Box regressor block

Your final task is to create a regressor block to predict bounding box coordinates. You decide to have a block with 2 fully connected layers with a ReLU activation in between, similar to the classifier you defined earlier.

Your vgg_model and input_dim are still available and torch and torchvision.models have been imported.

This exercise is part of the course

Deep Learning for Images with PyTorch

View Course

Exercise instructions

Create a variable num_coordinates with the number of boundary box coordinates to predict.
Define the appropriate input dimension for the first linear layer and set the output dimension to 32.
Define the appropriate output dimension in the regressor's last layer.

Hands-on interactive exercise

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

# Define the number of coordinates
____

bb = nn.Sequential(  
	# Add input and output dimensions
	nn.Linear(____, ____),
	nn.ReLU(),
	# Add the output for the last regression layer
	nn.Linear(32, ____),
)

Edit and Run Code