Not able to plot an image from a Pytorch Dataset

vision
1

Hi!
I am trying to extract an image and plot it using matplotlib.pyplot.imshow() but I am not able to do so. Here is my Dataset class:

class MelanomaDataset(Dataset):
    def __init__(self, df, meta_features=None, transforms=None):
        self.df = df.reset_index(drop=True)
        self.meta_features = meta_features
        self.transforms = transforms

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

    def __getitem__(self, index):
        img_name = self.df.iloc[index].image_name
        img_path = os.path.join('/kaggle/input/siim-isic-melanoma-classification/jpeg/train', img_name + '.jpg')

        image = cv2.imread(img_path)
        image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)

        if self.transforms is not None:
            res = self.transforms(image=image)
            image = res['image'].astype(np.float32)
        else:
            image = image.astype(np.float32)

        image = image.transpose(2, 0, 1)

        if self.meta_features is not None:
            img = torch.tensor(image).float()
            meta_feats = torch.tensor(
                self.df.iloc[index][self.meta_features]
            ).float()
        else:
            img = torch.tensor(image).float()
            meta_feats = None
        return {
            'image': img,
            'meta_features': meta_feats,
            'target': torch.tensor(self.df.iloc[index].target).long()
        }

And this is how I am trying to plot a sample from the dataset:

plt.imshow(dataset[1]["image"].permute(1, 2, 0))

It would be great if someone could help out with this problem.
Thanks!

2

What kind of error are you seeing?

3

I’m not getting an error. Just the figure that is being printed is white, so the img is not being printed. Here is a screenshot:

Screenshot 2023-01-15 at 1.56.26 AM
Screenshot 2023-01-15 at 1.56.26 AM788×530 20.1 KB

4

I guess you are ignoring a warning explaining you might be clipping the image?

Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).
5

I am not getting that warning I just double checked.

6

In that case I wouldn’t know what’s causing the issue as your code works fine after removing undefined parts and casing the output back to uint8:

class MelanomaDataset(Dataset):
    def __init__(self):
        self.paths = ["image.jpeg"]

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

    def __getitem__(self, index):
        img_path = self.paths[index]
        
        image = cv2.imread(img_path)
        image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)

        image = image.transpose(2, 0, 1)
        image = torch.from_numpy(image).float()

        return {
            'image': image,
        }
    
dataset = MelanomaDataset()
data = dataset[0]
img = data["image"]
plt.imshow(img.permute(1, 2, 0).numpy().astype(np.uint8))
7

That’s strange. Thanks a lot tho :slight_smile:

8

Could you double check the dtype and value range of your data["image"] output and check the warning I’ve posted?
Make sure that floating point numbers are in [0, 1] and ints in [0, 255].