边距(机器学习)
计算机科学
人工智能
GSM演进的增强数据速率
边缘检测
卷积(计算机科学)
分割
图层(电子)
深度学习
卷积神经网络
集合(抽象数据类型)
对象(语法)
图像分割
计算机视觉
Canny边缘检测器
视觉对象识别的认知神经科学
像素
模式识别(心理学)
目标检测
机器学习
图像(数学)
图像处理
人工神经网络
有机化学
化学
程序设计语言
作者
Zhiding Yu,Chen Feng,Mingyu Li,Srikumar Ramalingam
出处
期刊:Computer Vision and Pattern Recognition
日期:2017-07-01
被引量:230
标识
DOI:10.1109/cvpr.2017.191
摘要
Boundary and edge cues are highly beneficial in improving a wide variety of vision tasks such as semantic segmentation, object recognition, stereo, and object proposal generation. Recently, the problem of edge detection has been revisited and significant progress has been made with deep learning. While classical edge detection is a challenging binary problem in itself, the category-aware semantic edge detection by nature is an even more challenging multi-label problem. We model the problem such that each edge pixel can be associated with more than one class as they appear in contours or junctions belonging to two or more semantic classes. To this end, we propose a novel end-to-end deep semantic edge learning architecture based on ResNet and a new skip-layer architecture where category-wise edge activations at the top convolution layer share and are fused with the same set of bottom layer features. We then propose a multi-label loss function to supervise the fused activations. We show that our proposed architecture benefits this problem with better performance, and we outperform the current state-of-the-art semantic edge detection methods by a large margin on standard data sets such as SBD and Cityscapes.
科研通智能强力驱动
Strongly Powered by AbleSci AI