摘要
为提高汽车自动驾驶系统中视觉感知模块的鲁棒性,提出了使用图像语义分割方法进行交通场景理解。采用基于深度学习的语义分割方法,设计了兼顾运行速度和准确率的轻量化卷积神经网络。在特征提取部分,用轻量化特征提取模型MobileNetV2结构,用可变形卷积代替步长为2的卷积层;在特征解码部分,缩减卷积核数目、引入多尺度的空洞可变形卷积,补充低层特征细节。用扩充的Pascal VOC 2012数据集进行预训练和评估,用交通场景数据集Cityscapes进行测试。结果表明:该网络结构的准确率达到了平均交互比(mean IoU) 69.2%,超过了用MobileNetV2的DeepLab语义分割网络,运行速度127 ms/帧,占内存1.073 GB,优于使用VGG-16、ResNet-101的结果。
A method of traffic scene understanding was proposed using image semantic segmentation method to improve the robustness of a visual perception model in an automotive autonomous driving system.A lightweight convolutional-neural-network was designed adopting semantic segmentation using deep learning with striking an optimal balance between efficiency and performance.The lightweight model,Mobile Net V2,was adopted in the feature-extraction section,and the convolution layers were replaced using stride=2with deformable convolution layers;In feature-decoder section,multi-scale Atrous deformable convolution module was designed and low-level features were also used to add more detail information.Augmented PASCAL VOC2012dataset was used to pre-train and evaluate the network and the traffic scene dataset,Cityscapes,was used to fine-tune and test.The results show that the new network achieves an accuracy of mean IoU(intersection over union)of69.2%,and has better performances than that from DeepLab semantic segmentation networks with MobileNetV2.The new network takes only127ms per frame and1.073GB memory and is more efficient than that by the networks with VGG-16and ResNet-101.
作者
白傑
郝培涵
陈思汉
BAI Jie;HAO Peihan;CHEN Sihan(School of Automotive Studies, Tongji University, Shanghai 201804, China)
出处
《汽车安全与节能学报》
CAS
CSCD
2018年第4期433-440,共8页
Journal of Automotive Safety and Energy
基金
国家重点研发计划(2016YFB0101101)
关键词
汽车自动驾驶
场景理解
视觉感知
图像语义分割
轻量化卷积神经网络
深度学习
automotive autonomous driving
scene understanding
visual perception
image semantic segmentation
lightweight convolutional neural network
deep learning
