AI Paper Atlas

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

PatchPerPix for Instance Segmentation

Lisa Mais, Peter Hirsch, Dagmar Kainmueller

ECCV 2020 · Detection and segmentation › Open-set and few-shot detection · approach: Amodal and open-world instance segmentation · 42 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 consensus affinity computation (ablation)PatchPerPix (ppp+dec) · BBBC010avAP: avAP -0.176 when absent (MWS-Dense)Improved
add patch selection by consensus score (ablation)PatchPerPix (ppp+dec) · BBBC010avAP: avAP - 0.131 when absentImproved
add thinning of the patch selection (ablation)PatchPerPix (ppp+dec) · BBBC010avAP: 0.723 -> 0.727Improved
use decoder with increasing feature maps instead of standard U-Net (ablation)ppp+dec · BBBC010avAP: 0.719 -> 0.727Improved
use mutex watershed instead of connected components for partitioning (ablation)ppp+dec · BBBC010avAP: 0.723 -> 0.727No clear effect
replace separately trained autoencoder (ppp+ae) with joint decoder (ppp+dec)PatchPerPix · BBBC010avAP: 0.617 -> 0.727Improved
use full U-Net encoder with decoder instead of sliding-window encoder-decoder (ed-ppp)PatchPerPix · BBBC010avAP: 0.675 -> 0.727Improved
add decoder over raw patch prediction (ppp+dec vs ppp)PatchPerPix · BBBC010avAP: 0.689 -> 0.727Improved
add decoder over raw patch prediction (ppp+dec vs ppp)PatchPerPix · nuclei3davAPNo clear effect
use PatchPerPix consensus patch assembly instead of dense affinities with MWS (ablation)PatchPerPix · ISBI 2012 EMrRAND: 0.979112 -> 0.988290Improved
use PatchPerPix instead of MWSISBI 2012 EM segmentation · ISBI 2012 EMrRAND: 0.987922 -> 0.988290Improved

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