BiSeNetv2(pytorch)测试、训练cityscapes

**BiSeNetv2 (PyTorch) 测试与训练 Cityscapes**

###1. 概述本文将介绍如何使用 PyTorch 实现 BiSeNetv2 网络，并在 Cityscapes 数据集上进行测试和训练。

###2. 环境准备首先，我们需要安装必要的库：

bashpip install torch torchvision

bashwget  -O cityscapes.zipunzip cityscapes.zip

import torchimport torch.nn as nnclass BiSeNetv2(nn.Module):
 def __init__(self, num_classes=19):
 super(BiSeNetv2, self).__init__()
 self.encoder = nn.Sequential(
 nn.Conv2d(3,64, kernel_size=3),
 nn.ReLU(),
 nn.MaxPool2d(kernel_size=2),
 nn.Conv2d(64,128, kernel_size=3),
 nn.ReLU(),
 nn.MaxPool2d(kernel_size=2),
 nn.Conv2d(128,256, kernel_size=3),
 nn.ReLU(),
 nn.MaxPool2d(kernel_size=2)
 )
 self.decoder = nn.Sequential(
 nn.Upsample(scale_factor=2),
 nn.Conv2d(256,128, kernel_size=3),
 nn.ReLU(),
 nn.Upsample(scale_factor=2),
 nn.Conv2d(128,64, kernel_size=3),
 nn.ReLU(),
 nn.Conv2d(64, num_classes, kernel_size=1)
 )
 def forward(self, x):
 x = self.encoder(x)
 x = self.decoder(x)
 return x

from torch.utils.data import Dataset, DataLoaderimport torchvision.transforms as transformsclass CityscapesDataset(Dataset):
 def __init__(self, root_dir, transform=None):
 self.root_dir = root_dir self.transform = transform def __len__(self):
 return len(os.listdir(self.root_dir))
 def __getitem__(self, idx):
 img_path = os.path.join(self.root_dir, f"train/{idx}.jpg")
 label_path = os.path.join(self.root_dir, f"label/{idx}.png")
 image = Image.open(img_path)
 label = Image.open(label_path)
 if self.transform:
 image = self.transform(image)
 return image, labeltransform = transforms.Compose([transforms.ToTensor()])
dataset = CityscapesDataset(root_dir="path/to/cityscapes", transform=transform)

dataloader = DataLoader(dataset, batch_size=32, shuffle=True)

model = BiSeNetv2(num_classes=19)
criterion = nn.CrossEntropyLoss()
optimizer = torch.optim.Adam(model.parameters(), lr=0.001)

for epoch in range(10):
 for images, labels in dataloader:
 optimizer.zero_grad()
 outputs = model(images)
 loss = criterion(outputs, labels)
 loss.backward()
 optimizer.step()
 print(f"Epoch {epoch+1}, Loss: {loss.item():.4f}")

test_dataset = CityscapesDataset(root_dir="path/to/cityscapes", transform=transform)
test_dataloader = DataLoader(test_dataset, batch_size=32, shuffle=False)

model.eval()
total_correct =0with torch.no_grad():
 for images, labels in test_dataloader:
 outputs = model(images)
 _, predicted = torch.max(outputs, dim=1)
 total_correct += (predicted == labels).sum().item()
print(f"Test Accuracy: {total_correct / len(test_dataset) *100:.2f}%")