AI Paper Atlas

Research map / Video / Video-language understanding

SARL-STG: A Spatially Aware Reinforcement Learning Framework for Refining MLLMs in Spatio-Temporal Video Grounding

Hong Gao, Xiangkai Xu, Bin Zhong, Junjie Yin, Fangyu Kang, Yutong Xu, Xiugang Dong, Xurui Gao, Min-Ling Zhang

CVPR 2026 · Video › Video-language understanding · approach: Long video understanding

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 spatio-temporal dynamic reward in GRPO stage on top of joint SFT (ablation)SARL-STG · HCSTVGm tIoU: 58.6 -> 64.2Improved
use dynamic tIoU-weighted reward instead of plain spatio-temporal reward (ablation)SARL-STG GRPO · VidSTGm tIoU: 51.5 -> 52.3Improved
add spatial knowledge to reward (spatio-temporal reward vs temporal-only reward) (ablation)SARL-STG GRPO · HCSTVGm tIoU: 62.3 -> 63.8Improved
switch from temporal-only SFT to joint spatio-temporal SFT (ablation)SARL-STG Stage 1 · HCSTVGm tIoU: 62.5 -> 58.6Hurt
switch from temporal-only SFT to joint spatio-temporal SFT (ablation)SARL-STG Stage 1 · VidSTGm tIoU: 47.7 -> 50.5Improved
add GRPO with temporal reward after SFT (ablation)SARL-STG · VidSTGm tIoU: 50.5 -> 51.3Improved
add Temporal Enhancement Module and Global-Local Feature Fusion Module to spatial grounding (ablation)GroundingDINO · HCSTVG, GT inputm sIoU: 65.6 -> 72.1Improved
add Temporal Enhancement Module alone (ablation)GroundingDINO · HCSTVG, GT inputm sIoU: 65.6 -> 71.0Improved
shift input clip randomly by 40% from ground truth (ablation)spatial grounding module · HCSTVGm sIoU: 72.1 -> 66.4Hurt
shift input clip randomly by 20% from ground truth (ablation)spatial grounding module · HCSTVGm sIoU: 72.1 -> 70.3Hurt
train on STVG-Wild instead of STVG data only (ablation)SARL-STG · ST-Align (OOD)m tIoU: 42.6 -> 49.5Improved
add TVG data to STVG training data (ablation)SARL-STG · ActivityNet (OOD)m tIoU: 28.5 -> 33.4Improved

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