COMP/ENGN 6528 Mid-Semester test S1 2022








The Australian National University College of Engineering and Computer Science Mid-semester test, First Semester 2022

COMP/ENGN6528 Computer Vision

Question Booklet

Reading time: 15 minutes
Writing time: 1 hour
Uploading time: 15 minutes




















Instructions on next page



Allotted Time

You will have 1 hour to complete the exam plus 15 minutes of reading time (you are allowed to write
during this time. An additional 15 minutes has also been allowed to accommodate the additional task
of uploading your completed exam to the final exam Turnitin submission portal on the
COMP/ENGN6528 Wattle site. Thus you have 1 hour and 30 minutes to complete the exam. NO late
exams and submissions will be accepted. You may begin the exam as soon as you download it.
Minimal requirements:

You may attempt all questions

You SHOULD NOT include an assignment cover sheet

You must type your ANU student identification number at the top of the first page of your submission

You must monitor your own time (i.e. there is no invigilator to tell you how many minutes are left).

Your answers must be clear enough that another person can read, understand and mark your
answer. 11 or 12 point font with 1.5 spacing is preferred. Scanned images of handwritten equations
or diagrams must be legible and of a suitable size. Please be aware that your submitted document
should have at least 20 words based on the requirement of Turnitin. If it is not satisfied, it may lead to
unsuccessful submission.

Numbering questions
● You must specify the question you are answering by typing the relevant question number at
the top the page
● Each question should begin on a new page
● Multi-part questions (e.g. question 1 parts a and b) may be addressed on the same page but
should be clearly labelled (e.g. 1a, 1b )
● Questions should be answered in order

You must upload your completed answers in a single document file within the allotted time using a
compatible file type for Turnitin (Preference: MS Word’s .doc or .docx or .pdf format) It is the
student’s responsibility to check that the file has uploaded correctly within Turnitin. No late
submission will be accepted. Access to the Turnitin practise site can be found here:
https://www.anu.edu.au/students/academic-skills/academic-integrity/turnitin

Academic integrity

Students are reminded of the declaration that they agree to when submitting this exam paper via
Turnitin:
I declare that this work:
● upholds the principles of academic integrity as defined in the University Academic
Misconduct Rules;
● is original, except where collaboration (for example group work) has been authorised in
writing by the course convener in the course outline and/or Wattle site;
● is produced for the purposes of this assessment task and has not been submitted for
assessment in any other context, except where authorised in writing by the course convener;
● gives appropriate acknowledgement of the ideas, scholarship and intellectual property of
others insofar as these have been used;
● in no part involves copying, cheating, collusion, fabrication, plagiarism or recycling.



Mid-Semester Exam: How to
Do's and Don't's:
• You are recommended to record your exam process and keep the video to yourself
(for at least 4 weeks).
• Do not submit your video recording. This file stays private to you, unless we have
a specific reason to request this file from you.
• Feel free to ask question directly posting to "Instructors" on Piazza (public posting will be
disabled for you during the exam) or send email to course convenors. Our class link on
Piazza is: http://piazza.com/anu.edu.au/spring2022/engn6528
• Do not use any communication system (other than direct posts on Piazza to Instructors or
sending emails to your course convenor) during your exam - if you do, it will be counted
instantly as collusion and will have serious academic honesty consequences.
• Do not upload any material anywhere (other than to the Wattle upload link at the end of the
exam). If you do so, you will also become part of an academic collusion case which will stay
on your permanent record at the ANU.
• Be very careful searching for any material on-line. If you find yourself seeing references to
material which might be the result of collusion (which hopefully will not exist), you are one
click away from becoming part of a serious academic honesty case yourself. Remember that
all of your activity must appear in your recording, and according to basic academic standards,
we also expect you to reference in your pdf file anything which you might have included in
your working. You will likely waste valuable time for your exam and expose yourself to serious
risks, so we recommend to defer from doing so.
Step-by-step guide on how to sit your actual on-line mid-semester exam:
1. Find yourself a cozy spot and power down all communication channels, besides this forum
here.
2. Start your full screen recording now.
3. Download the exam paper from wattle (link will become active at the time of the exam).
4. Open the exam in the pdf reader, which you tested before.
5. Take a moment to read the whole document. You don’t need to spend exactly 15 minutes for
reading and you can start working on the exam after you finish reading it at your pace.
6. Fill in your answers. You can initially write your answer using your favourite editor. You are
strongly recommended to submit a word document or Pdf document. Whichever way you
chose, all your activities must appear on your recording.
7. Don't forget to save regularly, or use a system which does that for you.
8. When you are complete, upload a single file (word or PDF are OK), and upload via the Wattle
Link. You can make multiple uploads, they will overwrite the last. Ensure you upload before
the end of the time. We have allowed 15 minutes upload time so submit with some time to
spare.
9. Stop your screen recording and make sure the video file is saved (keep this file for at least
4 weeks). Do not submit your video recording.






