TY - GEN
T1 - An Efficient Method for Indoor Layout Estimation with FPN
AU - Wang, Aopeng
AU - Wen, Shiting
AU - Gao, Yunjun
AU - Li, Qing
AU - Deng, Ke
AU - Pang, Chaoyi
N1 - Funding Information:
Acknowledgments. The authors would like to thank the data providers of [23] for the testing data sets. This work was partially supported by the Natural Science Foundation of China (No. 61802344), the Ningbo Science and Technology Special Project(No. 2021Z019), the Hebei “One Hundred Plan” Project (No. E2012100006) and National Talent Program (No. G20200218015).
Publisher Copyright:
© 2021, Springer Nature Switzerland AG.
PY - 2021
Y1 - 2021
N2 - 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.
AB - 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.
KW - Feature pyramid network
KW - Layout estimation
KW - Scene understanding
UR - http://www.scopus.com/inward/record.url?scp=85121930633&partnerID=8YFLogxK
U2 - 10.1007/978-3-030-91560-5_7
DO - 10.1007/978-3-030-91560-5_7
M3 - Conference article published in proceeding or book
AN - SCOPUS:85121930633
SN - 9783030915599
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 94
EP - 106
BT - Web Information Systems Engineering - WISE 2021 - 22nd International Conference on Web Information Systems Engineering, WISE 2021, Proceedings
A2 - Zhang, Wenjie
A2 - Zou, Lei
A2 - Maamar, Zakaria
A2 - Chen, Lu
PB - Springer Science and Business Media Deutschland GmbH
T2 - 22nd International Conference on Web Information Systems Engineering, WISE 2021
Y2 - 26 October 2021 through 29 October 2021
ER -