Corpus ID: 232404462

Motion Basis Learning for Unsupervised Deep Homography Estimation with Subspace Projection

  title={Motion Basis Learning for Unsupervised Deep Homography Estimation with Subspace Projection},
  author={Nianjin Ye and Chuan Wang and Haoqiang Fan and Shuaicheng Liu},
In this paper, we introduce a new framework for unsupervised deep homography estimation. Our contributions are 3 folds. First, unlike previous methods that regress 4 offsets for a homography, we propose a homography flow representation, which can be estimated by a weighted sum of 8 pre-defined homography flow bases. Second, considering a homography contains 8 Degree-of-Freedoms (DOFs) that is much less than the rank of the network features, we propose a Low Rank Representation (LRR) block that… Expand

Figures and Tables from this paper


Deep Image Homography Estimation
Two convolutional neural network architectures are presented for HomographyNet: a regression network which directly estimates the real-valued homography parameters, and a classification network which produces a distribution over quantized homographies. Expand
Deep Homography Estimation for Dynamic Scenes
This paper investigates and discusses how to design and train a deep neural network that handles dynamic scenes, and develops a multi-scale network such that it jointly estimates the dynamics masks and homographies. Expand
Unsupervised Deep Homography: A Fast and Robust Homography Estimation Model
An unsupervised learning algorithm that trains a deep convolutional neural network to estimate planar homographies and has superior adaptability and performance compared to the corresponding supervised deep learning method. Expand
Learning Two-View Correspondences and Geometry Using Order-Aware Network
This paper proposes Order-Aware Network, which infers the probabilities of correspondences being inliers and regresses the relative pose encoded by the essential matrix, and is built hierarchically and comprises three novel operations. Expand
SuperPoint: Self-Supervised Interest Point Detection and Description
This paper presents a self-supervised framework for training interest point detectors and descriptors suitable for a large number of multiple-view geometry problems in computer vision and introduces Homographic Adaptation, a multi-scale, multi-homography approach for boosting interest point detection repeatability and performing cross-domain adaptation. Expand
LSM: Learning Subspace Minimization for Low-Level Vision
This work replaces the heuristic regularization term with a data-driven learnable subspace constraint, and preserves the data term to exploit domain knowledge derived from the first principles of a task to solve the energy minimization problem in low-level vision tasks. Expand
SuperGlue: Learning Feature Matching With Graph Neural Networks
SuperGlue is introduced, a neural network that matches two sets of local features by jointly finding correspondences and rejecting non-matchable points and introduces a flexible context aggregation mechanism based on attention, enabling SuperGlue to reason about the underlying 3D scene and feature assignments jointly. Expand
MAGSAC: Marginalizing Sample Consensus
  • D. Baráth, Jiri Matas
  • Computer Science
  • 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
  • 2019
Applying sigma-consensus, MAGSAC is proposed with no need for a user-defined sigma and improving the accuracy of robust estimation significantly, and is superior to the state-of-the-art in terms of geometric accuracy on publicly available real-world datasets for epipolar geometry and homography estimation. Expand
Robust Recovery of Subspace Structures by Low-Rank Representation
It is shown that the convex program associated with LRR solves the subspace clustering problem in the following sense: When the data is clean, LRR exactly recovers the true subspace structures; when the data are contaminated by outliers, it is proved that under certain conditions LRR can exactly recover the row space of the original data. Expand
Efficient sparse-to-dense optical flow estimation using a learned basis and layers
A novel sparse layered flow method in which each layer is represented by PCA-Flow, which is more accurate and significantly faster than popular methods such as LDOF and Classic+NL and achieves state-of-the-art performance in occluded regions on MPI-Sintel. Expand