Skip to content

昇思开源实习：在线推理案例04-FCN基于PyNative模式改写成以mint接口实现 #116

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 1 commit into
base: master
Choose a base branch
from
Open

昇思开源实习：在线推理案例04-FCN基于PyNative模式改写成以mint接口实现 #116

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
91 changes: 46 additions & 45 deletions Online/inference/04-FCN/mindspore_fcn8s.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -382,6 +382,7 @@
"outputs": [],
"source": [
"import mindspore.nn as nn\n",
"import mindspore as mint\n",
"\n",
"class FCN8s(nn.Cell):\n",
" def __init__(self, n_class):\n",
Expand All @@ -395,89 +396,89 @@
" self.n_class = n_class\n",
" # 卷积块1：两个3x3卷积层和BN层，后接最大池化层\n",
" self.conv1 = nn.SequentialCell(\n",
" nn.Conv2d(in_channels=3, out_channels=64,\n",
" mint.nn.Conv2d(in_channels=3, out_channels=64,\n",
" kernel_size=3, weight_init='xavier_uniform'),\n",
" nn.BatchNorm2d(64),\n",
" nn.ReLU(),\n",
" nn.Conv2d(in_channels=64, out_channels=64,\n",
" mint.nn.BatchNorm2d(64),\n",
" mint.nn.ReLU(),\n",
" mint.nn.Conv2d(in_channels=64, out_channels=64,\n",
" kernel_size=3, weight_init='xavier_uniform'),\n",
" nn.BatchNorm2d(64),\n",
" nn.ReLU()\n",
" mint.nn.BatchNorm2d(64),\n",
" mint.nn.ReLU()\n",
" )\n",
" self.pool1 = nn.MaxPool2d(kernel_size=2, stride=2)\n",
" # 卷积块2：两个3x3卷积层和BN层，后接最大池化层\n",
" self.conv2 = nn.SequentialCell(\n",
" nn.Conv2d(in_channels=64, out_channels=128,\n",
" mint.nn.Conv2d(in_channels=64, out_channels=128,\n",
" kernel_size=3, weight_init='xavier_uniform'),\n",
" nn.BatchNorm2d(128),\n",
" nn.ReLU(),\n",
" nn.Conv2d(in_channels=128, out_channels=128,\n",
" mint.nn.BatchNorm2d(128),\n",
" mint.nn.ReLU(),\n",
" mint.nn.Conv2d(in_channels=128, out_channels=128,\n",
" kernel_size=3, weight_init='xavier_uniform'),\n",
" nn.BatchNorm2d(128),\n",
" nn.ReLU()\n",
" mint.nn.BatchNorm2d(128),\n",
" mint.nn.ReLU()\n",
" )\n",
" self.pool2 = nn.MaxPool2d(kernel_size=2, stride=2)\n",
" # 卷积块3：三个3x3卷积层和BN层，后接最大池化层\n",
" self.conv3 = nn.SequentialCell(\n",
" nn.Conv2d(in_channels=128, out_channels=256,\n",
" mint.nn.Conv2d(in_channels=128, out_channels=256,\n",
" kernel_size=3, weight_init='xavier_uniform'),\n",
" nn.BatchNorm2d(256),\n",
" nn.ReLU(),\n",
" nn.Conv2d(in_channels=256, out_channels=256,\n",
" mint.nn.BatchNorm2d(256),\n",
" mint.nn.ReLU(),\n",
" mint.nn.Conv2d(in_channels=256, out_channels=256,\n",
" kernel_size=3, weight_init='xavier_uniform'),\n",
" nn.BatchNorm2d(256),\n",
" nn.ReLU(),\n",
" nn.Conv2d(in_channels=256, out_channels=256,\n",
" mint.nn.BatchNorm2d(256),\n",
" mint.nn.ReLU(),\n",
" mint.nn.Conv2d(in_channels=256, out_channels=256,\n",
" kernel_size=3, weight_init='xavier_uniform'),\n",
" nn.BatchNorm2d(256),\n",
" nn.ReLU()\n",
