AI Paper Atlas

Research map / 3D vision / 3D generation

Manhattan Room Layout Reconstruction from a Single $$360^{\circ }$$ Image: A Comparative Study of State-of-the-Art Methods

Chuhang Zou, Jheng-Wei Su, Chi‐Han Peng, Alex Colburn, Qi Shan, Peter Wonka, Hung‐Kuo Chu, Derek Hoiem

IJCV 2021 · 3D vision › 3D generation · approach: Generative indoor scene synthesis · 72 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 ImageNet pre-training to the encoder (ablation)LayoutNet v2 · PanoContext3D IoU (%): 78.71 -> 85.02Improved
use gradient ascent instead of sampling in layout fitting (ablation)LayoutNet v2 · PanoContext3D IoU (%): 83.60 -> 85.02Improved
add random stretching augmentation (ablation)LayoutNet v2 · PanoContext3D IoU (%): 83.97 -> 85.02Improved
freeze batch norm layers during end-to-end training (ablation)LayoutNet v2 · PanoContext3D IoU (%): 83.98 -> 85.02Improved
use equirectangular-view post-processing instead of DuLa-Net ceiling-view post-processing (ablation)LayoutNet v2 · PanoContext3D IoU (%): 81.45 -> 85.02Improved
use equirectangular-view post-processing instead of HorizonNet post-processing (ablation)LayoutNet v2 · PanoContext3D IoU (%): 82.70 -> 85.02Improved
add semantic loss term to layout optimization (ablation)LayoutNet v2 · PanoContext3D IoU (%): 85.02 -> 84.35No clear effect
add random stretching augmentation (ablation)DuLa-Net v2 · Stanford 2D-3D3D IoU (%): 85.03 -> 86.60Improved
use ResNet-50 instead of ResNet-18 encoder (ablation)DuLa-Net v2 · Stanford 2D-3D3D IoU (%): 84.93 -> 86.60Improved
use ResNet-50 instead of ResNet-34 encoderLayoutNet v2 · PanoContext3D IoU (%): 85.02 -> 82.44Hurt
use ResNet-50 instead of ResNet-18 encoderHorizonNet · PanoContext3D IoU (%): 80.27 -> 82.63Improved
use HorizonNet instead of LayoutNet v2 on general layoutsHorizonNet · MatterportLayout3D IoU (%): 75.82 -> 79.11Improved

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