AI Paper Atlas

Research map / Detection and segmentation / 3D detection and driving

Virtual Task Prompting for Multi-Task Scene Understanding

Zedong Wang, Yucheng Wang, Dan Xu

NeurIPS 2026 · Detection and segmentation › 3D detection and driving · approach: LiDAR perception for driving · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Proposes Virtual Task Prompting (VTP), which casts multi-task prompting as retrieving task-adaptive subsets from a shared virtual state to enable multi-task scene understanding on off-the-shelf vision foundation models.

Problem

Multi-task scene understanding models typically need modified vision encoders or complex decoders for cross-task interaction, limiting flexibility and compatibility with off-the-shelf vision foundation models.

Approach

Frames multi-task prompting as retrieving task-adaptive subsets from a shared virtual state, where both the state and access mechanism are learned, so cross-task interaction happens without modifying the vision encoder or using complex decoders.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
replace TaskPrompter prompting with Virtual Task Prompting (ImageNet ViT-L)TaskPrompter · Pascal-ContextDelta_p: 0.00 -> +2.77Improved
use DINOv3 backbone with VTPVTP-timm · Pascal-ContextDelta_p: +2.77 -> +5.15Improved
use DINOv3 instead of ImageNet ViT with VTPVTP-timm · Pascal-Context normalmErr: 13.56 -> 13.85Hurt
add virtual tasks V=2 to layer-wise routing (V=0) (ablation)VTP · Pascal-ContextDelta_p: +1.52 -> +2.77Improved
increase virtual tasks beyond V=2 (V=4, V=8) (ablation)VTP · Pascal-ContextDelta_p: +2.77 -> +2.41 / +2.43No clear effect
double decoder parameters (ablation)TaskPrompter · Pascal-ContextDelta_p: -0.18Hurt
increase task vectors per task (2x, 4x) (ablation)TaskPrompter · Pascal-ContextDelta_p: +0.09 / -0.14No clear effect
use VTP-timm instead of TaskPrompter on 13 Taskonomy tasksTaskPrompter · TaskonomyDelta_p: +4.50 overall; Edge 3D L1 0.0819 -> 0.0948Mixed
use ViT-5 backbone with VTP on Taskonomy normalsVTP · Taskonomy normalmErr: 29.046 -> 33.060Hurt
use VTP instead of TaskPrompter for efficiencyTaskPrompter · efficiencyparameters/throughput: 401M -> 342M; 14.394 -> 16.318 imgs/sImproved

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