AI Paper Atlas

Research map / Vision-language and multimodal / Visual question answering

Q-SiT: Teaching LMMs for Image Quality Scoring and Interpreting

Zicheng Zhang, Haoning Wu, Ziheng Jia, Weisi Lin, Guangtao Zhai

TPAMI 2026 · Vision-language and multimodal › Visual question answering · approach: Multimodal LLM question answering · 21 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add D1 scoring data (IQA datasets in QA form) (ablation)Q-SiT · scoring (avg SRCC/PLCC)score: 0.800 -> 0.861Improved
add D1 scoring data (ablation)Q-SiT · interpreting (LLVisionQA)accuracy: 0.754 -> 0.756Mixed
add D2 low-level interpreting data (ablation)Q-SiT · interpreting (LLVisionQA)accuracy: 0.721 -> 0.756Improved
add D3 extended conversations (ablation)Q-SiT · interpreting (LLVisionQA)accuracy: 0.744 -> 0.756Improved
add D3 extended conversations (ablation)Q-SiT · scoring (avg SRCC/PLCC)score: 0.859 -> 0.861No clear effect
add coarse-grained data mix ratio prediction (ablation)Q-SiT · interpreting (LLVisionQA)accuracy: 0.732 -> 0.752Improved
add coarse-grained data mix ratio prediction (ablation)Q-SiT · scoring (avg SRCC/PLCC)score: 0.855 -> 0.859Improved
add fine-grained loss-based ratio adjustment on top of CGR (ablation)Q-SiT · interpreting (LLVisionQA)accuracy: 0.752 -> 0.756Improved
add fine-grained loss-based ratio adjustment on top of CGR (ablation)Q-SiT · scoring (avg SRCC/PLCC)score: 0.859 -> 0.861Improved
predict mix ratio with 0.5B model instead of searching on 7BQ-SiT · training costA800 GPU hours: 2,800 -> 72Improved
use Q-SiT scoring-plus-interpreting training instead of interpreting-only Q-InstructQ-SiT 7B · LLVisionQAoverall accuracy: 72.30% -> 75.65%Improved
shrink Q-SiT to 0.5B (Q-SiT-mini)Q-SiT · inferencelatency, VRAM, FLOPs: 49% latency reduction, 25% VRAM, 92% fewer FLOPsImproved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy