Deep Crisp Boundaries: From Boundaries to Higher-level Tasks | |
Wang, Yupei; Zhao, Xin; Li, Yin; Huang, Kaiqi | |
刊名 | IEEE Transactions on Image Processing |
2019 | |
卷号 | 28期号:3页码:1285-1298 |
关键词 | Boundary Detection, Deep Learning |
英文摘要 | Edge detection has made significant progress with the help of deep convolutional networks (ConvNet). These ConvNet-based edge detectors have approached human level performance on standard benchmarks. We provide a systematical study of these detectors’ outputs. We show that the detection results did not accurately localize edge pixels, which can be adversarial for tasks that require crisp edge inputs. As a remedy, we propose a novel refinement architecture to address the challenging problem of learning a crisp edge detector using ConvNet. Our method leverages a top-down backward refinement pathway, and progressively increases the resolution of feature maps to generate crisp edges. Our results achieve superior performance, surpassing human accuracy when using standard criteria on BSDS500, and largely outperforming the state-of-the-art methods when using more strict criteria. More importantly, we demonstrate the benefit of crisp edge maps for several important applications in computer vision, including optical flow estimation, object proposal generation, and semantic segmentation. |
内容类型 | 期刊论文 |
源URL | [http://ir.ia.ac.cn/handle/173211/23349] |
专题 | 中国科学院自动化研究所 |
作者单位 | 1.Institute of Automation, Chinese Academy of Sciences 2.University of Chinese Academy of Sciences |
推荐引用方式 GB/T 7714 | Wang, Yupei,Zhao, Xin,Li, Yin,et al. Deep Crisp Boundaries: From Boundaries to Higher-level Tasks[J]. IEEE Transactions on Image Processing,2019,28(3):1285-1298. |
APA | Wang, Yupei,Zhao, Xin,Li, Yin,&Huang, Kaiqi.(2019).Deep Crisp Boundaries: From Boundaries to Higher-level Tasks.IEEE Transactions on Image Processing,28(3),1285-1298. |
MLA | Wang, Yupei,et al."Deep Crisp Boundaries: From Boundaries to Higher-level Tasks".IEEE Transactions on Image Processing 28.3(2019):1285-1298. |
个性服务 |
查看访问统计 |
相关权益政策 |
暂无数据 |
收藏/分享 |
除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。
修改评论