Integrating Parametric and Non-Parametric Models for Scene Labeling

Bing Shuai, Gang Wang, Zhen Zuo, Bing Wang, Lifan Zhao

CVPR 2015

doi:10.1109/CVPR.2015.7299053 /cvpr/2015/shuai2015cvpr-integrating/

Abstract

We adopt Convolutional Neural Networks (CNN) as our parametric model to learn discriminative features and classifiers for local patch classification. As visually similar pixels are indistinguishable from local context, we alleviate such ambiguity by putting a global scene constraint. We estimate the global potential in a non-parametric framework. Furthermore, a large margin based CNN metric learning method is proposed for better global potential estimation. The final pixel class prediction is performed by integrating local and global beliefs. Even without any post-processing, we achieve state-of-the-art on SiftFlow and competitive results on Stanford Background benchmark.

PDF CVPR Semantic Scholar

Cite

Text

Shuai et al. "Integrating Parametric and Non-Parametric Models for Scene Labeling." Conference on Computer Vision and Pattern Recognition, 2015. doi:10.1109/CVPR.2015.7299053

Markdown

[Shuai et al. "Integrating Parametric and Non-Parametric Models for Scene Labeling." Conference on Computer Vision and Pattern Recognition, 2015.](https://mlanthology.org/cvpr/2015/shuai2015cvpr-integrating/) doi:10.1109/CVPR.2015.7299053

BibTeX

@inproceedings{shuai2015cvpr-integrating,
  title     = {{Integrating Parametric and Non-Parametric Models for Scene Labeling}},
  author    = {Shuai, Bing and Wang, Gang and Zuo, Zhen and Wang, Bing and Zhao, Lifan},
  booktitle = {Conference on Computer Vision and Pattern Recognition},
  year      = {2015},
  doi       = {10.1109/CVPR.2015.7299053},
  url       = {https://mlanthology.org/cvpr/2015/shuai2015cvpr-integrating/}
}