Understanding Pure Textual Reasoning for Blind Image Quality Assessment

TL;DR

This paper investigates how textual reasoning influences blind image quality assessment, revealing that current models rely heavily on image data and that certain paradigms can better bridge the gap between image and text-based predictions.

Contribution

It compares different textual reasoning paradigms in BIQA, highlighting the effectiveness of Self-Consistency in reducing prediction gaps and providing insights for future improvements.

Findings

Self-Consistency paradigm significantly reduces image-text prediction gap

Existing models perform poorly with only textual information

Autoencoder paradigm offers directions for further optimization

Abstract

Textual reasoning has recently been widely adopted in Blind Image Quality Assessment (BIQA). However, it remains unclear how textual information contributes to quality prediction and to what extent text can represent the score-related image contents. This work addresses these questions from an information-flow perspective by comparing existing BIQA models with three paradigms designed to learn the image-text-score relationship: Chain-of-Thought, Self-Consistency, and Autoencoder. Our experiments show that the score prediction performance of the existing model significantly drops when only textual information is used for prediction. Whereas the Chain-of-Thought paradigm introduces little improvement in BIQA performance, the Self-Consistency paradigm significantly reduces the gap between image- and text-conditioned predictions, narrowing the PLCC/SRCC difference to 0.02/0.03. The Autoencoder-like paradigm is less effective in closing the image-text gap, yet it reveals a direction for further optimization. These findings provide insights into how to improve the textual reasoning for BIQA and high-level vision tasks.