AI Paper Atlas

Research map / Multi-agent systems and games / LLM agents

R^2-Bench: Benchmarking the Robustness of Referring Perception Models under Perturbations

Xiang Li, Kai Qiu, Jinglu Wang, Xiaohao Xu, Kashu Yamazaki, Hao Chen, Rita Singh, Xiaonan Huang, Bhiksha Raj

ECCV 2024 · Multi-agent systems and games › LLM agents · approach: Agentic benchmarks and orchestration · 3 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
use polygon-vertex sequential mask prediction (PolyFormer) instead of per-pixel classificationreferring image segmentation models · RefCOCO/+/g under perturbationmIoU and APC: RefCOCO APC -0.12 vs LAVT -0.21Improved
multi-task joint training (SEEM)referring image segmentation · RefCOCOAPC: clean mIoU 0.65 but APC -0.12Mixed
decoupled visual and ID embeddings (DeAOT) instead of pixel/object correspondence (Cutie)video object segmentation · YTVOSG APC: G_s APC -2.7 (DeAOT) vs -4.1 (Cutie)Improved
unseen object categories instead of seenVOS models · YTVOSperformance drop: AOT APC -5.4 seen vs -7.7 unseenHurt
frame-by-frame feature processing (OnlineRefer) instead of relying on temporal correspondencereferring VOS with image-level backbone · Ref-YTVOSJ APC: -9.4 vs ReferFormer -17.4Improved
quantization and local-global distillation for acoustic representation (QSD)audiovisual segmentation · AVS-s4F APC: F APC -7.9Improved
frost perturbation vs color-based (brightness, saturate) perturbations (ablation)perturbation types · visual perturbationsperformance degradationHurt
acoustic perturbations on AVS (ablation)AVS models · AVS-s4/ms3performance: e.g. AVSegFormer 81.5-81.6 across most typesNo clear effect
impulse response acoustic perturbation (ablation)AVS models · AVS-s4performance: QSD 79.1 under IR vs about 81.5 for othersHurt
dynamic (frame-varying) instead of static perturbation (ablation)referring VOS models · Ref-DAVISperformance: ReferFormer Ref-YTVOS J -13.7 static vs -17.8 dynamicHurt
increase R2-Agent data verification iterations from 0 to 2 (ablation)R2-Agent data sample selection · agent data selection user studyACC: 0.53 -> 0.77Improved
increase verification iterations from 2 to 3 (ablation)R2-Agent perturbation selection · agent data selection user studyACC: 0.86 -> 0.87No clear effect

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