AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

Zero Pixel Directional Boundary by Vector Transform

Edoardo Mello Rella, Ajad Chhatkuli, Yun Liu, Ender Konukoglu, Luc Van Gool

ICLR 2022 · Label-efficient and robust learning › Recognition and long-tail · approach: Loss design and classifier theory · 1 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 vector-field boundary representation instead of binary boundary classificationBoundary detection with the shared network architecture · Cityscapesassd: 4.71 -> 4.33Improved
use vector-field boundary representation instead of binary boundary classificationBoundary detection with the shared network architecture · CityscapesODS F: 0.764 -> 0.845Improved
use vector-field boundary representation instead of binary boundary classificationBoundary detection with the shared network architecture · Synthiaassd: 1.91 -> 1.67Improved
use vector-field boundary representation instead of binary boundary classificationBoundary detection with the shared network architecture · SynthiaODS F: 0.900 -> 0.819Hurt
use vector-field boundary representation instead of binary boundary classificationBoundary detection with the shared network architecture · Mapillary Vistasassd: 4.35 -> 3.60Improved
use vector-field boundary representation instead of binary boundary classificationBoundary detection with the shared network architecture · Mapillary VistasODS F: 0.760 -> 0.806Improved
use vector-field boundary representation instead of binary boundary classificationBoundary detection with the shared network architecture · BSDS500assd: 5.66 -> 5.83Hurt
use vector-field boundary representation instead of binary boundary classificationBoundary detection with the shared network architecture · BSDS500ODS F: 0.688 -> 0.676Hurt
replace vector-field prediction with distance-transform prediction trained using L1 loss (ablation)VT boundary detection · Cityscapesassd: 4.33 -> 5.63Hurt
replace vector-field prediction with distance-transform prediction trained using L1 loss (ablation)VT boundary detection · Synthiaassd: 1.67 -> 3.74Hurt
replace vector-field prediction with distance-transform prediction trained using L1 loss (ablation)VT boundary detection · Mapillary Vistasassd: 3.60 -> 7.13Hurt
replace vector-field prediction with distance-transform prediction trained using L1 loss (ablation)VT boundary detection · BSDS500assd: 5.83 -> 5.87No 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