R-FCN: Object Detection via Region-based Fully Convolutional Networks (NIPS 16), Arxiv 16.05

最新推荐文章于 2022-07-19 18:39:35 发布

_menglan_Zi

最新推荐文章于 2022-07-19 18:39:35 发布

阅读量5.1k

点赞数 2

分类专栏： cvpr Deep Learning detection 文章标签： FCN

本文链接：https://blog.youkuaiyun.com/zimenglan_sysu/article/details/52384316

版权

 
 R-FCN: Object Detection via Region-based Fully Convolutional Networks (NIPS 16), Arxiv 16.05 

 
 ===== 

 
 blogs： 

  论文阅读]R-FCN: Object Detection via Region-based Fully Convolutional Networks： 
  http://blog.youkuaiyun.com/u012361214/article/details/51507590 

  论文笔记 | R-FCN: Object Detection via Region-based Fully Convolutional Networks： 
 http://blog.youkuaiyun.com/bea_tree/article/details/51817263 

  R-FCN: Object Detection via Region-based Fully Convolutional Networks： 
  http://blog.youkuaiyun.com/surgewong/article/details/51873372 

 
 ===== 

 
 论文： 

  R-FCN - Object Detection via Region-based Fully Convolutional Networks 

 
 https://arxiv.org/pdf/1605.06409v2.pdf 

 
 ===== 

 
 开源代码： 

 
 https://github.com/daijifeng001/R-FCN 

  现在有开源的py code： 

  py-R-FCN: 
  https://github.com/Orpine/py-R-FCN

  py-R-FCN-multiGPU: 
  https://github.com/bharatsingh430/py-R-FCN-multiGPU/

 
 ===== 

 
 如何将R-FCN整合到py-faster-rcnn？ 

  详见： 
 https://github.com/daijifeng001/R-FCN/issues/9 

  成功将rfcn的matlab代码整合到py-faster-rcnn中，跑出来的实验为： 

  train：07+12 trainval 

  test： 07 test 

  resnet50：74.24%mAP （0.10s～0.12s per image） 

  resnet101： 76.23% mAP 
 （0.14s～0.16s per image） 

 
 --------------------------