期刊文献+

基于自注意和对抗学习的道路场景水体检测方法

Water Puddle Detection Method for Road Scene Based on Self-Attention and Adversarial Learning
下载PDF
导出
摘要 无人驾驶车辆近年来一直是研究的热点.无人车运行环境复杂、不确定因素多,尤其当其意外驶入水坑、泥潭等地形时可能直接导致抛锚,造成不可估量的损失,因此水体检测对无人车的运行有着重要意义.借助深度网络的强大学习能力,本文首先将反射注意力单元和自注意力机制相结合,并在(U shape Network,U-Net)模型基础上添加残差卷积块和上采样卷积模块,得到了新的道路场景水体区域检测模型(U shape Network with Attention for Road,URA-Net),该模型能够更好地捕捉特征依赖关系,提高水体语义特征的表示能力.进一步,本文提出了一种基于双生成器对抗学习的训练模型(Redundant With Dual Generative Adversarial Network,RWD-GAN),它对URA-Net稍做修改,拓展成两个生成器,通过在对抗网络框架下让生成器与鉴别器、生成器与生成器之间实现对抗学习,促进不同网络模型之间的信息传递.在公开数据集上的大量实验表明URA-Net达到了87.18%的F1指标,而RWD-GAN模型能够进一步提高水体检测的精度,使提升到了88.54%,URA-Net和RWD-GAN均超出现有深度网络水体检测方法的性能表现. There has been much interest in the research of self-driving cars.Yet the detection of potentially danger⁃ous obstacles on road makes this investigation more challenging.Water puddles,a typical obstacle of this kind,could trap a self-driving car or even cause serious accidents.Therefore,detecting water puddles is of great importance.To this end,this paper propose a novel water puddle detection model,URA-net(U shape Network with Attention for Road).Building its back⁃bone on U-net(U shape Network)with residual and upsampling blocks added,URA-net combines both the reflection atten⁃tion units and self-attention units,which can better characterize the dependence among image features so as to improve the representative capability to locate water puddles.Furthermore,a two-generator conditional adversarial network RWD-GAN(Redundant With Dual Generative Adversarial Network)is proposed,where two URA-Nets with a minor revision become the two generators to facilitate the information interaction in the adversarial learning process between the generators and the discriminator,as well as between the two generators themselves.Experiments on the public water puddle dataset demon⁃stration that URA-net achieves 87.18%measure,while RWD-GAN can further improve the accuracy of URA-net,pushing F1-score to 88.54%.Both URA-net and RWD-GAN outperforms the state-of-the-arts.
作者 王臣毅 王欢 孟策 WANG Chen-yi;WANG Huan;MENG Ce(School of Computer Science and Engineering,Nanjing University of Science&Technology,Nanjing,Jiangsu 210094,China)
出处 《电子学报》 EI CAS CSCD 北大核心 2023年第8期2213-2225,共13页 Acta Electronica Sinica
基金 国家自然科学基金(No.61703209)。
关键词 水体检测 自注意机制 对抗学习 深度学习 water puddle detection self-attention Adversarial learning deep learning
  • 相关文献

参考文献4

二级参考文献23

  • 1A Rankin, L Matthies, A Huertas. Daytime water detection based on sky reflections[ A]. Proceedings of the IEEF. Interna- tional Conference on Robotics and Automation (ICRA) [ C ]. Shanghai, China: Shanghai Intemonal Conference Center, 2011.5329 - 5336.
  • 2J M Geusebroek, R van den Boomgaard, A W M Smeulders, H Geerts. Color invariance [ J ]. IEEE Transactions on Pattern Analysis and Machine InteUigence(PAMI) ,2001,23(12) : 1338 - 1350.
  • 3Larry Matthies ,Paolo BeUutta , Mike Mchertry. Detecting wa- ter hazards for autonomous off road navigation[ A]. Proceedings of SPIE Conference 5083:Unmanned Ground Vehicle Technol- ogy V[ C]. Orlando, FL: SPIE, 2003.263 - 352.
  • 4Sarwal A, Nett J, Simond. Detection of Small Water Bodies [R]. USA:Perce Teak Robotics,2004.68 - 92.
  • 5M Iqbal, O Morel, F Meriaudeau. A survey on outdoor water hazard detection[ A] .Proceedings of the 5th International Con- ference on Information & Communication Technology and Sys- tems(ICTS) [ C]. Indonesia, 2009.1 - 7.
  • 6Kay M. Investigating the effect of vasodilator durgs on the tur- bulent sound caused by femoral artery stenosis using short- time fourier and wavelet transform methods[ J]. IEEE. Trans- actions on BME, 1994,41 (10) : 921 - 930.
  • 7A Rankin,L Matthies, A Huertas. Daytime water detection by fusing multiple cues for autonomous off-road navigation[ A ]. Proceedings of the 24th Army Science Conference[ C]. Odan- do, FL, 2004 .177 - 184.
  • 8A Rankin, L Matthies. Daytime water detection based on color variation[ A]. Presedings of the IEEE/RSJ International Con- ference on Intelligent Robots and Systems(IROS ) [ C ]. Taipei, Taiwan,2010.215 - 221.
  • 9Hoiem D, Efros A, Hebert M. Automatic photo pop-up[A]. ACM SIGGRAPH[ C]. Los Angeles, USA: ACM Press, 2005. 577 - 584.
  • 10高宇,卓力,王素玉,沈兰荪.一种基于H.264/AVC压缩域的GOP级视频场景转换检测算法[J].电子学报,2010,38(2):382-386. 被引量:4

共引文献42

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部