NPTEL Deep Learning For Computer Vision Assignment 3 Answers 2022

1 point
For this question, please see Question 1 in the iPython notebook(.ipynb file) provided alongside. Complete your implementation under the “YOUR CODE STARTS HERE” segment therein. Apply the following chain of transformations on the generated coordinates:1. Rotation by 135 degrees.
2. Translation of x coordinate by -2 units and y coordinate by 2 units.
3. Scaling by 2 units.
4. Rotation by 135 degrees.

Ans – D

The following figure represents 5 different set of points(each in different color) plotted on a cartesian plane.
One of them is the correct output that will be produced when we apply the transformations mentioned above to the given input.
Which of the following is the correct output which is produced after applying these transformations?
A
B
C
D
E

Ans -C

1 point
For this question, please see Question 2 in the iPython notebook(.ipynb file) provided alongside. Complete your implementation under the “YOUR CODE STARTS HERE” segment therein. What is the floor value of the mean?
15
5
17
9

Ans – D
1 point
For this question, please see Question 3 in the iPython notebook(.ipynb file) provided alongside. Complete your implementation under the “YOUR CODE STARTS HERE” segment therein. Follow the below mentioned steps:Step 1. Given a folder of images, use SIFT to extract their feature descriptors.
Step 2. Perform k-means clustering on the extracted descriptors.
Step 3. Create histograms by taking help of the clusters calculated in the previous step.
Step 4. Estimate 5 nearest neighbours to a given image and add the distances of those 5 neighbours.

What is the range of the sum as calculated in step 4?
0-200
200-400
400-600
600-800

Ans – C

1 point
Match the following:1) Dense registration through optical flow i) Assumption of rigid transformations
2) Wide baseline spatial matchingof same object ii) Detect different instances
3) Rotation in x − y plane iii) Two degrees of freedom
4) Bag-of-words iv) Aperture problem
v) One degree of freedom
1→iv, 2→i, 3→iii, 4→ii
1→iv, 2→i, 3→v, 4→ii
1→v, 2→ii, 3→iv, 4→i
1→v, 2→i, 3→iii, 4→ii

Ans – B
1 point
Consider the two images A and B below. In A, a dog is present on the left part of the image and in B, the same dog appears on the right part of the image. These two images have ________ bag-of-words representations because bag-of-words representation ________ .same, preserves spatial information
different, preserves spatial information
same, does not preserve spatial information
different, does not preserve spatial information

Ans A

1 point
Given a set of n images represented by a matrix Z∈Rk×n and a query image q, which one of the following quantities gives the index of best matched image from the pool of n images to the query image q? Here, k is the dimensionality of image representations (e.g. bag-of-words).arg min {ZTq}

arg max {qTZ}

arg min {qTZ}

arg max {ZTq}

Ans – D
1 point
Match the following:

Ans – Not Sure