1
Return

Edge-aware prompt-driven multi-task network for dense visual prediction

delete2026-08-04
delete0
PRE
AI
H
Huilan Luo *
M
Minghao Yu
DOI：10.1016/j.asoc.2026.116082delete
deleteOriginal
deleteOriginal request for help
deleteShare
deleteSave
Abstract

Abstract

En 中文
• Proposes EATP-MTL, a Transformer-based framework for dense visual multi-task learning. • Introduces an Edge-Enhanced Attention Block to preserve structural and spatial details. • Designs a Prompt-Guided Attention Module for selective and task-aware feature fusion. • Develops a Task-Specific Enhancement Block to refine and decouple task representations. • Achieves superior performance on NYUD-v2 and PASCAL-Context benchmarks.

Journal

Applied Soft Computing cover
Applied Soft Computing
IF:
6.6
Papers:
1.4W
Citations:
4.8W

Organization

No organization information available
Cited Papers

Cited Papers

Citing Papers

Citing Papers