Question

Q1)You are training a U-Net model. In addition to the feature image files, what do you need to provide as an input in the training dataset?

A) A string label indicating the class of the image.

B) A text file for each image, indicating the objects it contains and their bounding boxes.

C) An image file for each class in each image, containing a mask that indicates the pixels belonging to the class.

D) Nothing. You only need the feature images.

Q2)

$loss_history_2.jpg$

Q2) You train a convolutional neural network, tracking the training and validation loss as shown above? What does this loss history indicate?A

A) The input images are the wrong size for the model.

C) The model is not overfitting.

B) The model is overfitting.

D) The batch size is too large.

Note: i need explanation

Answer 1

Answer1) A).A string label indicating the class of the image

It takes images and classes as input.

Answer2) The model is overfitting.

Training loss is the error on the training dataset. Validation loss is the error after running the validation dataset on the trained network.

Validation and training error as the number of epoch increase. But sfter a certain point training error continues to drop but validation error starts increase.