Understanding Pure Textual Reasoning for Blind Image Quality Assessment
Yuan Li, Shin’ya Nishida · 2026
Read Understanding Pure Textual Reasoning for Blind Image Quality Assessment by Yuan Li, Shin’ya Nishida (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2601-02441.txt.
Description
Yuan Li and Shin'ya Nishida ask what written reasoning actually contributes to judging image quality without a reference picture, a task called blind image quality assessment. Recent systems let a model describe an image in words and reason toward a score. But it was unclear how much of the score the words alone can carry, and where the image itself remains essential.
They compare existing systems with three designs for learning the image, text, and score relationship, based on chain of thought, self consistency, and an autoencoder. Scores drop sharply when only text is used. Chain of thought adds little, self consistency closes most of the gap between text based and image based prediction. And the autoencoder design is weaker at closing the gap but points to a new direction for improvement.
Catalog record
| CREATOR | Yuan Li, Shin’ya Nishida |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 26 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2601-02441