AI Paper Atlas

Research map / Vision-language and multimodal / Human and embodied modeling

Learning What Helps: Task-Aligned Context Selection for Vision Tasks

Jingyu Guo, Emir Konuk, Fredrik Strand, Christos Matsoukas, Kevin Smith

CVPR 2026 · Vision-language and multimodal › Human and embodied modeling · approach: Scene graph generation · 0 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
learn task-aligned context selection instead of frozen DINO similarity retrievalViT-S/16 classifier · 11 fine-grained datasets (average)accuracy: 86.6 -> 88.3Improved
learn task-aligned context selection instead of frozen DINO similarityViT-S/16 classifier · 4 medical classification datasets (average)average score: 92.0 -> 92.9Improved
learn task-aligned context selection instead of frozen DINO similarityViT-S/16 with gated cross-attention · Kvasir-SEGIoU: 80.0 -> 81.1Improved
pair query with random context imageViT-S/16 classifier · Fine-grained averageaccuracy: 86.3 -> 86.0Hurt
provide blank image as context (ablation)TACS architecture · 15 classification datasetsaverage score: 87.6 -> 87.5No clear effect
provide duplicate of query as context (ablation)TACS architecture · 15 classification datasetsaverage score: 87.6 -> 87.7No clear effect
provide noisy query as context (ablation)TACS architecture · 15 classification datasetsaverage score: 87.6 -> 87.8No clear effect
use frozen DINOv3 similar-image retrieval as context (ablation)TACS architecture · 15 classification datasetsaverage score: 87.6 -> 88.0Improved
add learned TACS selection over frozen similarity retrieval (ablation)TACS architecture · 15 classification datasetsaverage score: 88.0 -> 89.5Improved
add policy-gradient path to differentiable selection (ablation)TACS selector · Fine-grained classificationaverage score: 86.7 -> 88.3Improved
add Gumbel-Softmax differentiable path to policy-only selection (ablation)TACS selector · Fine-grained classificationaverage score: 87.2 -> 88.3Improved
differentiable selection only vs frozen retrieval (ablation)TACS selector · Medical classificationaverage score: 92.0 -> 92.1No clear effect

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