COMP-ENGN-6528-Computer Vision-Midterm Exam
There are three question sessions
Q1-Q3.
Please name your submission as
COMP/ENGN6528 midsem u1234567.docx
(or PDF as you prefer)
1
Q1.(26 points) Multiple Choices 26 Marks. There is at least one correct answer for each
question. You can get the full mark only if all correct answers are selected. No mark will be
granted for partially selected answers. (2 marks each)
1. (2 points) Which filter(s) could be used for removing salt-and-pepper noise?
A. Gaussian Filter
B. Bilaterial Filter
C. Median Filter
D. Sobel Filter
2. (2 points) Please select the linear filter(s).
A. Gaussian filter
B. Median filter
C. Bilateral filter
D. Sobel filter
3. (2 points) Suppose a camera has a calibration matrix follows:
0@1.5 0 00 1.5 0
0 0 1
1A
What does this tell you about the camera?
A. This is not a valid camera matrix.
B. The focal length is 1.5.
C. The camera has been rotated.
D. The camera has been translated.
E. None of the above.
4. (2 points) Which of the following are advantages of the HSV colour-space with respect to the RGB
colour-space ?
A. Brightness information is separated into a single channel.
B. It uses less memory to encode each pixel.
C. Its fast to compute, so good for compression.
D. The Hue and Saturation components of HSV are less sensitive to lighting variation.
E. It is a perceptually uniform space - distances in the colour space itself correspond to di↵erences
in human perception of colour.
5. (2 points) Which of the following is a (are) good property(ies) of an interest point?
A. Strong 2D structure
B. Two eigenvalues of a local auto-correleation matrix (M) are both large.
C. Edges make good interest points.
D. The derivatives are small.
E. There are bright colours.
6. (2 points) Which of the following are not advantages of the SIFT feature over the Harris corner?
A. Response is invariant to image rotation
B. Response is invariant to image scaling
C. Response in invariant to image image translation
Page 2
D. Less computation required for computation of a SIFT feature.
E. SIFT provides distinctive features for matching across images.
7. (2 points) Which of the following are properties of a Gaussian Filter?
A. It is separable.
B. Pixel weight is scaled according to the nearness of the pixel to the centre of the filter.
C. The only parameter is the sigma of the Gaussian.
D. Applying it always increase the overall image brightness.
E. Its application sharpens edges in the image.
F. Cascading the filter with size = a followed by = b has the same e↵ect as applying a single
filter with size = (a+ b)
8. (2 points) Suppose you have a simple fully connected linear network, with bias, that takes a 32x32
greyscale image as input. The input is connected directly to a classification space with 10 output units.
How many free parameters are there in the network?
A. 10240
B. 10250
C. 10260
D. 320
E. 330
F. 340
9. (2 points) Suppose we have a softmax function at the end of our network that has 4 outputs correspond-
ing to 4 classes. Which of the following are possible outputs from the function?
A. 0.1 1 0.3 0.7
B. 0.3 0.4 0.2 0.1
C. 1 0 1 0
D. 0.4 0.1 0.4 0.1
E. 0.5 0 1.5 0
F. 0.5 0 -0.5 0
G. 0.5 0 0.7 -0.2
H. None of the above
10. (2 points) Which loss function is the best suited to classification problems? Note that y is the estimated
value, y? defines the ground truth. (Please select multiple answers if you consider them to be equally
good).
A. 12 (y y⇤)2
B. y ⇤ log(y) + (1 y⇤)log(1 y)
C. |(y y⇤)|
D. None of these
11. (2 points) Suppose we have a 7x7 convolutional layer, followed by a ReLU, then sequentially another
two 3x3 convolutional layers, each of them is followed by a ReLU. Note that the stride is equal to 1
