AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

Visual Saliency Transformer

Nian Liu, Ni Zhang, Kaiyuan Wan, Ling Shao, Junwei Han

ICCV 2021 · Architectures and efficiency › Attention and transformers · approach: Efficient and robust vision transformers · 634 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
add cross-modality transformer fusion instead of concatenation (ablation)RGB-D VST baseline · LFSDS-measure: 0.842 -> 0.849Improved
add cross-modality transformer fusion instead of concatenation (ablation)RGB-D VST baseline · DUTLF-DepthS-measure: 0.889 -> 0.889No clear effect
add bilinear upsampling to full resolution (ablation)+CMT · NJUDS-measure: 0.873 -> 0.906Improved
replace bilinear upsampling with RT2T token upsampling (ablation)+CMT+Bili · NJUDS-measure: 0.906 -> 0.915Improved
replace bilinear upsampling with RT2T token upsampling (ablation)+RC+Bili · DUTSS-measure: 0.867 -> 0.881Improved
add multi-level token fusion of low-level tokens (ablation)+CMT+RT2T · STERES-measure: 0.896 -> 0.910Improved
add multi-level token fusion of low-level tokens (ablation)+RC+RT2T · DUTSS-measure: 0.881 -> 0.895Improved
add token-based multi-task decoder with boundary detection (ablation)+CMT+RT2T+F · NJUDS-measure: 0.923 -> 0.922No clear effect
add token-based multi-task decoder with boundary detection (ablation)+CMT+RT2T+F · LFSDS-measure: 0.876 -> 0.882Improved
add token-based multi-task decoder with boundary detection (ablation)+RC+RT2T+F · DUTSS-measure: 0.895 -> 0.896Improved
use task tokens decoder instead of conventional two-stream decoder (ablation)+CMT+RT2T+F · LFSDS-measure: 0.874 -> 0.882Improved
use task tokens decoder instead of conventional two-stream decoder (ablation)+CMT+RT2T+F · MACsMACs (G): 28.27 -> 17.70Improved

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