当前位置:网站首页>Real time semantic segmentation

Real time semantic segmentation

2020-11-18 06:11:33 ShellCollector

 

A conversion method for video object segmentation , In the form of tag propagation , Simple high performance 、 high efficiency .

[9].A Transductive Approach for Video Object Segmentation

author | Yizhuo Zhang, Zhirong Wu, Houwen Peng, Stephen Lin

Company | Microsoft research Asia ; Carnegie Mellon University

Code | https://github.com/microsoft/transductive-vos.pytorch

Model :46m davis_train.pth.tar

 

author | Andreas Robinson, Felix Jaremo Lawin, Martin Danelljan, Fahad Shahbaz Khan, Michael Felsberg

Company | Linkoping University, Sweden;ETH Zurich, Switzerland;IIAI

Code | https://github.com/andr345/frtm-vos

need c++ Code :

rn18_ytvos.pth ResNet18 YouTubeVOS download
rn18_all.pth ResNet18 YouTubeVOS + DAVIS download
rn101_ytvos.pth ResNet101 YouTubeVOS download
rn101_all.pth ResNet101 YouTubeVOS + DAVIS download
rn101_dv.pth ResNet101 DAVIS download

rn18_ytvos.pth 4.7m

 

Some references :https://blog.csdn.net/jacke121/article/details/108838979

 

The paper :DFANet: Deep Feature Aggregation for Real-Time Semantic Segmentation

github:https://github.com/huaifeng1993/DFANet

 

CVPR2019, Look beyond the works of science and technology

 

platform input size batch size inference time /ms
rk3399 320*200 1 960
rk3399 200*100 1 589
rk3399 80*80 1 259
rk3399 72*72 1 161
2080 1024*1024 4 40
2080 1024*1024 1 16
2080 2048*1024 1 17
2080 2048*1024 2 39
2080 512*512 1 39
2080 512*512 16 44

版权声明
本文为[ShellCollector]所创,转载请带上原文链接,感谢

随机推荐