AI Paper Atlas

Research map / Detection and segmentation / Open-set and few-shot detection

K-convexity shape priors for segmentation

Hossam Isack, Lena Gorelick, Karin Ng, Olga Veksler, Yuri Boykov

ECCV 2018 · Detection and segmentation › Open-set and few-shot detection · approach: Amodal and open-world instance segmentation · 10 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
allow convex parts to overlap (k-convexity) instead of disjoint partsmulti-convexity shape prior · 3D liver segmentation, theta=10 deg, lambda=0.1F1: 0.39 -> 0.57Improved
allow convex parts to overlap (k-convexity) instead of predefined regionsmulti-convexity shape prior · 3D liver segmentation, theta=10 deg, lambda=0.1F1: 0.07 -> 0.57Improved
k-convexity vs k-disjoint at theta=20 degmulti-convexity shape prior · 3D liver segmentation, lambda=1F1: 0.50 -> 0.73Improved
k-convexity vs k-regions at theta=45 degmulti-convexity shape prior · 3D liver segmentation, lambda=0.1F1: 0.84 -> 0.83No clear effect
loosen shape tightness theta from 10 to 45 degrees for k-regions (ablation)k-regions · 3D liver segmentation, lambda=0.1F1: 0.07 -> 0.84Improved
loosen shape tightness theta from 10 to 45 degrees for k-convexity (ablation)k-convexity · 3D liver segmentation, lambda=0.1F1: 0.57 -> 0.83Improved
increase smoothness lambda from 0.1 to 1 for k-disjoint at theta=10 (ablation)k-disjoint · 3D liver segmentationF1: 0.39 -> 0.29Hurt
k-convexity vs k-regions on connected components at theta=10, lambda=1multi-convexity shape prior · 3D liver segmentationconnected components: 5.83 -> 2.67Improved
k-convexity vs k-disjoint on connected components at theta=20, lambda=0.1multi-convexity shape prior · 3D liver segmentationconnected components: 1.25 -> 1.17No clear effect
allow overlapping labels with regular (non-submodular) convexityk-convexity with regular convexity · synthetic proof-of-conceptqualitativeImproved

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