for all convolutional layers. What is the e↵ective receptive field of a node after the third convolutional
layer?
A. 7x7
Page 3
B. 9x9
C. 10x10
D. 3x3
E. 11x11
F. None of the above.
12. (2 points) Suppose a network has an input image of 240x240. A convolution layer of 5x5 with no
padding, and a stride of two is applied. This is followed by a ReLU, and then a pooling layer of 2x2,
stride 2. What is the size of the output of the pooling layer?
A. 60x60
B. 59x59
C. 118x118
D. 240x240
E. 120x120
F. None of the above.
13. (2 points) Which of the following does regularization in Neural Networks help with?
A. keeping the weights small.
B. preventing overfitting.
C. making some of the weights zero.
D. None of the above.
Page 4
Q2. (34 points) Short Answer Questions. (No answer should be more than half a page.)
1. (3 points) Please provide one example each for a point operation and a neighborhood operation.
2. (5 points) Please describe the di↵erences between Gaussian filter and Bilateral filter. What are the
advantages and disadvantages of the bilateral filter?
3. (3 points) How does SIFT extract scale invariant features?
4. (3 points) What are the advantages of a separable filter?
5. (3 points) What are the advantages and disadvantages of forward warping and inverse warping for image
transformation?
6. (5 points) Why is Harris corner detector rotation invariant?
7. (5 points) Please discuss whether bilateral filtering can remove salt and pepper noise. Please provide
the reasoning.
8. (7 points) Compare the activation function shown in Fig. 1 with a Relu function. Discuss the di↵erences
and the behavior of this activation function when training a network.
Figure 1: Activation Function Comparison
Page 5
Figure 2: Image filter (a) and Image patch (b)
Q3. (40 points) Calculation questions. Please only provide the final answer.
1. (5 points) Given an input 5x5 image:
0BBBB@
233 22 9 4 7
210 33 24 12 2
109 2 6 3 2
33 4 2 6 7
21 5 6 6 3
1CCCCA,
suppose a Deep CNN uses a filter (correlation) that it has learned:
240.8 0.4 0.40.3 0.2 0.1
0.2 0.7 0.6
35 to this image patch.
Then it is followed by a ReLU activation function. What will be the result for the centre pixel of this
image?
2. (5 points) Suppose that Fig. 2 (b) is an image, and (a) is a filter.
Calculate the result if you were to apply the filter as a convolution operator for the red (upper left) and
the blue square (bottom right) (Note: you are going to obtain one value for each region).
3. (15 points) Calculate the value of the blue patch in Fig. 3 using bilateral filtering. Assume the Domain
kernel is of size 5⇥5, the standard deviation d = 3, provided as
0BBBB@
0.0318 0.0375 0.0397 0.0375 0.0318
0.0375 0.0443 0.0469 0.0443 0.0375
0.0397 0.0469 0.0495 0.0469 0.0397
0.0375 0.0443 0.0469 0.0443 0.0375
0.0318 0.0375 0.0397 0.0375 0.0318
1CCCCA
the Range kernel is of size 5⇥ 5 and the standard deviation r = 50. Please 1) provide the range filter
associated to the pixel high-lighted in the Fig. 3, and 2) show the filtered value for the high-lighted pixel.
4. (15 points) Back propagation through the computational graph (see Fig. 4). The current value for
w0 = 0.2, w1 = 0.2, w2 = 0.3, x0 = 2, x1 = 3. p and q define the intermediate variables that are
calculated during training, at the specified points in the computation graph. l is the output of the
computational graph. Please provide the gradient @l@p and
@l
@q based on the back-propagated gradient
calculation.
Page 6
Figure 3: Image Patch
Figure 4: Computational Graph. ⇤ is the multiplication operator, + is addition operator, exp denotes the
exponential operator, ’+1’ defines the sum of the input with 1 and ’-1’ defines the subtraction of 1 from
input, 1x defines the reciprocal operator.
————-This is the end of the exam. ————–
Page 7