AI Paper Atlas

Research map / 3D vision / Depth and stereo

360 Layout Estimation via Orthogonal Planes Disentanglement and Multi-View Geometric Consistency Perception

Zhijie Shen, Chunyu Lin, Junsong Zhang, Lang Nie, Kang Liao, Yao Zhao

TPAMI 2024 · 3D vision › Depth and stereo · approach: Cost-volume multi-view stereo · 4 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 cross-scale interaction (ablation)DOPNet · MatterportLayout ablation3DIoU (%): 80.71 -> 81.34Improved
add distortion-aware feature assembling (on top of cross-scale interaction) (ablation)DOPNet · MatterportLayout ablation3DIoU (%): 80.10 -> 81.34Improved
add orthogonal plane pre-segmentation (ablation)DOPNet · MatterportLayout ablation3DIoU (%): 80.60 -> 81.34Improved
add soft-flipping fusion (ablation)DOPNet · MatterportLayout ablation3DIoU (%): 80.35 -> 81.34Improved
add channel-wise graph convolution (discriminative channels) (ablation)DOPNet · MatterportLayout ablation3DIoU (%): 80.23 -> 81.34Improved
add self-attention for long-range dependencies (ablation)DOPNet · MatterportLayout ablation3DIoU (%): 80.97 -> 81.34Improved
add cross-attention residuals (ablation)DOPNet · MatterportLayout ablation3DIoU (%): 80.45 -> 81.34Improved
add multi-view geometric consistency perception during fine-tuning (ablation)DOPNet · MVL, pseudo-label fine-tuning3DIoU (%): 51.91 -> 53.99Improved
add ceiling-3D ratio loss (ablation)DOPNet · MVL, pseudo-label fine-tuning3DIoU (%): 51.91 -> 76.02Improved
add multi-view perception on top of ceiling-3D loss (ablation)DOPNet · MVL, pseudo-label fine-tuning3DIoU (%): 76.02 -> 77.72Improved
add improved normal and gradient losses on top of ratio loss (ablation)MV-DOPNet · MVL, pseudo-label fine-tuning3DIoU (%): 77.72 -> 79.42Improved
use DOPNet instead of HorizonNet as pre-trained model without fine-tuninglayout network · MVL, pre-trained on MatterportLayout3DIoU (%): 73.93 -> 75.13Improved

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