5. Facial Expression Recognition with Keras

Facial Expression Recognition with Keras

by Coursera Project Network

Graded Quiz Answers :

1.Question 1

In the Facial Expression Recognition project that we just completed, we used the Keras API with TensorFlow as its backend. True or False?

1 / 1 point

True

False

2.Question 2

Which of following is true about the FER 2013 dataset of images we used to train the CNN model? (Select all that apply)

1 / 1 point

The dataset contains RGB images of faces

The dataset contains grayscale images of faces

Each image has 48 rows and 48 columns

The dataset contains seven categories of facial expressions

3.Question 3

The Facial Expression Recognition model we created is comprised of 6 dense, fully connected layers. True or False?

1 / 1 point

True

False

4.Question 4

What technology, API, or framework did we use for face detection?

1 / 1 point

Keras

OpenCV

Flask

5.Question 5

What are the two types of regularization techniques that you used in this project? (Select all that apply)

1 / 1 point

Weight regularization

Dropout regularization

Batch normalization

6.Question 6

Select all statements that are true:

1 / 1 point

In order to get a predicted class, we used the np.argmax() function on the predictions. This is because the predictions are 7 dimensional vectors and we want to use the class with the highest probability as our final prediction.

The argmax function displays the image along with its prediction.

The probability scores of a prediction sum up to a total of 1.0

7.Question 7

What does the following code block accomplish?

1
2
3
model_json = model.to_json()
with open("model.json", "w") as json_file:
    json_file.write(model_json)

1 / 1 point

The code saves the trained model's weights to disk as a JSON file.

The code uses a JSON string to store the "config" or model architecture.

8.Question 8

In the hands-on project, which statement(s) would apply to the softmax activation function? (Select all that apply)

1 / 1 point

It can be used as an activation function for the output layer in classification problems.

It gives us linear output for input values higher than 0 and for input values less than 0, the output is set to 0.

It gives us probability scores for the classes.

9.Question 9

Which function returns the history object which has training metrics like training accuracy and loss?

1 / 1 point

model.fit()

model.compile()

10.Question 10

In the hands-on project, the trained model recognized facial expressions in videos saved to disk. Is it true that you need to change just one line of code to classify facial expressions in real-time from your webcam stream?

1 / 1 point

True

False