AI Paper Atlas

Research map / Detection and segmentation / CNN features

Corner Detection Using Second-Order Generalized Gaussian Directional Derivative Representations

Weichuan Zhang, Changming Sun

TPAMI 2019 · Detection and segmentation › CNN features · approach: Box regression and loss design · 50 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 SOGGDD proposed detector over ACJ and ANDDcorner detection · 'Geometric' imagetotal missed plus false corners: 28 -> 18Improved
use SOGGDD proposed detector over ANDDcorner detection · 'Lab' imagetotal missed plus false corners: 156 -> 141Improved
use SOGGDD proposed detector over Harriscorner detection · 'Geometric' imagelocalization error (pixels): 1.269 -> 1.177Improved
use SOGGDD proposed detector over Harriscorner detection · 'Lab' imagelocalization error (pixels): 1.471 -> 1.449Improved
increase anisotropic factor to rho^2=1.5 (K=8) (ablation)SOGGDD detector · noise repeatability, 30 imagesaverage repeatabilityHurt
use isotropic filters (rho^2=1) with K=4, 8, 16 directions (ablation)SOGGDD detector · affine/JPEG/noise repeatability, 30 imagesaverage repeatabilityImproved
use K=8 directions with sigma^2=1.5 and rho^2=1 over other K values (ablation)SOGGDD detector · repeatability, 30 imagesaverage repeatabilityImproved
use multi-direction second-order filtering rather than horizontal/vertical derivatives onlycorner detection · Star-polygons 30 sidesmissed and false cornersImproved
use SOGGDD detector on region repeatabilitycorner detection · Oxford sequences Trees, Bikes, Ubc, Leuven, Wallrepeatability scoreImproved
use SOGGDD detector on Graffiticorner detection · Oxford Graffiti sequencerepeatability scoreNo clear effect
use SOGGDD detector over ACJ in speedcorner detection · execution time, Bikesexecution time (s): 9.452 -> 4.521Improved
use SOGGDD detector over Harris in speedcorner detection · execution time, Bikesexecution time (s): 0.717 -> 4.521Hurt

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