AI Paper Atlas

Research map / Detection and segmentation / CNN features

Linear Span Network for Object Skeleton Detection

Chang Liu, Wei Ke, Fei Qin, Qixiang Ye

ECCV 2018 · Detection and segmentation › CNN features · approach: Box regression and loss design · 35 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 subspace linear span (3 subspaces) on top of feature linear span (ablation)LSN · SK-LARGEF-measure (%): 49.53 -> 66.15Improved
span 3 subspaces instead of 2 (SRN) (ablation)LSN · SK-LARGEF-measure (%): 65.88 -> 66.15Improved
span 4 subspaces instead of 3 (ablation)LSN · SK-LARGEF-measure (%): 66.15 -> 65.89Hurt
add iterative training with resolution alignment (iter1) (ablation)LSN_3 · SK-LARGEF-measure (%): 66.15 -> 66.82Improved
train end-to-end instead of iterative strategy (ablation)LSN_3 · SK-LARGEF-measure (%): 66.82 -> 66.63Hurt
use more training iterations (iter2, iter3) (ablation)LSN_3 · SK-LARGEF-measure (%): 66.82 -> 66.74 -> 66.68No clear effect
use LSN instead of SRNskeleton detector · SK-LARGEF-measure: 0.655 -> 0.668Improved
use LSN instead of SRNskeleton detector · WH-SYMMAXF-measure: 0.780 -> 0.797Improved
use LSN instead of SRNskeleton detector · SK-SMALLF-measure: 0.632 -> 0.633No clear effect
use LSN instead of SRNskeleton detector · SYMMAXF-measure: 0.446 -> 0.480Improved
use LSN instead of SRNskeleton detector · Sym-PASCALF-measure: 0.443 -> 0.425Hurt
use LSN instead of HEDedge detector · BSDS500ODS: 0.780 -> 0.790Improved

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