class SelfAttention(nn.Module): def __init__(self, in_channels, reduction=4): super(SelfAttention, self).__init__() self.avg_pool = nn.AdaptiveAvgPool1d(1) self.fc1 = nn.Conv1d(in_channels, in_channels // reduction, 1, bias=False) self.relu = nn.ReLU(inplace=True) self.fc2 = nn.Conv1d(in_channels // reduction, in_channels, 1, bias=False) self.sigmoid = nn.Sigmoid() def forward(self, x): b, c, n = x.size() y = self.avg_pool(x) y = self.fc1(y) y = self.relu(y) y = self.fc2(y) y = self.sigmoid(y) return x * y.expand_as(x) def get_model(input_channels=6, use_xyz=True): return Pointnet2MSG(input_channels=input_channels, use_xyz=use_xyz) class Pointnet2MSG(nn.Module): def __init__(self, input_channels=6, use_xyz=True): super().__init__() self.SA_modules = nn.ModuleList() channel_in = input_channels skip_channel_list = [input_channels] for k in range(cfg.RPN.SA_CONFIG.NPOINTS.len()): mlps = cfg.RPN.SA_CONFIG.MLPS[k].copy() channel_out = 0 for idx in range(mlps.len()): mlps[idx] = [channel_in] + mlps[idx] channel_out += mlps[idx][-1] mlps.append(channel_out) self.SA_modules.append( nn.Sequential( PointnetSAModuleMSG( npoint=cfg.RPN.SA_CONFIG.NPOINTS[k], radii=cfg.RPN.SA_CONFIG.RADIUS[k], nsamples=cfg.RPN.SA_CONFIG.NSAMPLE[k], mlps=mlps, use_xyz=use_xyz, bn=cfg.RPN.USE_BN ), SelfAttention(channel_out) ) ) skip_channel_list.append(channel_out) channel_in = channel_out self.FP_modules = nn.ModuleList() for k in range(cfg.RPN.FP_MLPS.len()): pre_channel = cfg.RPN.FP_MLPS[k + 1][-1] if k + 1 < len(cfg.RPN.FP_MLPS) else channel_out self.FP_modules.append( PointnetFPModule( mlp=[pre_channel + skip_channel_list[k]] + cfg.RPN.FP_MLPS[k] ) ) def _break_up_pc(self, pc): xyz = pc[..., 0:3].contiguous() features = ( pc[..., 3:].transpose(1, 2).contiguous() if pc.size(-1) > 3 else None ) return xyz, features def forward(self, pointcloud: torch.cuda.FloatTensor): xyz, features = self._break_up_pc(pointcloud) l_xyz, l_features = [xyz], [features] for i in range(len(self.SA_modules)): li_xyz, li_features = self.SA_modules[i](l_xyz[i], l_features[i]) l_xyz.append(li_xyz) l_features.append(li_features) for i in range(-1, -(len(self.FP_modules) + 1), -1): l_features[i - 1] = self.FP_modules[i]( l_xyz[i - 1], l_xyz[i], l_features[i - 1], l_features[i] ) return l_xyz[0], l_features[0]在forward函数中,如果我要使用channel_out变量传入SA_modules中,我该如何在forward函数中计算并得到它,再传入SA_modules中,你可以给我详细的代码吗?
时间: 2023-07-16 20:12:55 浏览: 162
PAPR.rar_PAPR_in_ofdm papr_ofdm papr reduction_papr reduction
5星 · 资源好评率100%
在 `Pointnet2MSG` 的构造函数中,我们可以看到 SA_modules 是一个 `nn.ModuleList()`,其中每个元素都是一个 `nn.Sequential()`,包含了一个 `PointnetSAModuleMSG` 和一个 `SelfAttention`。
在 `PointnetSAModuleMSG` 中,我们可以看到最后一个 MLP 输出的通道数即为 `channel_out`。因此,我们可以在每个 `PointnetSAModuleMSG` 的定义中记录 `channel_out`,并将其传递给下一个 `PointnetSAModuleMSG` 的输入。
以下是修改后的代码:
```python
class Pointnet2MSG(nn.Module):
def __init__(self, input_channels=6, use_xyz=True):
super().__init__()
self.SA_modules = nn.ModuleList()
channel_in = input_channels
skip_channel_list = [input_channels]
for k in range(cfg.RPN.SA_CONFIG.NPOINTS.len()):
mlps = cfg.RPN.SA_CONFIG.MLPS[k].copy()
channel_out = 0
for idx in range(mlps.len()):
mlps[idx] = [channel_in] + mlps[idx]
channel_out += mlps[idx][-1]
mlps.append(channel_out)
sa_module = nn.Sequential(
PointnetSAModuleMSG(
npoint=cfg.RPN.SA_CONFIG.NPOINTS[k],
radii=cfg.RPN.SA_CONFIG.RADIUS[k],
nsamples=cfg.RPN.SA_CONFIG.NSAMPLE[k],
mlps=mlps,
use_xyz=use_xyz,
bn=cfg.RPN.USE_BN
)
)
self.SA_modules.append(sa_module)
skip_channel_list.append(channel_out)
channel_in = channel_out
self.FP_modules = nn.ModuleList()
for k in range(cfg.RPN.FP_MLPS.len()):
pre_channel = cfg.RPN.FP_MLPS[k + 1][-1] if k + 1 < len(cfg.RPN.FP_MLPS) else channel_out
self.FP_modules.append(
PointnetFPModule(
mlp=[pre_channel + skip_channel_list[k]] + cfg.RPN.FP_MLPS[k]
)
)
def _break_up_pc(self, pc):
xyz = pc[..., 0:3].contiguous()
features = (
pc[..., 3:].transpose(1, 2).contiguous()
if pc.size(-1) > 3 else None
)
return xyz, features
def forward(self, pointcloud: torch.cuda.FloatTensor):
xyz, features = self._break_up_pc(pointcloud)
l_xyz, l_features = [xyz], [features]
for i in range(len(self.SA_modules)):
sa_module = self.SA_modules[i]
li_xyz, li_features = sa_module(l_xyz[i], l_features[i])
l_xyz.append(li_xyz)
l_features.append(li_features)
channel_out = sa_module[-1].fc2.out_channels
for i in range(-1, -(len(self.FP_modules) + 1), -1):
l_features[i - 1] = self.FP_modules[i](
l_xyz[i - 1], l_xyz[i], l_features[i - 1], l_features[i]
)
return l_xyz[0], l_features[0], channel_out
```
使用上述代码,我们可以在 forward 函数中得到 `channel_out`,并将其传递给下一个 `PointnetSAModuleMSG` 的输入。
阅读全文