AI Paper Atlas

Research map / 3D vision / Face and shape

FloorNet: A Unified Framework for Floorplan Reconstruction from 3D Scans

Chen Liu, Jiaye Wu, Yasutaka Furukawa

ECCV 2018 · 3D vision › Face and shape · approach: Learned mesh and geometric representations · 145 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 PointNet and image branches to the floorplan branch (full FloorNet) (ablation)floorplan branch · Floorplan benchmarkrelationship metric (%): 50.0 -> 56.8Improved
use FloorNet hybrid design instead of OctNet (ablation)FloorNet · Floorplan benchmarkrelationship metric (%): 13.5 -> 56.8Improved
add floorplan-to-PointNet unpooling (ablation)FloorNet · Floorplan benchmarkrelationship metric (%): 52.3 -> 56.8Improved
add PointNet-to-floorplan pooling (ablation)FloorNet · Floorplan benchmarkrelationship metric (%): 52.8 -> 56.8Improved
add image-to-floorplan pooling (ablation)FloorNet · Floorplan benchmarkrelationship metric (%): 56.8 -> 56.8No clear effect
add image-to-floorplan pooling (ablation)FloorNet · Floorplan benchmarkicon precision (%): 50.5 -> 56.1Improved
add image-to-floorplan pooling (ablation)FloorNet · Floorplan benchmarkroom precision/recall (%): 83.1 -> 85.2 recall, 84.2 -> 83.6 precisionMixed
add PointNet-to-floorplan pooling (ablation)FloorNet · Floorplan benchmarkicon recall (%): 42.2 -> 57.8Improved
use PointNet branch alone instead of floorplan branch (ablation)PointNet only · Floorplan benchmarkroom precision (%): 76.3 -> 27.0Hurt
use image branch alone instead of floorplan branch (ablation)Image branch only · Floorplan benchmarkrelationship metric (%): 50.0 -> 0.0Hurt
use FloorNet instead of Tango Navigator App for wall geometry (ablation)FloorNet · Wall line distanceaverage line distance (pixels): 2.72 -> 1.66Improved
train with all loss functions in one run instead of one by one (ablation)FloorNet training · Floorplan benchmarktest performanceHurt

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