University of Oulu

L. Huynh, P. Nguyen, J. Matas, E. Rahtu and J. Heikkilä, "Lightweight Monocular Depth with a Novel Neural Architecture Search Method," 2022 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), 2022, pp. 326-336, doi: 10.1109/WACV51458.2022.00040

Lightweight monocular depth with a novel neural architecture search method

Saved in:
Author: Huynh, Lam1; Nguyen, Phong1; Matas, Jiri2;
Organizations: 1University of Oulu
2Czech Technical University in Prague
3Tampere University
Format: article
Version: accepted version
Access: open
Online Access: PDF Full Text (PDF, 8.1 MB)
Persistent link:
Language: English
Published: Institute of Electrical and Electronics Engineers, 2022
Publish Date: 2022-06-07


This paper presents a novel neural architecture search method, called LiDNAS, for generating lightweight monocular depth estimation models. Unlike previous neural architecture search (NAS) approaches, where finding optimized networks is computationally demanding, the introduced novel Assisted Tabu Search leads to efficient architecture exploration. Moreover, we construct the search space on a pre-defined backbone network to balance layer diversity and search space size. The LiDNAS method outperforms the state-of-the-art NAS approach, proposed for disparity and depth estimation, in terms of search efficiency and output model performance. The LiDNAS optimized models achieve result superior to compact depth estimation state-of-the-art on NYU-Depth-v2, KITTI, and ScanNet, while being 7%-500% more compact in size, i.e the number of model parameters.

see all

Series: IEEE Winter Conference on Applications of Computer Vision
ISSN: 1550-5790
ISSN-E: 2472-6737
ISSN-L: 1550-5790
ISBN: 978-1-6654-0915-5
ISBN Print: 978-1-6654-0916-2
Pages: 326 - 336
DOI: 10.1109/WACV51458.2022.00040
Host publication: 2022 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), 3-8 Jan 2022, Waikoloa, HI, USA
Conference: IEEE/CVF Winter Conference on Applications of Computer Vision (WACV)
Type of Publication: A4 Article in conference proceedings
Field of Science: 113 Computer and information sciences
Copyright information: © 2021 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.