Skip to main content
Home
ENHU

Main navigation

  • Discover
    • News
    • Events
  • Research & development
    • Areas of application
    • Research topics
  • Resources
    • Publications
    • Lead researchers
  • Partners
    • Consortium members
    • International partners
    • Industry contacts
    • University contacts
  1. Home
  2. Publications
IEEE Robotics and Automation Letters, Volume 6, Issue 4 / Oct 2021

ChangeGAN: A Deep Network for Change Detection in Coarsely Registered Point Clouds

Széchenyi Plusz RRF

In this letter we introduce a novel change detection approach called ChangeGAN for coarsely registered point clouds in complex street-level urban environment. Our generative adversarial network-like (GAN) architecture compounds Siamese-style feature extraction, U-net-like use of multiscale features, and Spatial Trans-formation Network (STN) blocks for optimal transformation estimation. The input point clouds are represented by range images, which enables the use of 2D convolutional neural networks. The result is a pair of binary masks showing the change regions on each input range image, which can be backprojected to the input point clouds without loss of information. We have evaluated the proposed method on various challenging scenarios and we have shown its superiority against state-of-the-art change detection methods.

Url
10.1109/LRA.2021.3105721
Authors
Nagy, B.
Kovács. L.
Benedek, Cs.
Institutes

Kapcsolat

Prof. Dr. Péter Gáspár

Hungary, H-1111 Budapest,
Kende u. 13-17.
+36 1 279 6000
autonom@nemzetilabor.hu

© 2020-2021 National Laboratory for Autonomous Systems, Budapest