AI Paper Atlas

Research map / Detection and segmentation / CNN features

Large-Field Contextual Feature Learning for Glass Detection

Haiyang Mei, Xin She Yang, Letian Yu, Qiang Zhang, Xiaopeng Wei, Rynson W. H. Lau

TPAMI 2022 · Detection and segmentation › CNN features · approach: Context for hard detection conditions · 28 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 LCFI module to base network (ablation)GDNet base · GDDIoU: 84.89 -> 87.63Improved
add LCFI module to base network (ablation)GDNet base · GDDBER: 7.40 -> 5.62Improved
increase LCFI from one scale to two scales (ablation)LCFI · GDDIoU: 86.22 -> 86.78Improved
increase LCFI from two scales to four scales (ablation)LCFI · GDDIoU: 86.78 -> 87.63Improved
use dilated convolutions instead of local convolutions in LCFI (ablation)LCFI · GDDIoU: 86.93 -> 87.13Improved
use spatially separable convolutions instead of dilated convolutions (ablation)LCFI · GDDIoU: 87.13 -> 87.31Improved
add a second parallel spatially separable path in each LCFI block (ablation)LCFI · GDDIoU: 87.31 -> 87.63Improved
add boundary feature enhancement (BFE) module (ablation)GDNet · GDDIoU: 87.63 -> 87.83Improved
add boundary feature enhancement (BFE) module (ablation)GDNet · GDDBER: 5.62 -> 5.52Improved
use ResNeXt-101 instead of ResNet-101 backbone (ablation)GDNet-B · GDDIoU: 86.84 -> 87.83Improved
use SENet-154 instead of ResNeXt-101 backbone (ablation)GDNet-B · GDDIoU: 87.83 -> 88.09Improved
use GDNet-B over best prior MirrorNet on mirror segmentationMirrorNet · MSDIoU: 78.95 -> 81.70Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 4 of 4 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