AI Paper Atlas

Research map / Video / Video segmentation and tracking

Dynamic Context-Sensitive Filtering Network for Video Salient Object Detection

Miao Zhang, Jie Liu, Yifei Wang, Yongri Piao, Shunyu Yao, Wei Ji, Jingjing Li, Huchuan Lu, Zhongxuan Luo

ICCV 2021 · Video › Video segmentation and tracking · approach: Video restoration with temporal priors · 128 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
generate kernels by affinity matrix multiplication instead of stacked convolutions (ablation)DCFNet · DAVISmax F-measure: 0.889 -> 0.900Improved
add global attention to fuse multi-scale branches (ablation)DCFNet · DAVISmax F-measure: 0.895 -> 0.900Improved
use three dilation rates 1, 3, 5 instead of dilation rate 1 only (ablation)DCFNet · DAVISmax F-measure: 0.893 -> 0.900Improved
use three dilation rates instead of dilation rate 3 only (ablation)DCFNet · DAVISmax F-measure: 0.885 -> 0.900Improved
use three dilation rates instead of dilation rate 5 only (ablation)DCFNet · ViSalMAE: 0.012 -> 0.010Improved
make DCFM bidirectional instead of unidirectional (ablation)DCFNet · DAVISMAE: 0.017 -> 0.016Improved
make DCFM bidirectional instead of unidirectional (ablation)DCFNet · SegV2MAE: 0.017 -> 0.015Improved
add temporal branch to spatial-only model (ablation)DCFNet · DAVISmax F-measure: 0.887 -> 0.900Improved
add spatial branch to temporal-only model (ablation)DCFNet · DAVISmax F-measure: 0.867 -> 0.900Improved
use gated weighted fusion instead of element-wise addition (ablation)DCFNet · DAVISmax F-measure: 0.881 -> 0.900Improved
apply DCFM at all four decoder stages (ablation)DCFNet · DAVISS-measure and MAEImproved
append DCFM to CPD to add temporal modelling (ablation)CPD+F · DAVISMAE and max F-measure: 25.7% and 2.8%Improved

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