Transfer Learning from Synthetic Data in the Camera Pose Estimation Problem

被引:3
|
作者
Charco, Jorge L. [1 ,2 ]
Sappa, Angel D. [1 ,3 ]
Vintimilla, Boris X. [1 ]
Velesaca, Henry O. [1 ]
机构
[1] ESPOL Polytech Univ, Escuela Super Politecn Litoral, ESPOL, Campus Gustavo Galindo Km 30-5 Via Perimetral, Guayaquil, Ecuador
[2] Univ Guayaquil, Delta & Kennedy Av,PB EC090514, Guayaquil, Ecuador
[3] Comp Vis Ctr, Edif O,Campus UAB, Barcelona 08193, Spain
关键词
Relative Camera Pose Estimation; Siamese Architecture; Synthetic Data; Deep Learning; Multi-view Environments; Extrinsic Camera Parameters;
D O I
10.5220/0009167604980505
中图分类号
TP31 [计算机软件];
学科分类号
081202 ; 0835 ;
摘要
This paper presents a novel Siamese network architecture, as a variant of Resnet-50, to estimate the relative camera pose on multi-view environments. In order to improve the performance of the proposed model a transfer learning strategy, based on synthetic images obtained from a virtual-world, is considered. The transfer learning consists of first training the network using pairs of images from the virtual-world scenario considering different conditions (i.e., weather, illumination, objects, buildings, etc.); then, the learned weight of the network are transferred to the real case, where images from real-world scenarios are considered. Experimental results and comparisons with the state of the art show both, improvements on the relative pose estimation accuracy using the proposed model, as well as further improvements when the transfer learning strategy (synthetic-world data transfer learning real-world data) is considered to tackle the limitation on the training due to the reduced number of pairs of real-images on most of the public data sets.
引用
收藏
页码:498 / 505
页数:8
相关论文
共 50 条
  • [1] Transfer Learning for Driver Pose Estimation from Synthetic Data
    Sagmeister, Daniel
    Schoerkhuber, Dominik
    Nezveda, Matej
    Stiedl, Fabian
    Schimkowitsch, Maria
    Gelautz, Margrit
    [J]. 2023 IEEE INTELLIGENT VEHICLES SYMPOSIUM, IV, 2023,
  • [2] Accurate and Scalable Contour-based Camera Pose Estimation Using Deep Learning With Synthetic Data
    Hokmabadi, Ilyar Asl Sabbaghian
    Ai, Mengchi
    Minaretzis, Chrysostomos
    Sideris, Michael
    El-Sheimy, Naser
    [J]. 2023 IEEE/ION POSITION, LOCATION AND NAVIGATION SYMPOSIUM, PLANS, 2023, : 1385 - 1393
  • [3] Domain adaptation of networks for camera pose estimation: Learning camera pose estimation without pose labels
    Langerman, Jack
    Qiu, Ziming
    Sörös, Gábor
    Sebok, Dávid
    Wang, Yao
    Huang, Howard
    [J]. arXiv, 2021,
  • [4] Joint camera blur and pose estimation from aliased data
    LeBlanc, Joel W.
    Thelen, Brian J.
    Hero, Alfred O.
    [J]. JOURNAL OF THE OPTICAL SOCIETY OF AMERICA A-OPTICS IMAGE SCIENCE AND VISION, 2018, 35 (04) : 639 - 651
  • [5] Transfer Learning based Precise Pose Estimation with Insufficient Data
    Choi, Wonje
    Woo, Honguk
    [J]. PROCEEDINGS OF THE 5TH INTERNATIONAL CONFERENCE ON MACHINE VISION AND APPLICATIONS, ICMVA 2022, 2022, : 50 - 55
  • [6] Improving 2D Human Pose Estimation in Rare Camera Views with Synthetic Data
    Purkrabek, Miroslav
    Matas, Jiri
    [J]. 2024 IEEE 18TH INTERNATIONAL CONFERENCE ON AUTOMATIC FACE AND GESTURE RECOGNITION, FG 2024, 2024,
  • [7] Estimation of Camera Pose with Respect to Terrestrial LiDAR Data
    Guan, Wei
    You, Suya
    Pang, Guan
    [J]. 2013 IEEE WORKSHOP ON APPLICATIONS OF COMPUTER VISION (WACV), 2013, : 391 - 398
  • [8] A fast and accurate iterative method for the camera pose estimation problem
    Zheng, Maoteng
    Wang, Shiguang
    Xiong, Xiaodong
    Zhu, Junfeng
    [J]. IMAGE AND VISION COMPUTING, 2020, 94
  • [9] Minimal solutions for the multi-camera pose estimation problem
    Lee, Gim Hee
    Li, Bo
    Pollefeys, Marc
    Fraundorfer, Friedrich
    [J]. INTERNATIONAL JOURNAL OF ROBOTICS RESEARCH, 2015, 34 (07): : 837 - 848
  • [10] Camera Calibration and Pose Estimation from Planes
    Bazargani, Hamid
    Laganiere, Robert
    [J]. IEEE INSTRUMENTATION & MEASUREMENT MAGAZINE, 2015, 18 (06) : 20 - 27