AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

NuPlanQA: A Large-Scale Dataset and Benchmark for Multi-View Driving Scene Understanding in Multi-Modal Large Language Models

Sung-Yeon Park, Can Cui, Yunsheng Ma, Ahmadreza Moradipari, Rohit Gupta, Kyungtae Han, Ziran Wang

ICCV 2025 · Vision-language and multimodal › Multimodal LLMs · approach: VLM capability analysis · 28 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 BEV-Fusion module (multi-view, multi-frame) (ablation)LLaMA-3.2-Vision-11B baseline · NuPlanQA-Evalaccuracy total: 65.5 -> 78.7Improved
add BEV-Fusion module (multi-view, multi-frame) (ablation)LLaMA-3.2-Vision-11B baseline · NuPlanQA-Eval spatial relationsaccuracy: 60.8 -> 74.1Improved
add BEV-Fusion module (multi-view, multi-frame) (ablation)LLaMA-3.2-Vision-11B baseline · NuPlanQA-Eval ego-centric reasoningaccuracy: 67.0 -> 81.9Improved
add BEV-Fusion module (multi-view, multi-frame) (ablation)LLaMA-3.2-Vision-11B baseline · NuPlanQA-Eval road env. perceptionaccuracy: 68.7 -> 80.0Improved
add BEV-Fusion with single-view single-frame input (ablation)LLaMA-3.2-Vision-11B baseline · NuPlanQA-Evalaccuracy total: 65.5 -> 71.4Improved
add multi-view images (versus front view only) with BEV-Fusion, single-frame (ablation)BEV-LLM · NuPlanQA-Evalaccuracy total: 71.4 -> 76.2Improved
add historical frames to single-view BEV-Fusion (ablation)BEV-LLM · NuPlanQA-Evalaccuracy total: 71.4 -> 72.4Improved
add historical frames to multi-view BEV-Fusion (ablation)BEV-LLM · NuPlanQA-Evalaccuracy total: 76.2 -> 78.7Improved
use multi-frame instead of single-frame input for GPT-4oGPT-4o · NuPlanQA-Evalaccuracy total: 81.6 -> 81.5No clear effect
use multi-frame instead of single-frame input for Gemini-1.5-ProGemini-1.5-Pro · NuPlanQA-Evalaccuracy total: 79.4 -> 78.1Hurt
use multi-frame instead of single-frame input for LLaVA-OV 7BLLaVA-OV 7B · NuPlanQA-Evalaccuracy total: 75.8 -> 77.8Improved
scale LLaVA-NV from 7B to 32BLLaVA-NV · NuPlanQA-Eval multi-frameaccuracy total: 56.8 -> 71.3Improved

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