DSAC - Differentiable RANSAC for Camera Localization

Abstract

RANSAC is an important algorithm in robust optimization and a central building block for many computer vision applications. In recent years, traditionally hand-crafted pipelines have been replaced by deep learning pipelines, which can be trained in an end-to-end fashion. However, RANSAC has so far not been used as part of such deep learning pipelines, because its hypothesis selection procedure is non-differentiable. In this work, we present two different ways to overcome this limitation. The most promising approach is inspired by reinforcement learning, namely to replace the deterministic hypothesis selection by a probabilistic selection for which we can derive the expected loss w.r.t. to all learnable parameters. We call this approach DSAC, the differentiable counterpart of RANSAC. We apply DSAC to the problem of camera localization, where deep learning has so far failed to improve on traditional approaches. We demonstrate that by directly minimizing the expected loss of the output camera poses, robustly estimated by RANSAC, we achieve an increase in accuracy. In the future, any deep learning pipeline can use DSAC as a robust optimization component.

arXiv commentCVPR 2017

References in corpus (11)

• Adam: A Method for Stochastic Optimization
• Very Deep Convolutional Networks for Large-Scale Image Recognition
• Deep Residual Learning for Image Recognition
• Fully Convolutional Networks for Semantic Segmentation
• Deep Image Homography Estimation
• Predicting Depth, Surface Normals and Semantic Labels with a Common Multi-Scale Convolutional Architecture
• LIFT: Learned Invariant Feature Transform
• On the Sample Complexity of End-to-end Training vs. Semantic Abstraction Training
• WarpNet: Weakly Supervised Matching for Single-view Reconstruction
• Learning Analysis-by-Synthesis for 6D Pose Estimation in RGB-D Images
• DeepMatching: Hierarchical Deformable Dense Matching

Cited by in corpus (5)

• Cross-view Semantic Segmentation for Sensing Surroundings
• UR2KiD: Unifying Retrieval, Keypoint Detection, and Keypoint Description without Local Correspondence Supervision
• Image-based localization using LSTMs for structured feature correlation
• Purely Geometric Scene Association and Retrieval - A Case for Macro Scale 3D Geometry
• Eigendecomposition-Free Training of Deep Networks for Linear Least-Square Problems