AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

DGSeg: Dynamic Gating of Semantic-Spatial Guided Predictions for Reasoning Segmentation

Ruizhe Zeng, Siyu Cao, Lu Zhang, Zhi-yong Liu

ECCV 2026 · Vision-language and multimodal › Multimodal LLMs · approach: Spatial reasoning in VLMs · 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
fuse separate semantic and spatial branch predictions with dynamic gating instead of joint cue encoding (no fusion) (ablation)DGSeg (Qwen2.5-VL-3B) · ReasonSeg valgIoU: 60.8 -> 66.0Improved
use learnable fusion instead of average fusion (ablation)DGSeg · ReasonSeg valgIoU: 63.8 -> 66.0Improved
learned scalar weight instead of non-learnable confidence weighting (ablation)DGSeg · ReasonSeg val/testgIoU: val 64.1 -> 64.0; test 57.2 -> 57.8Mixed
replace dynamic gating with MoE fusion (ablation)DGSeg · ReasonSeg val/testgIoU: 66.0 -> 65.6 (val), 60.0 -> 60.6 (test)No clear effect
add dual-branch gating on top of single best branch (ablation)DGSeg · ReasonSeg valgIoU: 63.4 (spatial-only) -> 66.0Improved
add RL finetuning (Stage 1) of the MLLM (ablation)DGSeg · ReasonSeg valgIoU: 36.0 -> 66.0Improved
add semantic reward based on SAM3 mask IoU (ablation)Stage 1 RL · ReasonSeg valgIoU: 62.1 -> 66.0Improved
use segmentation-model preference reward instead of category-label reward (ablation)Stage 1 RL · ReasonSeg valgIoU: 64.7 -> 66.0Improved
add spatial (box IoU) reward (ablation)Stage 1 RL · ReasonSeg valgIoU: 64.1 -> 66.0Improved
add format reward (ablation)Stage 1 RL · ReasonSeg valgIoU: 64.6 -> 66.0Improved
add dual-branch dynamic gating overheadDGSeg · efficiencyFLOPs / FPS: +0.3% FLOPs, -2.0% FPS, +4.2% memoryNo clear effect
scale MLLM from 3B to 7BDGSeg · ReasonSeg testgIoU: 60.0 -> 67.3Improved

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