HPSv3: Towards Wide-Spectrum Human Preference Score
This work addresses the need for better human-aligned evaluation metrics in text-to-image generation, offering a dataset and method that improve over existing limited approaches, though it is incremental in building on prior human preference scoring.
The authors tackled the problem of evaluating text-to-image generation models by introducing HPSv3, which includes a wide-spectrum dataset (HPDv3 with 1.08M text-image pairs and 1.17M comparisons) and a VLM-based preference model, resulting in a robust metric and an iterative refinement method (CoHP) that enhances image quality without extra data.
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.