Self-Supervised Optical Flow Estimation by Projective Bootstrap

Dense optical flow estimation is complex and time consuming, with state-of-the-art methods relying either on large synthetic data sets or on pipelines requiring up to a few minutes per frame pair. In this paper, the authors address the problem of optical flow estimation in the automotive scenario in a self-supervised manner. The author's argue that optical flow can be cast as a geometrical warping between two successive video frames and devise a deep architecture to estimate such transformation in two stages. First, a dense pixel-level flow is computed with a projective bootstrap on rigid surfaces. The author's show how such global transformation can be approximated with a homography and extend spatial transformer layers so that they can be employed to compute the flow field implied by such transformation. Subsequently, the author's refine the prediction by feeding a second, deeper network that accounts for moving objects. A final reconstruction loss compares the warping of frame with the subsequent frame and guides both estimates.The model has the speed advantages of end-to-end deep architectures while achieving competitive performances, both outperforming recent unsupervised methods and showing good generalization capabilities on new automotive data sets.

Language

  • English

Media Info

Subject/Index Terms

Filing Info

  • Accession Number: 01718383
  • Record Type: Publication
  • Files: TLIB, TRIS
  • Created Date: Aug 29 2019 3:13PM