AI Paper Atlas

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

Don't Hit Me! Glass Detection in Real-World Scenes

Haiyang Mei, Xin Yang, Yang Wang, Yuanyuan Liu, Shengfeng He, Qiang Zhang, Xiaopeng Wei, Rynson W.H. Lau

CVPR 2020 · Detection and segmentation › Open-set and few-shot detection · approach: Industrial anomaly detection

Check 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 GDNet with large-field contextual feature integrationGlass detection · GDD test setIoU: 85.07 -> 87.63Improved
Add LCFI modules to the base network (ablation)Glass detection · GDD test setIoU: 84.89 -> 87.63Improved
Use one LCFI block scale (ablation)Base glass detection network · GDD test setIoU: 84.89 -> 86.22Improved
Use two LCFI block scales instead of one (ablation)Base glass detection network with LCFI · GDD test setIoU: 86.22 -> 86.78Improved
Replace spatially separable convolutions with local convolutions (ablation)Base glass detection network with LCFI · GDD test setIoU: 87.63 -> 86.93Hurt
Use dilated convolutions instead of local convolutions (ablation)Base glass detection network with LCFI · GDD test setIoU: 86.93 -> 87.13Improved
Use spatially separable convolutions instead of dilated convolutions (ablation)Base glass detection network with LCFI · GDD test setIoU: 87.13 -> 87.63Improved
Use two parallel spatially separable convolution paths per LCFI block (ablation)Base glass detection network with LCFI · GDD test setIoU: 87.31 -> 87.63Improved
Use GDNet for mirror segmentationMirror segmentation · MSD test setIoU: 78.95 -> 80.31Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 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