SA-IQA: Redefining Image Quality Assessment for Spatial Aesthetics with Multi-Dimensional Rewards

2025-12-04

AI Summary

Existing image quality assessment (IQA) methods lack systematic aesthetic evaluation tailored to AI-generated indoor scenes. Method: We propose Spatial Aesthetics—a novel paradigm modeling indoor image quality along four dimensions: layout, harmony, illumination, and geometric distortion. Based on this, we construct SA-BENCH, the first large-scale benchmark comprising 12K high-quality indoor images with fine-grained human annotations. Leveraging SA-BENCH, we design a multi-dimensional reward fusion mechanism and jointly optimize the SA-IQA framework via multimodal large language model (MLLM) fine-tuning and Generalized Reinforcement Learning from Preference Optimization (GRPO). We further integrate a Best-of-N sampling strategy to enhance AIGC generation quality. Contribution/Results: Experiments show SA-IQA significantly outperforms state-of-the-art IQA methods on SA-BENCH, achieving an average 18.7% improvement in Pearson correlation. SA-IQA effectively guides generative model optimization. The code and dataset will be publicly released.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Search and Optimization: Sampling/Simulation-based Search
• Computer Vision: Diffusion Models for Vision

Application Category

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

Abstract

In recent years, Image Quality Assessment (IQA) for AI-generated images (AIGI) has advanced rapidly; however, existing methods primarily target portraits and artistic images, lacking a systematic evaluation of interior scenes. We introduce Spatial Aesthetics, a paradigm that assesses the aesthetic quality of interior images along four dimensions: layout, harmony, lighting, and distortion. We construct SA-BENCH, the first benchmark for spatial aesthetics, comprising 18,000 images and 50,000 precise annotations. Employing SA-BENCH, we systematically evaluate current IQA methodologies and develop SA-IQA, through MLLM fine-tuning and a multidimensional fusion approach, as a comprehensive reward framework for assessing spatial aesthetics. We apply SA-IQA to two downstream tasks: (1) serving as a reward signal integrated with GRPO reinforcement learning to optimize the AIGC generation pipeline, and (2) Best-of-N selection to filter high-quality images and improve generation quality. Experiments indicate that SA-IQA significantly outperforms existing methods on SA-BENCH, setting a new standard for spatial aesthetics evaluation. Code and dataset will be open-sourced to advance research and applications in this domain.

Problem

• Assesses interior image aesthetics across layout, harmony, lighting, and distortion dimensions
• Evaluates current IQA methods and develops a comprehensive reward framework for spatial aesthetics
• Applies the framework to optimize AIGC generation and filter high-quality interior images

Innovation

• Multi-dimensional fusion approach for spatial aesthetics
• MLLM fine-tuning for comprehensive reward framework
• GRPO reinforcement learning integration for AIGC optimization