Mask R-CNN overview in Computer Vision - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to import the Mask R-CNN model from the torchvision library.

Computer Vision

from torchvision.models.detection import [1]

Drag options to blanks, or click blank then click option'

Amaskrcnn_resnet50_fpn

Bfasterrcnn_resnet50_fpn

Cretinanet_resnet50_fpn

Dssd300_vgg16

Attempts:

3 left

💡 Hint

Common Mistakes

Choosing Faster R-CNN import instead of Mask R-CNN.

Confusing RetinaNet or SSD with Mask R-CNN.

The function maskrcnn_resnet50_fpn loads the Mask R-CNN model with a ResNet-50 backbone and Feature Pyramid Network.

2fill in blank

medium

Complete the code to create a Mask R-CNN model pre-trained on COCO dataset.

Computer Vision

model = [1](pretrained=True)

Drag options to blanks, or click blank then click option'

Afasterrcnn_resnet50_fpn

Bssd300_vgg16

Cretinanet_resnet50_fpn

Dmaskrcnn_resnet50_fpn

Attempts:

3 left

💡 Hint

Common Mistakes

Using Faster R-CNN function instead of Mask R-CNN.

Forgetting to set pretrained=True.

Using maskrcnn_resnet50_fpn(pretrained=True) loads the Mask R-CNN model with weights trained on the COCO dataset.

3fill in blank

hard

Fix the error in the code to switch the model to evaluation mode.

Computer Vision

model.[1]()

Drag options to blanks, or click blank then click option'

Atrain

Bfit

Ceval

Dpredict

Attempts:

3 left

💡 Hint

Common Mistakes

Using train() instead of eval().

Using non-existing methods like fit() or predict().

Calling model.eval() sets the model to evaluation mode, disabling dropout and batch normalization updates.

4fill in blank

hard

Fill both blanks to prepare the input image tensor and move it to the device.

Computer Vision

image = image.to([1]).unsqueeze([2])

Drag options to blanks, or click blank then click option'

Adevice

B0

C1

Dcpu

Attempts:

3 left

💡 Hint

Common Mistakes

Using unsqueeze(1) instead of unsqueeze(0).

Moving image to cpu explicitly when device is GPU.

The image tensor is moved to the device (CPU or GPU) with to(device). Then unsqueeze(0) adds a batch dimension at position 0.

5fill in blank

hard

Fill all three blanks to run the model on input and get masks from the output.

Computer Vision

with torch.no_grad():
    output = model([1])
    masks = output[0][[2]].squeeze([3]) > 0.5

Drag options to blanks, or click blank then click option'

A[image]

Bmasks

C1

D0

Attempts:

3 left

💡 Hint

Common Mistakes

Passing image tensor directly instead of a list.

Accessing output with wrong key or index.

Using squeeze(0) instead of squeeze(1).

The model expects a list of images, so input is [image]. The masks are in output[0]['masks']. The squeeze(1) removes the channel dimension before thresholding.