School of Computer Science and Engineering

FMTH0301/Rev.5.3
Course Plan

Semester: V CS/AI Year: 2024-25

Course Title: Computer Vision Course Code: 24ECSE317
Total Contact hrs: 58hrs Duration of ESA: 2hrs
ISA Marks:50 ESA Marks:50
Lesson Plan Author: Prof. Lalitha.G.M Date: 08/08/2024
Checked By:Prof. Padmashree Desai Date: 14/08/2024

Brief description of the course: This course encompasses principles and techniques to
interpret and analyze visual data, such as images and videos in the field of artificial
intelligence. It's based on the idea of replicating how humans see and understand the world.

Prerequisites: Programming skills: C++ or Python , Mathematics: Linear Algebra, Calculus,

statistics, Machine Learning and Image Processing.

Course Outcomes (COs):

At the end of the course the student should be able to:

1. Identify the mathematical foundations for basics of computer vision.

2. Analyze principles and techniques of computer vision through features and filtering.

3. Apply the semantic segmentation for clustering and recognition.

4. Develop object detection and optical flow for motion

5. Apply advanced techniques to interpret results.

1
 School of Computer Science and Engineering

Course Articulation Matrix: Mapping of Course Outcomes (COs) with Program

Outcomes (POs)
Course Title: Computer Vision Semester: V
Course Code: 24ECSE317 Year: 2024-25

Course Outcomes (COs) / Program 1 2 3 4 5 6 7 8 9 10 11 12 13 14

Outcomes (POs)

1.Identify the mathematical M M

foundations for basics of computer
vision.

2.Analyze principles and techniques M M

of computer vision through features
and filtering.

3.Apply the semantic segmentation M M

for clustering and recognition.

4.Develop object detection and M M

optical flow for motion.

5.Apply advanced techniques to M M H M

interpret results.

Degree of compliance L: Low M: Medium H: High

2
 School of Computer Science and Engineering

Competency addressed in the Course and corresponding Performance Indicators

Competency Performance Indicators

1.4 Demonstrate competence in computer CSPI-1.4.1 Apply suitable data structures and
science engineering knowledge. programming paradigm to solve problems.
CSPI-1.4.3 Apply suitable database concepts to manage
data.
2.1 Demonstrate an ability to identify and CSPI-2.1.2 Identify processes, modules ,variables, and
characterize an engineering problem parameters of computer based system to solve the
problems.
2.2 Demonstrate an ability to formulate a CSPI-2.2.2 Identify functionalities and computing
solution plan and methodology for an resources.
engineering problem
2.3 Demonstrate an ability to formulate CSPI-2.3.1 Apply computer engineering principles to
and interpret a model. formulate models (mathematical or otherwise) of a
computer-based system or process that is appropriate
in terms of applicability and required accuracy.
5.3 Demonstrate an ability to apply IT CSPI-5.3.1 Demonstrate proficiency in using IT tools for
tools for the chosen engineering activity. performing engineering activity.
10.1 Demonstrate effective use of written CSPI-10.1.2 Write a technical report for software
communication skills relevant to the development life cycle activities using Standards.
engineering discipline that convey
information effectively to both technical
and non-technical stake holders.
10.2 Demonstrate competence in CSPI-10.2.2 Deliver effective oral presentations to
listening, speaking and presentation technical and non-technical audiences.

Eg: 1.2.3: Represents Program Outcome ‘1’, Competency ‘2’, and Performance Indicators ‘3’.

3
 School of Computer Science and Engineering

Course Content

Course Code: 24ECSE317 Course Title: Computer Vision

L-T-P : 2-0-1 Credits:3 Contact Hrs: 4hrs/week.
ISA Marks: 50 ESA Marks: 50 Total Marks: 100
Teaching Hrs: 30 Practical Hrs: 28 Exam Duration: 2hrs

Content Hrs
Unit – 1
Introduction: Computer Vision Overview, Pixels and image representation, 4hrs
1 Filters:
Linear systems, Convolutions and cross-correlations; Lab: Basics, Filters
Features and filtering: Edge detection: Gaussian, Sobel filters, Canny edge 8hrs
detector, Features and fitting: RANSAC Local features, Harris corner
2
detection, Feature descriptors: Difference of gaussians, Scale invariant
feature transform; Lab: Filters, Edges, Features
Semantic segmentation part I: Perceptual grouping, Agglomerative 3hrs
3
clustering, Super pixels and over segmentation;
Unit – 2
Semantic segmentation part II: Clustering: K-means, Mean shift; Visual Bag 3 hrs
3 of Words: Texture features, Visual bag of words; Lab: Resizing, clustering,
recognition.
Motion: Optical Flow, Lucas-Kanade method, Horn-Schunk Method, 6hrs
4 Pyramids for large motion, Tracking: Feature Tracking, Lucas KanadeTomasi
(KLT) tracker; Lab: Object detection, optical flow.
Advanced Techniques: Image stitching, Image pyramids, Object recognition, 6hrs
5
Dimensionality reduction, Face identification, Detecting objects by parts.

