S. Xie, and Z. Tu. (2015)cite arxiv:1504.06375Comment: v2 Add appendix A for updated results (ODS=0.790) on BSDS-500 in a new experiment setting. Fix typos and reorganize formulations. Add Table 2 to discuss the role of deep supervision. Add links to publicly available repository for code, models and data.
Abstract
We develop a new edge detection algorithm that tackles two important issues
in this long-standing vision problem: (1) holistic image training and
prediction; and (2) multi-scale and multi-level feature learning. Our proposed
method, holistically-nested edge detection (HED), performs image-to-image
prediction by means of a deep learning model that leverages fully convolutional
neural networks and deeply-supervised nets. HED automatically learns rich
hierarchical representations (guided by deep supervision on side responses)
that are important in order to approach the human ability resolve the
challenging ambiguity in edge and object boundary detection. We significantly
advance the state-of-the-art on the BSD500 dataset (ODS F-score of .782) and
the NYU Depth dataset (ODS F-score of .746), and do so with an improved speed
(0.4 second per image) that is orders of magnitude faster than some recent
CNN-based edge detection algorithms.
cite arxiv:1504.06375Comment: v2 Add appendix A for updated results (ODS=0.790) on BSDS-500 in a new experiment setting. Fix typos and reorganize formulations. Add Table 2 to discuss the role of deep supervision. Add links to publicly available repository for code, models and data
%0 Generic
%1 xie2015holisticallynested
%A Xie, Saining
%A Tu, Zhuowen
%D 2015
%K cs.CV
%T Holistically-Nested Edge Detection
%U http://arxiv.org/abs/1504.06375
%X We develop a new edge detection algorithm that tackles two important issues
in this long-standing vision problem: (1) holistic image training and
prediction; and (2) multi-scale and multi-level feature learning. Our proposed
method, holistically-nested edge detection (HED), performs image-to-image
prediction by means of a deep learning model that leverages fully convolutional
neural networks and deeply-supervised nets. HED automatically learns rich
hierarchical representations (guided by deep supervision on side responses)
that are important in order to approach the human ability resolve the
challenging ambiguity in edge and object boundary detection. We significantly
advance the state-of-the-art on the BSD500 dataset (ODS F-score of .782) and
the NYU Depth dataset (ODS F-score of .746), and do so with an improved speed
(0.4 second per image) that is orders of magnitude faster than some recent
CNN-based edge detection algorithms.
@misc{xie2015holisticallynested,
abstract = {We develop a new edge detection algorithm that tackles two important issues
in this long-standing vision problem: (1) holistic image training and
prediction; and (2) multi-scale and multi-level feature learning. Our proposed
method, holistically-nested edge detection (HED), performs image-to-image
prediction by means of a deep learning model that leverages fully convolutional
neural networks and deeply-supervised nets. HED automatically learns rich
hierarchical representations (guided by deep supervision on side responses)
that are important in order to approach the human ability resolve the
challenging ambiguity in edge and object boundary detection. We significantly
advance the state-of-the-art on the BSD500 dataset (ODS F-score of .782) and
the NYU Depth dataset (ODS F-score of .746), and do so with an improved speed
(0.4 second per image) that is orders of magnitude faster than some recent
CNN-based edge detection algorithms.},
added-at = {2020-12-17T09:35:03.000+0100},
author = {Xie, Saining and Tu, Zhuowen},
biburl = {https://www.bibsonomy.org/bibtex/29f6df63dbdfcfa9606c3184d050750a7/aerover},
description = {Holistically-Nested Edge Detection},
interhash = {ec7889e1bfec52d1c5ef8ae755ba8f2b},
intrahash = {9f6df63dbdfcfa9606c3184d050750a7},
keywords = {cs.CV},
note = {cite arxiv:1504.06375Comment: v2 Add appendix A for updated results (ODS=0.790) on BSDS-500 in a new experiment setting. Fix typos and reorganize formulations. Add Table 2 to discuss the role of deep supervision. Add links to publicly available repository for code, models and data},
timestamp = {2021-02-22T08:38:00.000+0100},
title = {Holistically-Nested Edge Detection},
url = {http://arxiv.org/abs/1504.06375},
year = 2015
}