AI Paper Atlas

Research map / Detection and segmentation / CNN features

Defocus Blur Detection via Multi-Stream Bottom-Top-Bottom Fully Convolutional Network

Wenda Zhao, Fan Zhao, Dong Wang, Huchuan Lu

CVPR 2018 · Detection and segmentation › CNN features · approach: Context for hard detection conditions · 87 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
Use BTBNet instead of a VGG16 fully convolutional network (ablation)DBD network · Shi’s datasetF-measure: 0.797 -> 0.853Improved
Use BTBNet instead of a VGG16 fully convolutional network (ablation)DBD network · Our datasetF-measure: 0.643 -> 0.730Improved
Add multi-scale streams and FNet, using two or three streams rather than one (ablation)BTBNet · Shi’s datasetF-measure: 0.853 -> 0.854Improved
Add multi-scale streams and FNet, using two or three streams rather than one (ablation)BTBNet · Our datasetF-measure: 0.730 -> 0.732Improved
Use three streams instead of two (ablation)BTBNet+FNet · Shi’s datasetF-measure: 0.853 -> 0.854Improved
Use three streams instead of two (ablation)BTBNet+FNet · Our datasetF-measure: 0.731 -> 0.732Improved
Add a fourth stream at scale 0.4 instead of using three streams (ablation)BTBNet+FNet · Shi’s datasetF-measure: 0.854 -> 0.853Hurt
Add a fourth stream at scale 0.4 instead of using three streams (ablation)BTBNet+FNet · Our datasetF-measure: 0.732 -> 0.731Hurt
Add RRNet recurrent reconstruction (ablation)BTBNet(3S)+FNet · Shi’s datasetF-measure: 0.854 -> 0.867Improved
Add RRNet recurrent reconstruction (ablation)BTBNet(3S)+FNet · Our datasetF-measure: 0.732 -> 0.761Improved
Add RRNet recurrent reconstruction (ablation)BTBNet(3S)+FNet · Shi’s datasetMAE: 0.130 -> 0.107Improved
Add RRNet recurrent reconstruction (ablation)BTBNet(3S)+FNet · Our datasetMAE: 0.230 -> 0.194Improved

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