Deep Crisp Boundaries: From Boundaries to Higher-level Tasks

TL;DR

This paper introduces a novel deep learning architecture for edge detection that produces crisp, accurately localized edges, significantly improving performance and benefiting various computer vision tasks.

Contribution

The paper proposes a top-down refinement network for learning crisp edges with ConvNets, surpassing previous methods and demonstrating advantages in multiple vision applications.

Findings

Achieved state-of-the-art edge detection performance on BSDS500.

Produced edges that outperform human accuracy on standard criteria.

Enhanced performance in optical flow, object proposals, and segmentation tasks.

Abstract

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 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.