Reference Books:
1. Richard Szeliski, Computer Vision: Algorithms and Applications, Springer,
2011.
2. D. Forsyth and J. Ponce, Computer Vision: A Modern Approach, Pearson
Education India, 2nd Ed, 2015.
3. R. I. Hartley and A. Zisserman, Multiple View Geometry in Computer Vision,
Cambridge University Press, 2nd Edition, 2004.
4
 School of Computer Science and Engineering

Evaluation Scheme

In-Semester Assessment Scheme

Assessment Conducted for Weightage in

marks Marks
ISA-1 (Theory) 30
33
ISA-2 (Theory) 30
Lab Activity 30 17
Total 50

End-Semester Assessment Scheme

Assessment Conducted Weightage in

for marks Marks
Theory 60 33
Lab 20 17
Total 50

5
 School of Computer Science and Engineering

Lab Activity plan

Credit: 1 Computer Vision Lab

Preamble:Computer Vision is concerned with the theory and technology for building
artificial system that obtain information from various images or multi-dimensional
data. Information about the environment for robotic systems could be provided by a
computer vision system, acting as a vision sensor and providing high-level information
about the environment. Computer Vision is inverse of computer graphics. Computer
Vision seeks to generate intelligent and useful descriptions of visual scenes and
sequences and of the objects that populate them, by performing operations on the
signals received from video cameras.
Objective: The student should be able to use Computer Vision Tools for handling,
processing, and analyzing advanced vision techniques.
Team size: Group of 4- 5

Type: Each batch will work for one distinct application area
Blooms Timeline PI
Sl. No. Experiments CO Hrs Marks
level wrt COE code
Tool Installation Assignment of
the following application areas
to each batch:
1) Semantic segmentation:
Perceptual grouping,
Agglomerative clustering
2) Motion: Optical Flow, 1st &2nd
1. CO1 L3 1.4.1 4 Nil
Pyramids for large motion. week
3) Advanced Techniques:
Image stitching, Image
pyramids, Object recognition,
Dimensionality reduction, Face
identification, Detecting objects
by parts.
2. Problem Identification CO1 L3 3rdWeek 2.1.2 2 05
Handling basic Features,
3. filtering,segmentation,clusterin L3 4th Week 2.3.1 6 Nil
CO2
g and object detection.
5th & 6th
4. Analysis CO2 L3 1.4.1 4 Nil
Week
th
5. Design CO2 L3 7 & 8 th 2.3.1 4 05
6
 School of Computer Science and Engineering

Week
9th , 10th
6. Implementation CO5 L3 & 11th 5.3.1 6 10
Week
CO5 14th 10.2.2
7. Presentation and Report L3 2 10
Week 10.1.2
Total 28 30

In Semester Assessment (ISA) Rubrics for Practical (P) component

Phases Exemplary Satisfactory Needs Notsatisfactory
( 8-10 ) ( 5-7 ) Improvement ( 0-1 )
(2 - 4)
Problem The problem The problem The problem is The problem
Identificati statement is clear, statement is clear vaguely defined or statement lacks
on concise, and but comprises of lacks significant clarity. It is unclear,
(PI - 2.1.2) specific. It clearly ambiguities. clarity, resulting in
doesn't provide
identifies the difficulties to
context, and doesn't
problem, its context, understand the
and its scope. issue. adequately convey
the issue.
Analysis Basic Features and The majority of The features and The appropriate
(PI - 1.4.3) filtering highly the features and filtering poorly features and
relevant.algorithmsc filtering prepared with filtering are missing
ollected are relevant and major issues in or irrelevant.
systematically and appropriate. But, handling. Inadequate,with
prepared with minor issues exist significant problems
thorough
knowledge.
Design Decision and design Decision and Design is Decision and design
(PI – 2.3.1) recommendations design inappropriate or recommendation
have a strong base recommendation poorly selected. are not relevant.
and justification. are reasonable. The approach lacks
structure.
Implemen Selection of
Selection of Significant technical Needs improvement
tation appropriate appropriate problems or major in the analytical
(PI-5.3.1) analytical tools and
analytical tools functionality issues. tools and/or
techniques. and techniques. The project may software
Developed code thatDeveloped code not work as engineering
that follows the intended or lacks techniques.
follows the design
design core features. The developed code
specification. specification, but is not mapping to
can be further the design
improved. specification.
7
 School of Computer Science and Engineering

