AI Paper Atlas

Research map / Multi-agent systems and games / Human-AI and goal-conditioned

RefAV: Towards Planning-Centric Scenario Mining

Cainan Davidson, Deva Ramanan, Neehar Peri

CVPR 2026 · Multi-agent systems and games › Human-AI and goal-conditioned · approach: Data-driven traffic simulation · 11 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
filter tracks by referred class instead of labelling all tracks referredground-truth tracks · RefAV scenario miningHOTA-Temporal: 13.3 -> 21.4Improved
use program synthesis (RefProg) instead of LLM API as black boxLe3DE2E tracks · RefAV scenario miningHOTA-Temporal: 37.2 -> 50.1Improved
use LLM black box instead of ReferGPTLe3DE2E tracks · RefAV scenario miningHOTA-Temporal: 20.2 -> 37.2Improved
use image-embedding similarity instead of class filtering on predicted tracksLe3DE2E tracks · RefAV scenario miningHOTA-Temporal: 19.2 -> 17.2Hurt
use image-embedding similarity instead of class filtering on oracle tracksground-truth tracks · RefAV scenario miningHOTA-Temporal: 21.4 -> 24.6Improved
replace ground-truth tracks with predicted Le3DE2E tracksRefProg · RefAV scenario miningHOTA-Temporal: 64.8 -> 50.1Hurt
use Claude 3.7 Sonnet instead of Qwen-2.5-7B-Instruct for code generation (ablation)RefProg · RefAV scenario mining (RefProg LLM ablation)HOTA-Temporal: 31.6 -> 50.1Improved
use Claude 3.7 Sonnet instead of Gemini 2.0 Flash for code generation (ablation)RefProg · RefAV scenario mining (RefProg LLM ablation)HOTA-Temporal: 45.2 -> 50.1Improved
use Claude 3.5 Sonnet instead of Qwen-2.5-7B-Instruct for code generation (ablation)RefProg · RefAV scenario mining (RefProg LLM ablation)program failure rate: 18.1 -> 0.5Improved
swap tracker from Le3DE2E to ReVoxelDetRefProg · RefAV scenario miningHOTA-Temporal: 50.1 -> 43.9Hurt
subsample tracker inputs from 10 Hz to 2 Hz (ablation)RefProg · RefAV scenario miningHOTA-TemporalNo clear effect
use RefProg zero-shot instead of trained PromptTrack on nuPromptPromptTrack (PETR) · nuPrompt (zero-shot RefProg)AMOTA: 0.259 -> 0.321Improved

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