AttributeError：“CarvanaDataset”对象没有属性“image”

我尝试制作一个 Unet 模型。这是我的代码：

import torch
import torch.nn as nn
import torch.optim as optim
import albumentations as A
from albumentations.pytorch import ToTensorV2
from tqdm import tqdm
from model import Unet
from utils import get_loaders
import numpy as np
import random


# Hyperparameters

LEARNING_RATE = 1e-8
BATCH_SIZE = 8
NUM_EPOCH = 2
NUM_WORKERS = 2
PIN_MEMORY = True
DEVICE = "cuda" if torch.cuda.is_available() else "cpu"
print(DEVICE)

TRAIN_IMG_DIR = "./dataset/train_img_dir"
TRAIN_MASK_DIR = "./dataset/train_mask_dir"
VAL_IMG_DIR = "./dataset/val_img_dir"
VAL_MASK_DIR = "./dataset/val_mask_dir"

IMAGE_HEIGHT = 160
IMAGE_WIDTH = 240

train_losses = []
val_acc = []
val_dice = []

# 设置随机种子
seed = random.randint(1, 100)
torch.manual_seed(seed)
torch.cuda.manual_seed(seed)
torch.cuda.manual_seed_all(seed) # 如果你使用多个GPU
np.random.seed(seed)
random.seed(seed)
torch.backends.cudnn.deterministic = True
torch.backends.cudnn.benchmark = False



def train_fn(loader, model, loss_fn, optimizer, scaler):
    loop = tqdm(loader)
    total_loss = 0.0
    for index, (data, target) in enumerate(loop):
        data = data.to(DEVICE)
        target = target.unsqueeze(1).float().to(DEVICE)

        with torch.cuda.amp.autocast():
            predict = model(data)
            loss = loss_fn(predict, target)

        optimizer.zero_grad()
        scaler.scale(loss).backward()
        scaler.step(optimizer)
        scaler.update()

        total_loss += loss.item()

        loop.set_postfix(loss=loss.item())
    return total_loss / len(loader)


def check_accuracy(loader, model, DEVICE="cuda"):
    num_correct = 0
    num_pixels = 0
    dice_score = 0
    model.eval()

    with torch.no_grad():
        for x, y in loader:
            x = x.to(DEVICE)
            y = y.unsqueeze(1).to(DEVICE)
            predictions = torch.sigmoid(model(x))
            predictions = (predictions > 0.5).float()
            num_correct += (predictions == y).sum()
            num_pixels += torch.numel(predictions)
            dice_score += (2 * (predictions * y).sum()) / (2 * (predictions * y).sum()+((predictions*y)<1).sum())

    accuracy = round(float(num_correct / num_pixels), 4)
    dice = round(float(dice_score/len(loader)), 4)

    print(f"Got {num_correct}/{num_pixels} with acc {num_correct/num_pixels * 100 :.2f}")
    print(f"Dice Score:{dice_score/len(loader)}")

    model.train()

    return accuracy, dice


def main():
    train_transform = A.Compose(
        [
            A.Resize(height=IMAGE_HEIGHT, width=IMAGE_WIDTH),
            A.Rotate(limit=35, p=1.0),
            A.HorizontalFlip(p=0.5),
            A.VerticalFlip(p=0.1),
            A.Normalize(
                mean=[0.0, 0.0, 0.0],
                std=[1.0, 1.0, 1.0],
                max_pixel_value=255.0,
            ),
            ToTensorV2(),
        ],)

    val_transform = A.Compose(
        [
            A.Resize(height=IMAGE_HEIGHT, width=IMAGE_WIDTH),
            A.Normalize(
                mean=[0.0, 0.0, 0.0],
                std=[1.0, 1.0, 1.0],
                max_pixel_value=255.0,
            ),
            ToTensorV2(),
        ], )

    train_loader, val_loader = get_loaders(TRAIN_IMG_DIR, TRAIN_MASK_DIR, VAL_IMG_DIR, VAL_MASK_DIR,
                                           train_transform, val_transform,
                                           BATCH_SIZE, NUM_WORKERS, PIN_MEMORY)

    model = Unet(in_channel=3, out_channel=1).to(device=DEVICE)
    loss_fn = nn.BCEWithLogitsLoss()
    optimizer = optim.Adam(model.parameters(), lr=LEARNING_RATE)
    scaler = torch.cuda.amp.GradScaler()

    for index in range(NUM_EPOCH):
        print("Current Epoch: ", index)
        train_loss = train_fn(train_loader, model, loss_fn, optimizer, scaler)
        train_losses.append(train_loss)

        accuracy, dice = check_accuracy(val_loader, model, DEVICE=DEVICE)
        val_acc.append(accuracy)
        val_dice.append(dice)


if __name__ == "__main__":
    main()

