History

Xia Li 李夏 dbca8b44a9 [Feature] Support EMANet (#34 ) * add emanet * fixed bug and typos * add emanet config * fixed padding * fixed identity * rename * rename * add concat_input * fallback to update last * Fixed concat * update EMANet * Add tests * remove self-implement norm Co-authored-by: Jiarui XU <xvjiarui0826@gmail.com>		2020-09-07 13:06:59 +08:00
..
README.md	[Feature] Support EMANet (#34 )	2020-09-07 13:06:59 +08:00
emanet_r50-d8_512x1024_80k_cityscapes.py	[Feature] Support EMANet (#34 )	2020-09-07 13:06:59 +08:00
emanet_r50-d8_769x769_80k_cityscapes.py	[Feature] Support EMANet (#34 )	2020-09-07 13:06:59 +08:00
emanet_r101-d8_512x1024_80k_cityscapes.py	[Feature] Support EMANet (#34 )	2020-09-07 13:06:59 +08:00
emanet_r101-d8_769x769_80k_cityscapes.py	[Feature] Support EMANet (#34 )	2020-09-07 13:06:59 +08:00

README.md

Expectation-Maximization Attention Networks for Semantic Segmentation

Introduction

@inproceedings{li2019expectation,
  title={Expectation-maximization attention networks for semantic segmentation},
  author={Li, Xia and Zhong, Zhisheng and Wu, Jianlong and Yang, Yibo and Lin, Zhouchen and Liu, Hong},
  booktitle={Proceedings of the IEEE International Conference on Computer Vision},
  pages={9167--9176},
  year={2019}
}

Results and models

Cityscapes

Method	Backbone	Crop Size	Lr schd	Mem (GB)	Inf time (fps)	mIoU	mIoU(ms+flip)	download
EMANet	R-50-D8	512x1024	80000	5.4	4.58	77.59	79.44	model \| log
EMANet	R-101-D8	512x1024	80000	6.2	2.87	79.10	81.21	model \| log
EMANet	R-50-D8	769x769	80000	8.9	1.97	79.33	80.49	model \| log
EMANet	R-101-D8	769x769	80000	10.1	1.22	79.62	81.00	model \| log