Research map / Vision-language and multimodal / Robot action and navigation
Agentic Collaborative Cognition for Zero-Shot 3D Understanding
Wenxin Wang, Bo Zhang, Feng Chen, Zixuan Wang, Wen Li, Changsheng Li, Yinjie Lei
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add object information table to the BEV cognitive map (ablation) | Ours · Nr3D | Acc@0.25: -8.9% Acc@0.25 when removed | Improved |
| add object information table to the BEV cognitive map (ablation) | Ours · SQA3D | EM: -3.9% EM when removed | Improved |
| add candidate filtering and feedback (ablation) | Ours · Nr3D | Acc@0.25: -7.4% Acc@0.25 when disabled | Improved |
| add rendered images to real images (ablation) | Ours · ScanQA | EM: -5.1% EM without rendered images | Improved |
| add real images to rendered images (ablation) | Ours · ScanQA | EM: -4.7% EM without real images | Improved |
| use cognitive-map viewpoint planning instead of random viewpoints (ablation) | Ours · Nr3D | Acc@0.25: +42.9% Acc@0.25 | Improved |
| use cognitive-map viewpoint planning instead of co-visible template (ablation) | Ours · Nr3D | Acc@0.25: +13.6% Acc@0.25 | Improved |
| increase Perception Agents from 1 to 3 (ablation) | Ours · Nr3D (250 queries) | Acc@0.25: 65.6 -> 66.8 | Improved |
| increase planned viewpoints K from 8 to 9 (ablation) | Ours · Nr3D, ScanRefer, SQA3D | accuracy | Hurt |
| raise viewpoint matching threshold above 0.8 (ablation) | Ours · multiple benchmarks | accuracy | Hurt |
| replace Qwen2.5-VL-72B with GPT-4o (ablation) | Ours · Nr3D | Acc@0.25: 56.7 -> 61.2 | Improved |
| shrink MLLM from Qwen2.5-VL-72B to 7B (ablation) | Ours · Nr3D | Acc@0.25: 56.7 -> 49.3 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.
- SeeGround — SeeGround: See and Ground for Zero-Shot Open-Vocabulary 3D Visual Grounding (CVPR 2025) · better on 3 of 3 shared results
- SPAZER — SPAZER: Spatial-Semantic Progressive Reasoning Agent for Zero-shot 3D Visual Grounding (NeurIPS 2025) · better on 3 of 3 shared results
- Agent3D-Zero — Agent3D-Zero: An Agent for Zero-shot 3D Understanding (ECCV 2024) · better on 2 of 2 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
- IR3D-Bench: Evaluating Vision-Language Model Scene Understanding as Agentic Inverse Rendering (NeurIPS 2025 · similarity 0.89)
- Visual Agentic AI for Spatial Reasoning with a Dynamic API (CVPR 2025 · similarity 0.88)
- Pursuing Minimal Sufficiency in Spatial Reasoning (ICLR 2026 · similarity 0.88)
- An Extensible Framework for Open Heterogeneous Collaborative Perception (ICLR 2024 · similarity 0.88)
- D3D-VLP: Dynamic 3D Vision-Language-Planning Model for Embodied Grounding and Navigation (CVPR 2026 · similarity 0.87)
- Agent-X: Evaluating Deep Multimodal Reasoning in Vision-Centric Agentic Tasks (ICLR 2026 · similarity 0.87)
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.
