PyTorch torch.nn.Conv3d 函数
PyTorch torch.nn 参考手册torch.nn.Conv3d 是 PyTorch 中的三维卷积模块。
它处理三维输入，如视频或医学图像（深度 x 高 x 宽）。
函数定义
torch.nn.Conv3d(in_channels, out_channels, kernel_size, stride=1, padding=0, dilation=1, groups=1, bias=True)
输入形状
(batch, channels, depth, height, width)
使用示例
示例 1: 基本用法
实例
import torch
import torch.nn as nn
# 3D 卷积
conv3d = nn.Conv3d(in_channels=3, out_channels=32, kernel_size=3, padding=1)
# 输入：batch=2，通道=3，深度=16，高=32，宽=32
x = torch.randn(2, 3, 16, 32, 32)
output = conv3d(x)
print("输入形状:", x.shape)
print("输出形状:", output.shape)
import torch.nn as nn
# 3D 卷积
conv3d = nn.Conv3d(in_channels=3, out_channels=32, kernel_size=3, padding=1)
# 输入：batch=2，通道=3，深度=16，高=32，宽=32
x = torch.randn(2, 3, 16, 32, 32)
output = conv3d(x)
print("输入形状:", x.shape)
print("输出形状:", output.shape)
示例 2: 视频分类
实例
import torch
import torch.nn as nn
class VideoCNN(nn.Module):
def __init__(self, num_classes=10):
super(VideoCNN, self).__init__()
self.conv1 = nn.Conv3d(3, 32, kernel_size=3, padding=1)
self.pool = nn.MaxPool3d(2)
self.conv2 = nn.Conv3d(32, 64, kernel_size=3, padding=1)
self.gap = nn.AdaptiveAvgPool3d(1)
self.fc = nn.Linear(64, num_classes)
def forward(self, x):
x = torch.relu(self.conv1(x))
x = self.pool(x)
x = torch.relu(self.conv2(x))
x = self.gap(x)
x = x.view(x.size(0), -1)
return self.fc(x)
model = VideoCNN()
x = torch.randn(4, 3, 16, 112, 112) # 16帧视频
output = model(x)
print("视频:", x.shape, "-> 类别:", output.shape)
import torch.nn as nn
class VideoCNN(nn.Module):
def __init__(self, num_classes=10):
super(VideoCNN, self).__init__()
self.conv1 = nn.Conv3d(3, 32, kernel_size=3, padding=1)
self.pool = nn.MaxPool3d(2)
self.conv2 = nn.Conv3d(32, 64, kernel_size=3, padding=1)
self.gap = nn.AdaptiveAvgPool3d(1)
self.fc = nn.Linear(64, num_classes)
def forward(self, x):
x = torch.relu(self.conv1(x))
x = self.pool(x)
x = torch.relu(self.conv2(x))
x = self.gap(x)
x = x.view(x.size(0), -1)
return self.fc(x)
model = VideoCNN()
x = torch.randn(4, 3, 16, 112, 112) # 16帧视频
output = model(x)
print("视频:", x.shape, "-> 类别:", output.shape)
示例 3: 参数计算
实例
import torch
import torch.nn as nn
conv = nn.Conv3d(64, 128, kernel_size=3, padding=1)
# 参数量计算: out_ch * in_ch * k_d * k_h * k_w + bias
params = 128 * 64 * 3 * 3 * 3
print("参数量:", params)
print("权重形状:", conv.weight.shape)
import torch.nn as nn
conv = nn.Conv3d(64, 128, kernel_size=3, padding=1)
# 参数量计算: out_ch * in_ch * k_d * k_h * k_w + bias
params = 128 * 64 * 3 * 3 * 3
print("参数量:", params)
print("权重形状:", conv.weight.shape)
使用场景
- 视频处理: 动作识别
- 医学影像: CT、MRI
- 3D 分割: 体素数据
注意：3D 卷积计算量大，需要较多显存。