Abstract | ||
---|---|---|
Our aim is to provide a pixel-wise instance-level labeling of a monocular image in the context of autonomous driving. We build on recent work [32] that trained a convolutional neural net to predict instance labeling in local image patches, extracted exhaustively in a stride from an image. A simple Markov random field model using several heuristics was then proposed in [32] to derive a globally consistent instance labeling of the image. In this paper, we formulate the global labeling problem with a novel densely connected Markov random field and show how to encode various intuitive potentials in a way that is amenable to efficient mean field inference [15]. Our potentials encode the compatibility between the global labeling and the patch-level predictions, contrast-sensitive smoothness as well as the fact that separate regions form different instances. Our experiments on the challenging KITTI benchmark [8] demonstrate that our method achieves a significant performance boost over the baseline [32]. |
Year | DOI | Venue |
---|---|---|
2016 | 10.1109/CVPR.2016.79 | 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) |
Keywords | Field | DocType |
instance-level segmentation,pixel-wise instance-level labeling,monocular image,autonomous driving,convolutional neural net,local image patches,globally consistent instance labeling,densely connected Markov random field,patch-level predictions,contrast-sensitive smoothness | ENCODE,Pattern recognition,Markov random field,Computer science,Inference,Segmentation,Heuristics,Artificial intelligence,Smoothness,Artificial neural network,Machine learning,Zhàng | Conference |
Volume | Issue | ISSN |
2016 | 1 | 1063-6919 |
ISBN | Citations | PageRank |
978-1-4673-8852-8 | 18 | 0.67 |
References | Authors | |
13 | 3 |
Name | Order | Citations | PageRank |
---|---|---|---|
Ziyu Zhang | 1 | 112 | 10.19 |
Sanja Fidler | 2 | 2087 | 116.71 |
Raquel Urtasun | 3 | 6810 | 304.97 |