" mint.nn.BatchNorm2d(256),\n",
" mint.nn.ReLU()\n",
" )\n",
" self.pool3 = nn.MaxPool2d(kernel_size=2, stride=2)\n",
" # 卷积块4：三个3x3卷积层和BN层，后接最大池化层\n",
" self.conv4 = nn.SequentialCell(\n",
" nn.Conv2d(in_channels=256, out_channels=512,\n",
" mint.nn.Conv2d(in_channels=256, out_channels=512,\n",
" kernel_size=3, weight_init='xavier_uniform'),\n",
" nn.BatchNorm2d(512),\n",
" nn.ReLU(),\n",
" nn.Conv2d(in_channels=512, out_channels=512,\n",
" mint.nn.BatchNorm2d(512),\n",
" mint.nn.ReLU(),\n",
" mint.nn.Conv2d(in_channels=512, out_channels=512,\n",
" kernel_size=3, weight_init='xavier_uniform'),\n",
" nn.BatchNorm2d(512),\n",
" nn.ReLU(),\n",
" nn.Conv2d(in_channels=512, out_channels=512,\n",
" mint.nn.BatchNorm2d(512),\n",
" mint.nn.ReLU(),\n",
" mint.nn.Conv2d(in_channels=512, out_channels=512,\n",
" kernel_size=3, weight_init='xavier_uniform'),\n",
" nn.BatchNorm2d(512),\n",
" nn.ReLU()\n",
" mint.nn.BatchNorm2d(512),\n",
" mint.nn.ReLU()\n",
" )\n",
" self.pool4 = nn.MaxPool2d(kernel_size=2, stride=2)\n",
" # 卷积块5：三个3x3卷积层和BN层，后接最大池化层\n",
" self.conv5 = nn.SequentialCell(\n",
" nn.Conv2d(in_channels=512, out_channels=512,\n",
" mint.nn.Conv2d(in_channels=512, out_channels=512,\n",
" kernel_size=3, weight_init='xavier_uniform'),\n",
" nn.BatchNorm2d(512),\n",
" nn.ReLU(),\n",
" nn.Conv2d(in_channels=512, out_channels=512,\n",
" mint.nn.BatchNorm2d(512),\n",
" mint.nn.ReLU(),\n",
" mint.nn.Conv2d(in_channels=512, out_channels=512,\n",
" kernel_size=3, weight_init='xavier_uniform'),\n",
" nn.BatchNorm2d(512),\n",
" nn.ReLU(),\n",
" nn.Conv2d(in_channels=512, out_channels=512,\n",
" mint.nn.BatchNorm2d(512),\n",
" mint.nn.ReLU(),\n",
" mint.nn.Conv2d(in_channels=512, out_channels=512,\n",
" kernel_size=3, weight_init='xavier_uniform'),\n",
" nn.BatchNorm2d(512),\n",
" nn.ReLU()\n",
" mint.nn.BatchNorm2d(512),\n",
" mint.nn.ReLU()\n",
" )\n",
" self.pool5 = nn.MaxPool2d(kernel_size=2, stride=2)\n",
" # 全卷积层6：7x7卷积层和BN层，用于代替全连接层\n",
" self.conv6 = nn.SequentialCell(\n",
" nn.Conv2d(in_channels=512, out_channels=4096,\n",
" mint.nn.Conv2d(in_channels=512, out_channels=4096,\n",
" kernel_size=7, weight_init='xavier_uniform'),\n",
" nn.BatchNorm2d(4096),\n",
" nn.ReLU(),\n",
" mint.nn.BatchNorm2d(4096),\n",
" mint.nn.ReLU(),\n",
" )\n",
" # 全卷积层7：1x1卷积层和BN层，用于代替全连接层\n",
" self.conv7 = nn.SequentialCell(\n",
" nn.Conv2d(in_channels=4096, out_channels=4096,\n",
" mint.nn.Conv2d(in_channels=4096, out_channels=4096,\n",
" kernel_size=1, weight_init='xavier_uniform'),\n",
" nn.BatchNorm2d(4096),\n",
" nn.ReLU(),\n",
" mint.nn.BatchNorm2d(4096),\n",
" mint.nn.ReLU(),\n",
" )\n",
" # 产生分数图的层\n",
" self.score_fr = nn.Conv2d(in_channels=4096, out_channels=self.n_class,\n",
Expand Down