- 博客(6)
- 收藏
- 关注
RSS订阅原创 Pytorch1.5升级1.6 CPP代码修改（RuntimeError: Error compiling objects for extension）
Pytorch1.5升级1.6 CPP代码修改（RuntimeError: Error compiling objects for extension） AT_CHECK和THCState_getCurrentStream(state) 用 TORCH_CHECK 和 at::cuda::getCurrentCUDAStream()替代。 参考https://github.com/hyangwinter/flownet3d_pytorch/issues/8#issuecomment-663925414 .
2020-10-14 00:24:34
3225
3
原创 SEMEDA: Enhancing Segmentation Precision with Semantic Edge Aware Loss小记
论文地址：https://arxiv.org/abs/1905.01892 方法流程如下图所示： E∗E^*E∗, E^\hat{E}E^, S∗S^*S∗, S^\hat{S}S^分别表示边缘真值图，边缘预测图，掩模真值图，掩模预测图，E∗^\hat{E^*}E∗^, 表示将掩模真值图输入边缘预测网络ϕ\phiϕ后得到的边缘预测图谱。首先训练边缘提取网络ϕ\phiϕ。再使用论文中提出的SEME...
2019-05-09 19:38:03
624
原创 Fully Convolutional Networks for Semantic Segmentation
1. Training the whole image is more efficient and equally effective.When these receptive fields overlap significantly, both feedforward computation and backpropagation are much more efficient when c
2017-12-29 16:44:16
335
原创 Classification with an edge: Improving semantic image segmentation with boundary detection
Classification with an edge: Improving semantic image segmentation with boundary detectionNetwroks:SEG-H encoder-decoder networkIt’s a crossbreed of FCN and encoder-decoder architecture. Use pyramid-bo
2017-12-25 17:04:15
1228
原创 A new deep convolutional neural network for fast hyperspectral image classification Review
A new deep convolutional neural network for fast hyperspectral image classification ReviewBasically, this paper imply a 3D CNN.Points 1:Three ways to apply CNN on remote sensing image Extract only spec
2017-12-21 11:30:34
844
原创 Pytorch小记（一）
一. 类型转换#define a tensor a = torch.zeros(10).type(torch.LongTensor) a = torch.IntTensor(a)
2017-12-21 10:36:33
284
空空如也
空空如也
TA创建的收藏夹 TA关注的收藏夹
TA关注的人