Unmixing Convolutional Features for Crisp Edge Detection

Linxi Huan, Nan Xue, Xianwei Zheng, Wei He, Jianya Gong, Gui-Song Xia

Abstract: This article presents a context-aware tracing strategy (CATS) for crisp edge detection with deep edge detectors, based on an observation that the localization ambiguity of deep edge detectors is mainly caused by the mixing phenomenon of convolutional neural networks: <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Feature mixing</i> in edge classification and <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">side mixing</i> during fusing side predictions. The CATS consists of two modules: A novel <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">tracing loss</i> that performs feature unmixing by tracing boundaries for better side edge learning, and a <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">context-aware fusion block</i> that tackles the side mixing by aggregating the complementary merits of learned side edges. Experiments demonstrate that the proposed CATS can be integrated into modern deep edge detectors to improve localization accuracy. With the vanilla VGG16 backbone, in terms of BSDS500 dataset, our CATS improves the F-measure (ODS) of the RCF and BDCN deep edge detectors by 12 and 6 percent, respectively when evaluating without using the morphological non-maximal suppression scheme for edge detection.