The presentation is The presentation There is ambiguity Communication is

Presentati professionally is understandable in the presentation. ineffective. Time
on (PI- delivered. The but may lack The speaker is management may
10.2.2) speaker exhibits engagement. The unconfident. There be inconsistent.
strong confidence. speaker shows is clear time
Demo is delivered some confidence management.
within the allotted but may struggles
time. with articulation.
Documentation is Documentation is Documentation is Documentation is
and comprehensive, ordered and unclear, poorly desperately
Report (PI- well-organized. comprehensive. structured, Key organized, with
10.1.2) could be points don’t carry significant problems
structured better. clarity. with clarity.

Lab End Semester Examination Parameters

Write_Up: Solution for the Problem statement 5
Parameters in brief
Implementation 10
Result Analysis 5
Total Marks 20

End Semester Assessment (ESA) Rubrics for Practical (P) component

Parameters
Scope for
(Evaluated Excellent Competent PI Code
Improvement
for 20 M)
Algorithm Able to write Able to write solution Able to write solution 2.1.2
Writing solution without by considering some with warnings or
(5 M) errors. Incorporated inputs and errors. (0-1M)
all inputs and conditions. (2-3M)
conditions. (4-5M)
Execution Able to execute Able to execute Able to execute 2.1.2
(10 M) solution without solution but solution with
errors for all inputs exceptional warnings or errors (1-
and exceptional conditions are 4M)
conditions. (8-10 M) missing. (5-7M)

8
 School of Computer Science and Engineering

Result Comprehensively Has difficulty in Fails to identify 1.4.3

Analysis (5M) analyzes and applies identifying or appropriate solution
correct solution.(4- characterizing the for given problem.
5M) correct solution for (0-1M)
given problem.

(2-3M)

End Semester Assessment (ESA) Rubrics for Practical (P) component

Course Unitization for ISA and ESA
Topics / Chapters Teaching No. of No. of No. of No. of
Credits Questions in Questions Questions in Questions in
ISA-1 in ISA-2 Activity ESA
Unit I
Computer Vision overview 4 0.5 -- -- 0.5
Features and filtering 8 2 -- -- 2

Semantic segmentation I 3 0.5 -- -- 0.5

Unit II
Semantic segmentation II 3 -- 0.5 -- 0.5

Motion 6 -- 1.5 -- 1.5

Advanced Techniques 6 -- 1 -- 1
Note
1. Each Question carries 15 marks and may also consists of sub-questions.
2. Mixing of sub-questions from different chapters within a Unit I and Unit II is allowed in
ISA-I, ISA-II and ESA
3. Answer 4 full questions of 15 marks each (two full questions from Unit I and Unit II)
out of 6 questions in ESA.

Date: 15.8.2024 Head, SoCSE

9
School of Computer Science and Engineering

10
 School of Computer Science and Engineering

Course Assessment Plan

Course Title: Computer Vision Code: 24ECSE317

Weightage Assessment Methods

Course outcomes (COs) in
assessment ISA-I ISA-II Lab ESA ESA(Lab
Assessment (Theory) )
1.Identify the mathematical
foundations for basics of 10% ✓ ✓ ✓
computer vision.

2.Analyze principles and

techniques of computer 25% ✓ ✓
vision through features and
filtering.

3.Apply the semantic

segmentation for clustering 25% ✓ ✓ ✓ ✓ ✓
and recognition.

4.Develop object detection 20% ✓ ✓ ✓ ✓

and optical flow for motion.

5.Apply advanced
techniques to interpret 20% ✓ ✓ ✓ ✓
results.

Weightage 100% 16% 17% 17% 33% 17%

11
 School of Computer Science and Engineering

Chapterwise Plan

Course Code and Title: 24ECSE317/Computer Vision

Chapter Number and Title: 1. Introduction. Planned Hours:4 hrs

Learning Outcomes:-
At the end of the topic the student should be able to:
Topic Learning Outcomes COs BL CA Code
1. Explain the concept of digital image representation. 1 L2 1.4
2. Discuss linear systems in the context of image processing. 2 L3 2.1
3. Describe convolution and cross-correlation in the context of 2 L3 2.3
image processing.

