Faster R-CNN usage in PyTorch - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to load a pre-trained Faster R-CNN model from torchvision.

PyTorch

import torchvision.models.detection as detection
model = detection.[1](pretrained=True)

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

Afasterrcnn_resnet50_fpn

Bresnet50

Cmaskrcnn_resnet50_fpn

Dfasterrcnn_vgg16

Attempts:

3 left

💡 Hint

Common Mistakes

Using 'resnet50' which is a classification model, not detection.

Using 'maskrcnn_resnet50_fpn' which is for instance segmentation, not Faster R-CNN.

The function fasterrcnn_resnet50_fpn loads the Faster R-CNN model with a ResNet-50 backbone and Feature Pyramid Network, pretrained on COCO dataset.

2fill in blank

medium

Complete the code to put the Faster R-CNN model in evaluation mode before inference.

PyTorch

model.[1]()

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

Atrain

Bpredict

Cfit

Deval

Attempts:

3 left

💡 Hint

Common Mistakes

Calling 'train()' instead of 'eval()' which keeps the model in training mode.

Trying to call 'predict()' which is not a PyTorch model method.

Calling eval() sets the model to evaluation mode, disabling dropout and batch normalization updates, which is needed before inference.

3fill in blank

hard

Fix the error in the code to move the input image tensor to the same device as the model.

PyTorch

device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
model.to(device)
image = image.[1](device)

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

Adevice

Bcuda

Cto

Dmove

Attempts:

3 left

💡 Hint

Common Mistakes

Using 'cuda' which fails if CUDA is not available.

Using 'device' or 'move' which are not valid tensor methods.

The correct method to move a tensor to a device is to(device). Using cuda() only works for CUDA devices and device() or move() are invalid.

4fill in blank

hard

Fill both blanks to prepare the input image tensor as a list and normalize pixel values between 0 and 1.

PyTorch

image = image.[1](dtype=torch.float32) / 255.0
inputs = [image.[2](0)]

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

Ato

Bunsqueeze

Creshape

Dview

Attempts:

3 left

💡 Hint

Common Mistakes

Using 'reshape' or 'view' which can change data layout incorrectly.

Forgetting to convert to float before dividing by 255.

First, convert the image to float32 with to(dtype=torch.float32) and scale by 255. Then add a batch dimension with unsqueeze(0) to make it a list of tensors for the model.

5fill in blank

hard

Fill all three blanks to run inference and extract predicted boxes and labels from the output.

PyTorch

with torch.no_grad():
    outputs = model([1])

boxes = outputs[0][[2]].cpu().numpy()
labels = outputs[0][[3]].cpu().numpy()

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

Ainputs

B'boxes'

C'labels'

D'scores'

Attempts:

3 left

💡 Hint

Common Mistakes

Passing a tensor instead of a list to the model.

Using wrong keys like 'scores' instead of 'labels'.

Pass the input list inputs to the model. The output is a list of dicts. Extract 'boxes' and 'labels' keys from the first output dict, then move to CPU and convert to numpy arrays.