Skip to content

In Defense of Pre-trained ImageNet Architectures for Real-time Semantic Segmentation of Road-driving Images #27

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
DeepTecher opened this issue Mar 25, 2019 · 1 comment
Labels

Comments

@DeepTecher
Copy link
Owner

In Defense of Pre-trained ImageNet Architectures for Real-time Semantic Segmentation of Road-driving Images 🏆 2nd best model for Real-Time Semantic Segmentation on Cityscapes,CVPR2019

提交日期:2019-03-20
团队:萨格勒布大学 电气工程与计算学院
作者:Marin Oršić, Ivan Krešo, Petra Bevandić, Siniša Šegvić

摘要:最近在要求道路驾驶数据集上的语义分割方法的成功引起了对许多相关应用领域的兴趣。其中许多应用涉及移动平台上的实时预测,例如汽车,无人机和各种机器人。由于涉及非凡的计算复杂性,实时设置具有挑战性。许多以前的工作通过定制轻量级架构解决了这一挑战,通过相对于通用架构减少深度,宽度和层容量来降低计算复杂性。我们提出了一种替代方法,可以在各种计算预算中实现明显更好的性能。首先,我们依靠轻量级通用架构作为主要识别引擎。然后,我们利用横向连接的轻量级上采样作为恢复预测分辨率的最具成本效益的解决方案。最后,我们建议通过以新颖的方式融合多种分辨率的共享特征来扩大感受野。几个道路驾驶数据集的实验显示了所提出方法的显着优势,无论是使用ImageNet预训练参数还是我们从头学习。我们的Cityscapes测试提交名为SwiftNetRN-18,可提供75.5%的MIoU,并在GTX1080Ti上的1024x2048图像上实现39.9Hz。

@DeepTecher
Copy link
Owner Author

代码:orsic/swiftnet

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
Projects
None yet
Development

No branches or pull requests

1 participant