception, yet existing human-centric metrics are constrained by limited data coverage, suboptimal feature extraction, and inefficient loss functions. To address these challenges, we introduce Human Preference Score v3 (HPSv3). (1) We release HPDv3, the first wide-spectrum human preference dataset integrating 1.08M text-image pairs and 1.17M annotated pairwise comparisons from state-of-the-art generative models and low to high-quality real-world images. (2) We introduce a VLM-based preference model trained using an uncertainty-aware ranking loss for fine-grained ranking. Besides, we propose Chain-of-Human-Preference (CoHP), an iterative image refinement method that enhances quality without extra data, using HPSv3 to select the best image at each step. Extensive experiments demonstrate that HPSv3 serves as a robust metric for wide-spectrum image evaluation, and CoHP offers an efficient and human-aligned approach to improve image generation quality. The code and dataset are available at the HPSv3 Homepage.