我得到了这个错误：

D:\anaconda3\envs\Unet\python.exe C:\Users\xxx\PycharmProjects\Study\pythonProject3\train.py 库达当前纪元：0 0it [00:00, ?it/s] 回溯（最近一次调用最后一次）：文件“C:\Users\xxx\PycharmProjects\Study\pythonProject3\train.py”，第 142 行，位于主要的（）文件“C:\Users\xxx\PycharmProjects\Study\pythonProject3\train.py”，第 133 行，在 main 中 train_loss = train_fn(train_loader、模型、loss_fn、优化器、缩放器) 文件“C:\Users\xxx\PycharmProjects\Study\pythonProject3\train.py”，第 50 行，在 train_fn 中对于枚举（循环）中的索引（数据，目标）：文件“D:\anaconda3\envs\Unet\lib\site-packages\tqdm\std.py”，第 1181 行，在 iter for obj in iterable 中：文件“D:\anaconda3\envs\Unet\lib\site-packages\torch\utils\data\dataloader.py”，第 435 行， iter return self._get_iterator() 文件“D:\anaconda3\envs\Unet\lib\site-packages\torch\utils\data\dataloader.py”，第 381 行，在 _get_iterator 中返回 _MultiProcessingDataLoaderIter(self) 文件“D:\anaconda3\envs\Unet\lib\site-packages\torch\utils\data\dataloader.py”，第 1072 行，in init self._reset(loader,first_iter=True) 文件“D:\anaconda3\envs\Unet\lib\site-packages\torch\utils\data\dataloader.py”，第 1105 行，在 _reset 中 self._try_put_index() 文件“D:\anaconda3\envs\Unet\lib\site-packages\torch\utils\data\dataloader.py”，第 1339 行，位于 _try_put_index 中索引 = self._next_index() 文件“D:\anaconda3\envs\Unet\lib\site-packages\torch\utils\data\dataloader.py”，第 618 行，在 _next_index 中 return next(self._sampler_iter) # 可能会引发 StopIteration 文件“D:\anaconda3\envs\Unet\lib\site-packages\torch\utils\data\sampler.py”，第 254 行，in iter for self.sampler 中的 idx：文件“D:\anaconda3\envs\Unet\lib\site-packages\torch\utils\data\sampler.py”，第 76 行，in iter return iter(range(len(self.data_source) ））文件“C:\Users\xxx\PycharmProjects\Study\pythonProject3\dataset.py”，第 15 行，in len return len(self.image) AttributeError：“CarvanaDataset”对象没有属性“image” 进程已完成，退出代码为 1

我该如何解决此问题？

错误信息 "AttributeError: 'CarvanaDataset' object has no attribute 'image'" 表明正在尝试访问 CarvanaDataset 类中不存在的 image 属性。这个错误出现在尝试获取数据集长度时，这意味着问题出在的 CarvanaDataset 类的 __len__ 函数中。

需要确保在 CarvanaDataset 类中正确定义了 self.image 。它应该在 __init__ 方法中初始化，并且存储图像路径的列表或图像数据本身。

以下是修复此错误的步骤以及改进代码的建议：

1. 检查 CarvanaDataset 类:

确保在 utils.py 文件中定义了 CarvanaDataset 类。
验证 __init__ 方法是否正确初始化了 self.image ，并将其设置为图像路径列表或图像数据。

```python class CarvanaDataset(Dataset): def init (self, image_dir, mask_dir, transform=None): self.image_dir = image_dir self.mask_dir = mask_dir self.transform = transform self.images = os.listdir(image_dir) # 获取所有图像文件名

   def __len__(self):
       return len(self.images)

   def __getitem__(self, index):
       img_path = os.path.join(self.image_dir, self.images[index])
       mask_path = os.path.join(self.mask_dir, self.images[index].replace('.jpg', '_mask.gif')) 
       # 假设掩码文件名与图像文件名相关

       image = np.array(Image.open(img_path).convert("RGB"))
       mask = np.array(Image.open(mask_path).convert("L"), dtype=np.float32) 
       # 将掩码转换为单通道

       if self.transform is not None:
           augmentations = self.transform(image=image, mask=mask)
           image = augmentations["image"]
           mask = augmentations["mask"]

       return image, mask

```

2. 检查 get_loaders 函数：

确保 get_loaders 函数在创建 DataLoader 实例时正确地实例化了 CarvanaDataset 类。

```python def get_loaders( train_dir, train_maskdir, val_dir, val_maskdir, train_transform, val_transform, batch_size, num_workers=4, pin_memory=True, ): train_ds = CarvanaDataset( image_dir=train_dir, mask_dir=train_maskdir, transform=train_transform, )

   train_loader = DataLoader(
       train_ds,
       batch_size=batch_size,
       num_workers=num_workers,
       pin_memory=pin_memory,
       shuffle=True,
   )

   val_ds = CarvanaDataset(
       image_dir=val_dir,
       mask_dir=val_maskdir, 
       transform=val_transform,
   )

   val_loader = DataLoader(
       val_ds,
       batch_size=batch_size,
       num_workers=num_workers,
       pin_memory=pin_memory,
       shuffle=False,
   )

   return train_loader, val_loader

```

通过进行这些更改，应该能够解决 "AttributeError: 'CarvanaDataset' object has no attribute 'image'" 错误并继续进行 U-Net 模型的训练。

标签：python,unet-neural-network
From： 78845269

AttributeError：“CarvanaDataset”对象没有属性“image”

相关文章

赞助商

阅读排行