DSAC - Differentiable RANSAC for Camera Localization

被引:383
作者
Brachmann, Eric [1 ]
Krull, Alexander [1 ]
Nowozin, Sebastian [2 ]
Shotton, Jamie [2 ]
Michel, Frank [1 ]
Gumhold, Stefan [1 ]
Rother, Carsten [1 ]
机构
[1] Tech Univ Dresden, Dresden, Germany
[2] Microsoft, Coimbatore, Tamil Nadu, India
来源
30TH IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR 2017) | 2017年
基金
欧洲研究理事会;
关键词
D O I
10.1109/CVPR.2017.267
中图分类号
TP18 [人工智能理论];
学科分类号
140502 [人工智能];
摘要
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 In recent years, traditionally hand-crafted pipelines have been replaced by deep learning pipelines, which can be trained in an end-to-end fashion.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(1).
引用
收藏
页码:2492 / 2500
页数:9
相关论文
共 42 条
[1]
[Anonymous], 2008, ECCV
[2]
[Anonymous], 2013, CVPR
[3]
[Anonymous], 2004, IJCV
[4]
[Anonymous], 2015, CORR ABS150602640
[5]
[Anonymous], 2016, P IEEE C COMP VIS PA
[6]
[Anonymous], 2014, ECCV
[7]
[Anonymous], 2015, ICCV
[8]
[Anonymous], 2015, CVPR
[9]
[Anonymous], 2016, ECCV
[10]
[Anonymous], 2009, IJCV