Creating binary masks

Images for segmentation tasks are typically annotated with pixel-level masks. Consider this image of an Egyptian Mau cat.

cat image

In this and the next exercise, you will use the corresponding mask to segment the cat out of the image. First, you will need to load the mask and binarize it.

Image from PIL, transforms from torchvision, and torch have already been imported for you.

This exercise is part of the course

Deep Learning for Images with PyTorch

View Course

Exercise instructions

Load the mask image stored in annotations/Egyptian_Mau_123.png and assign it to mask.
Create a binary_mask from mask_tensor where each pixel equal to 1/255 is assigned a tensor value of 1.0, and the remaining pixels are assigned a tensor value of 0.0.

Hands-on interactive exercise

Have a go at this exercise by completing this sample code.

# Load mask image
mask = ____

# Transform mask to tensor
transform = transforms.Compose([transforms.ToTensor()])
mask_tensor = transform(mask)

# Create binary mask
binary_mask = ____(
    ____, 
    ____,
    ____,
)

# Print unique mask values
print(binary_mask.unique())

Edit and Run Code