AI Paper Atlas

Research map / 3D vision / Face and shape

AtlantaNet: Inferring the 3D Indoor Layout from a Single 360(∘) Image beyond the Manhattan World Assumption

Giovanni Pintore, Marco Agus, Enrico Gobbetti

ECCV 2020 · 3D vision › Face and shape · approach: Primitive-based shape abstraction

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
train two instances jointly with a summed loss instead of separately (ablation)AtlantaNet · Ablation average (Manhattan+Atlanta)3D IoU: 75.48 -> 76.04Improved
use a single instance with mixed ceiling/floor feeding instead of two jointly trained instances (ablation)AtlantaNet · Ablation average (Manhattan+Atlanta)3D IoU: 76.04 -> 79.26Improved
use a single instance with mixed feeding instead of two separate instances (also halves parameters) (ablation)AtlantaNet · Ablation average (Manhattan+Atlanta)3D IoU: 75.48 -> 79.26; 200M -> 100M paramsImproved
add the gradient (Sobel) loss term (ablation)AtlantaNet (mixed feeding, ResNet50) · Ablation average (Manhattan+Atlanta)3D IoU: 79.26 -> 80.79Improved
add the gradient loss term (ablation)AtlantaNet (mixed feeding, ResNet50) · Ablation average (Manhattan+Atlanta)delta_i: 0.854 -> 0.902Improved
use ResNet101 backbone instead of ResNet50 (ablation)AtlantaNet · Ablation average (Manhattan+Atlanta)3D IoU: 80.79 -> 83.22; 100M -> 119M paramsImproved
fine-tune with selected Atlanta scenes added to trainingAtlantaNet · AtlantaLayout Atlanta overall3D IoU: 72.50 -> 80.01Improved
replace HorizonNet with AtlantaNet on >10-corner Manhattan layoutslayout estimator · MatterportLayout Manhattan >10 corners3D IoU: 68.32 -> 73.89Improved
replace HorizonNet with AtlantaNet on 6-corner Manhattan layoutslayout estimator · MatterportLayout Manhattan 6 corners3D IoU: 82.26 -> 80.10Hurt
replace HorizonNet with AtlantaNet on cuboid layouts trained on PanoContextlayout estimator · Cuboid layout (train PanoContext)3D IoU: 75.57 -> 75.56No clear effect

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 2 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