AI Paper Atlas

Research map / Detection and segmentation / CNN features

HoughNet: Integrating near and long-range evidence for bottom-up object detection

Nermin Samet, Samet Hicsonmez, Emre Akbas

ECCV 2020 · Detection and segmentation › CNN features · approach: Region-based and query-based detectors · 62 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 voting module to OAP baseline (ablation)OAP R-101-DCN · COCO valAP: 36.2 -> 37.3Improved
replace voting module with dilated convolution (ablation)OAP R-101-DCN · COCO valAP: 37.3 -> 36.6Hurt
add dilated convolution to baseline (ablation)OAP R-101-DCN · COCO valAP: 36.2 -> 36.6Improved
add voting module to OAP baseline for small objects (ablation)OAP R-101-DCN · COCO valAP_S: 16.3 -> 16.8Improved
keep only the center ring of the vote field (remove outer rings) (ablation)HoughNet · COCO val (ablation setting)AP: 23.8 -> 24.6Improved
add the center ring when using only outer rings (ablation)HoughNet · COCO val (ablation setting)AP: 24.4 -> 24.6Improved
add inner rings to context-only voting (ablation)HoughNet · COCO val (ablation setting)AP: 23.6 -> 24.6Improved
vary number of angle bins 60 to 360 degrees (ablation)HoughNet · COCO val (ablation setting)AP: 24.6 -> 24.5No clear effect
reduce rings from 5 to 3 (ablation)HoughNet · COCO val (ablation setting)AP: 24.6 -> 24.8Improved
reduce rings from 5 to 3 (ablation)HoughNet · COCO val (ablation setting)FPS: 3.5 -> 15.6Improved
add ExtremeNet weight initializationHoughNet HG-104 (SS) · COCO test-devAP: 40.8 -> 43.1Improved
add voting module to CycleGANCycleGAN · CityscapesPer-pixel acc.: 0.43 -> 0.52Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 2 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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