Deep Crisp Boundaries: From Boundaries to Higher-level Tasks
This work addresses the need for crisp edge maps in computer vision applications like optical flow and segmentation, representing a novel method for a known bottleneck rather than a foundational advancement.
The paper tackled the problem of inaccurate edge localization in deep convolutional network-based edge detectors, which is adversarial for tasks requiring crisp edges, and proposed a refinement architecture that achieved superior performance, surpassing human accuracy on BSDS500 with standard criteria and largely outperforming state-of-the-art methods under stricter criteria.
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.