AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

Exploring Efficient Reasoning Segmentation with Small Language Models

Changsong Wen, Zelin Peng, Yu Huang, Xiaokang Yang, Wei Shen

ECCV 2026 · Vision-language and multimodal › Multimodal LLMs · approach: Chain-of-thought multimodal reasoning · 0 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 register tokens to the <SEG> token (ablation)LReSeg · ReasonSeg valcIoU: 47.8 -> 52.8Improved
add register tokens to the <SEG> token (ablation)LReSeg · ReasonSeg testcIoU: 46.6 -> 50.0Improved
add <SEG> token alongside register tokens (ablation)LReSeg · ReasonSeg valcIoU: 48.2 -> 52.8Improved
use learnable instead of fixed register tokens (ablation)LReSeg · ReasonSeg valcIoU: 48.7 -> 52.8Improved
select visual tokens by attention instead of randomly (ablation)LReSeg · ReasonSeg valcIoU: 50.8 -> 52.8Improved
selective injection instead of uniform fusion (ablation)LReSeg · ReasonSeg valcIoU: 50.6 -> 52.8Improved
inject at one selected layer instead of every layer (ablation)LReSeg · ReasonSeg valcIoU: 52.1 -> 52.8Improved
keep causal attention mask (ablation)LReSeg · ReasonSeg valcIoU: 49.5 -> 52.8Improved
keep vision-to-text attention mask (ablation)LReSeg · ReasonSeg valcIoU: 52.2 -> 52.8Improved
use DINO unified encoder instead of SigLIP (ablation)LReSeg · ReasonSeg valcIoU: 50.5 -> 52.8Improved
use DINO unified encoder instead of CLIP (ablation)LReSeg · ReasonSeg valcIoU: 49.4 -> 52.8Improved
use unified DINO encoder instead of DINO & SAM (ablation)LReSeg · ReasonSeg valcIoU: 51.2 -> 52.8Improved

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