AI Paper Atlas

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

Faster Multi-Object Segmentation Using Parallel Quadratic Pseudo-Boolean Optimization

Niels Jeppesen, Patrick M. Jensen, Anders N. Christensen, Anders B. Dahl, Vedrana A. Dahl

ICCV 2021 · Detection and segmentation › Open-set and few-shot detection · approach: Amodal and open-world instance segmentation · 4 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 bottom-up merging with a single thread (P-QPBO(1) instead of K-QPBO)QPBO · Nerve segmentation N2solve time: 4,987 s -> 2,429 s (51% reduction)Improved
add bottom-up merging with a single thread (P-QPBO(1) instead of K-QPBO)QPBO · Nerve segmentation N1solve time: 836 s -> 558 s (33% reduction)Improved
use compact data structures (M-QPBO instead of K-QPBO)K-QPBO · Nerve segmentation N1solve time and memory: 836 s -> 628 s (25% reduction); 134.2 GB -> 60.1 GBImproved
use compact data structures (M-QPBO instead of K-QPBO)K-QPBO · Nerve segmentation N2solve time and memory: 4,987 s -> 3,704 s; 306.8 GB -> 182.7 GBImproved
increase threads from 1 to 16P-QPBO · Nerve segmentation N1solve time: 558 s -> 94 sImproved
increase threads from 32 to 40P-QPBO · Nerve segmentation N1solve time: 80 s -> 74 sImproved
increase threads from 40 to 48P-QPBO · Nerve segmentation N1solve time: 74 s -> 76 sNo clear effect
increase threads from 40 to 48P-QPBO · Nerve segmentation N2solve time: 245 s -> 239 sNo clear effect
add the two-stage QPBO strategy to parallel merging (P-QPBO instead of Liu-Sun maxflow) (ablation)P-QPBO · Nerve segmentation N1speed-up vs K-QPBO: 6.07 -> 10.47Improved
use P-QPBO instead of parallel EIBFS with bottom-up merging (ablation)P-EIBFS · BBBC038v1 S1mean speed-up vs K-QPBO: 0.68 -> 2.96Improved
use P-QPBO instead of serial EIBFS maxflow on full graph (ablation)EIBFS · Nerve segmentation N1speed-up vs K-QPBO: 1.08 -> 10.47Improved
use P-QPBO instead of best-case EIBFS-QPBO estimate (ablation)EIBFS-QPBO · BBBC038v1 S1mean speed-up vs K-QPBO: 0.66 -> 2.96Improved

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