An Efficient Method for Indoor Layout Estimation with FPN

Aopeng Wang, Shiting Wen, Yunjun Gao, Qing Li, Ke Deng, Chaoyi Pang

Research output: Chapter in book / Conference proceedingConference article published in proceeding or bookAcademic researchpeer-review

Abstract

As a fundamental part of indoor scene understanding, the research of indoor room layout estimation has attracted much attention recently. The task is to predict the structure of a room from a single image. In this article, we illustrate that this task can be well solved even without sophisticated post-processing program, by adopting Feature Pyramid Networks (FPN) to solve this problem with adaptive changes. Besides, an optimization step is devised to keep the order of key points unchanged, which is an essential part for improving the model’s performance but has been ignored from the beginning. Our method has demonstrated great performance on the benchmark LSUN dataset on both processing efficiency and accuracy. Compared with the state-of-the-art end-to-end method, our method is two times faster at processing speed (32 ms) than its speed (86 ms), with 0.71 % lower key point error and 0.2 % higher pixel error respectively. Besides, the advanced two-step method is only 0.02 % better than our result on key point error. Both the high efficiency and accuracy make our method a good choice for some real-time room layout estimation tasks.

Original languageEnglish
Title of host publicationWeb Information Systems Engineering - WISE 2021 - 22nd International Conference on Web Information Systems Engineering, WISE 2021, Proceedings
EditorsWenjie Zhang, Lei Zou, Zakaria Maamar, Lu Chen
PublisherSpringer Science and Business Media Deutschland GmbH
Pages94-106
Number of pages13
ISBN (Print)9783030915599
DOIs
Publication statusPublished - 2021
Event22nd International Conference on Web Information Systems Engineering, WISE 2021 - Melbourne, Australia
Duration: 26 Oct 202129 Oct 2021

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume13081 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference22nd International Conference on Web Information Systems Engineering, WISE 2021
Country/TerritoryAustralia
CityMelbourne
Period26/10/2129/10/21

Keywords

  • Feature pyramid network
  • Layout estimation
  • Scene understanding

ASJC Scopus subject areas

  • Theoretical Computer Science
  • General Computer Science

Fingerprint

Dive into the research topics of 'An Efficient Method for Indoor Layout Estimation with FPN'. Together they form a unique fingerprint.

Cite this