摘要
In this paper,we propose a simple but effective framework for lane boundary detection,called Spin Net.Considering that cars or pedestrians often occlude lane boundaries and that the local features of lane boundaries are not distinctive,therefore,analyzing and collecting global context information is crucial for lane boundary detection.To this end,we design a novel spinning convolution layer and a brand-new lane parameterization branch in our network to detect lane boundaries from a global perspective.To extract features in narrow strip-shaped fields,we adopt stripshaped convolutions with kernels which have 1×n or n×1 shape in the spinning convolution layer.To tackle the problem of that straight strip-shaped convolutions are only able to extract features in vertical or horizontal directions,we introduce the concept of feature map rotation to allow the convolutions to be applied in multiple directions so that more information can be collected concerning a whole lane boundary.Moreover,unlike most existing lane boundary detectors,which extract lane boundaries from segmentation masks,our lane boundary parameterization branch predicts a curve expression for the lane boundary for each pixel in the output feature map.And the network utilizes this information to predict the weights of the curve,to better form the final lane boundaries.Our framework is easy to implement and end-to-end trainable.Experiments show that our proposed Spin Net outperforms state-of-the-art methods.
In this paper, we propose a simple but effective framework for lane boundary detection, called Spin Net. Considering that cars or pedestrians often occlude lane boundaries and that the local features of lane boundaries are not distinctive, therefore, analyzing and collecting global context information is crucial for lane boundary detection. To this end, we design a novel spinning convolution layer and a brand-new lane parameterization branch in our network to detect lane boundaries from a global perspective. To extract features in narrow strip-shaped fields, we adopt stripshaped convolutions with kernels which have 1 × n or n × 1 shape in the spinning convolution layer. To tackle the problem of that straight strip-shaped convolutions are only able to extract features in vertical or horizontal directions, we introduce the concept of feature map rotation to allow the convolutions to be applied in multiple directions so that more information can be collected concerning a whole lane boundary. Moreover,unlike most existing lane boundary detectors, which extract lane boundaries from segmentation masks, our lane boundary parameterization branch predicts a curve expression for the lane boundary for each pixel in the output feature map. And the network utilizes this information to predict the weights of the curve, to better form the final lane boundaries. Our framework is easy to implement and end-to-end trainable. Experiments show that our proposed Spin Net outperforms state-of-the-art methods.
基金
supported by the National Natural Science Foundation of China(Project No.61572264)
Research Grant of Beijing Higher Institution Engineering Research Center
Tsinghua–Tencent Joint Laboratory for Internet Innovation Technology.
作者简介
Ruochen Fan,is a master student at Computer Science Department,Tsinghua University under the supervision of Prof.Shi-Min Hu.He currently focuses on perception system for autonomous driving,especially for point cloud segmentation and RGB detection.Before that,He did some research on saliency detection and weakly-supervised segmentation.E-mail:R.Fan,frc16@mails.tsinghua.edu.cn;Xuanrun Wang is currently a senior undergraduate student in the Department of Computer Science and Technology at Tsinghua University.His research interest is computer vision.E-mail:X.Wang,xuanrun-16@mails.tsinghua.edu.cn;Qibin Hou is at present a third-year Ph.D.student under Prof.Ming-Ming Cheng’s supervision.Before joining in the media group at Nankai University,he was a machine learning engineer in Baidu.His research interests include low-level vision,deep learning,and multimedia applications.E-mail:andrewhoux@gmail.com;Hanchao Liu is currently a master student in the Department of Computer Science and Technology,Tsinghua University.His research interests include image/video processing and computer vision.E-mail:H.Liu,liuhc17@mails.tsinghua.edu.cn;Tai-Jiang Mu is currently a assistant researcher in the Department of Computer Science and Technology,Tsinghua University,where he received his Ph.D.and B.S.degrees in 2016 and 2011,respectively.His research area is computer graphics,mainly focusing on stereoscopic image and video processing,and stereoscopic perception.E-mail:T.-J.Mu,mmmutj@gmail.com;
