ann

Asymmetric Non-local Neural Networks for Semantic Segmentation

Introduction

@inproceedings{annn,
  author    = {Zhen Zhu and
               Mengde Xu and
               Song Bai and
               Tengteng Huang and
               Xiang Bai},
  title     = {Asymmetric Non-local Neural Networks for Semantic Segmentation},
  booktitle={International Conference on Computer Vision},
  year      = {2019},
  url       = {http://arxiv.org/abs/1908.07678},
}

Results and models

Cityscapes

Method	Backbone	Crop Size	Lr schd	Mem (GB)	Inf time (fps)	mIoU	mIoU(ms+flip)	download
ANN	R-50-D8	512x1024	40000	6	3.71	77.40	78.57	model | log
ANN	R-101-D8	512x1024	40000	9.5	2.55	76.55	78.85	model | log
ANN	R-50-D8	769x769	40000	6.8	1.70	78.89	80.46	model | log
ANN	R-101-D8	769x769	40000	10.7	1.15	79.32	80.94	model | log
ANN	R-50-D8	512x1024	80000	-	-	77.34	78.65	model | log
ANN	R-101-D8	512x1024	80000	-	-	77.14	78.81	model | log
ANN	R-50-D8	769x769	80000	-	-	78.88	80.57	model | log
ANN	R-101-D8	769x769	80000	-	-	78.80	80.34	model | log

ADE20K

Method	Backbone	Crop Size	Lr schd	Mem (GB)	Inf time (fps)	mIoU	mIoU(ms+flip)	download
ANN	R-50-D8	512x512	80000	9.1	21.01	41.01	42.30	model | log
ANN	R-101-D8	512x512	80000	12.5	14.12	42.94	44.18	model | log
ANN	R-50-D8	512x512	160000	-	-	41.74	42.62	model | log
ANN	R-101-D8	512x512	160000	-	-	42.94	44.06	model | log

Pascal VOC 2012 + Aug

Method	Backbone	Crop Size	Lr schd	Mem (GB)	Inf time (fps)	mIoU	mIoU(ms+flip)	download
ANN	R-50-D8	512x512	20000	6	20.92	74.86	76.13	model | log
ANN	R-101-D8	512x512	20000	9.5	13.94	77.47	78.70	model | log
ANN	R-50-D8	512x512	40000	-	-	76.56	77.51	model | log
ANN	R-101-D8	512x512	40000	-	-	76.70	78.06	model | log