SOURCE-LINKED INTELLIGENCE

Task-State Adaptation with Prototype Memory for Multi-Task Dense Prediction

arXiv · AI, language, vision and robotics · article · Aug 28, 2026 · UTC

Vision foundation backbones provide strong representations for dense prediction, yet a single shared feature still needs to support tasks with different, image-dependent adaptation requirements. We propose MemMTL, a multi-task dense prediction framework that estimates a compact task state from global visual context and refines it through a learnable task-state prototype memory. The refined state is converted into task-conditioned expert logits and combined with token-level logits before sparse top-$k$ selection over a local expert bank shared by all tasks. A separate task-agnostic residual ban

Read original source ↗ Open in workspace

recordType

paper

region

Global