Lesson Schedule
Class No. - Portion covered per hour
1.Computer Vision Overview,
2. Pixels and image representation,
3. Filters: Linear systems,
4. Convolutions and cross-correlations;

Review Questions
Sl.No. - Questions TLOs BL PI Code
1.Explain the concept of digital image representation. How are 1 L3 1.4.1
images stored digitally? Discuss the role of pixels in representing
images and how pixel intensity contributes to image information.
2.Define linear systems in the context of image processing. Explain 2 L2 2.3.1
the properties of linearity and shift-invariance. How are linear
systems used in filtering operations in computer vision?
3.Illustrate convolution and cross-correlation in the context of 3 L3 2.1.3
image processing. Explain the mathematical formulations of both
operations and their differences

12
 School of Computer Science and Engineering

Course Code and Title: 24ECSE317/Computer Vision

Chapter Number and Title: 2. Features and filtering Planned Hours: 8hrs

Learning Outcomes:-
At the end of the topic the student should be able to:
Topic Learning Outcomes COs BL PI Code
1.Discuss the edge detection in the context of image processing. 2 L2 2.1
2.Explain the Canny edge detection algorithm. 2 L3 2.3
3. Discuss Features and fitting. 2 L2 2.3
4.Explain the Feature descriptors. 2 L3 2.3

Lesson Schedule
Class No. - Portion covered per hour
1. Edge detection: Gaussian,
2. Sobel filters,
3. Canny edge detector,
4. Features and fitting: RANSAC Local features,
5. Features and fitting: RANSAC Local features contd.
6. Harris corner detection,
7. Feature descriptors: Difference of gaussians,
8. Scale invariant feature transform;

13
 School of Computer Science and Engineering

Review Questions
Sl.No. - Questions TLOs BL PI Code
1.Discuss edge detection in the context of image processing. Explain
its significance in computer vision tasks and its applications in image 1 L2 2.1.3
analysis and feature extraction.
2.Explain the Canny edge detection algorithm and its advantages
over other edge detection techniques. Discuss the steps involved in
the Canny edge detection process, including Gaussian smoothing, 2 L3 2.3.1
gradient computation, non-maximum suppression, and edge tracing
by hysteresis.
3. Discuss RANSAC (Random Sample Consensus) in the context of
computer vision. Explain its significance in robust estimation and its 3 L2 2.3.1
applications in tasks such as image stitching and object recognition.
4. Compare and contrast the performances of local feature
descriptors like SIFT and SURF with respect to robustness,
computational efficiency, and effectiveness in matching scenarios. 3 L3 2.3.1
Discuss the advantages and limitations of Harris corner detection
compared to other corner detection techniques.
5.Illustrate the Difference of Gaussians (DoG) method in the context
of computer vision. Explain how DoG is used for feature detection 4 L3 2.3.1
and its significance in scale-space representation.

14
 School of Computer Science and Engineering

Course Code and Title: 24ECSE317/Computer Vision

Chapter Number and Title: 3.Semantic Segmentation-I Planned Hours: 3 hrs

Learning Outcomes:-
At the end of the topic the student should be able to:
Topic Learning Outcomes COs BL PI Code
1. Explain the perceptual grouping algorithms. 3 L2 2.1
2. Discuss semantic segmentation in the context of computer vision. 3 L2 2.3

Lesson Schedule
Class No. - Portion covered per hour
1. Perceptual grouping,
2. Agglomerative clustering,
3. Super pixels and over segmentation;

Review Questions
Sl.No. - Questions TLOs BL PI Code
1.Discuss semantic segmentation in the context of computer vision.
Explain its significance in scene understanding and pixel-level 2.1.3
1 L2
classification tasks. Provide examples of real-world applications
where semantic segmentation is used.
2.Discuss the concept of perceptual grouping in image
segmentation. Explain how perceptual grouping algorithms group
image regions based on visual similarity cues such as color, texture, 1 L3 2.1.3
and proximity. Provide examples of perceptual grouping techniques
and their role in pre-processing for semantic segmentation.

3.Illustrate the clustering in the context of computer vision and its

significance in image segmentation. Explain how clustering 2.3.1
2 L3
algorithms group similar pixels together to form coherent regions in
images.
4.Describe the application of the K-means clustering algorithm in 2 L3 2.1.3
image segmentation. Explain how K-means partitions the image
into clusters based on pixel similarities in color space. Discuss the

15
 School of Computer Science and Engineering

challenges and considerations in using K-means for image

segmentation tasks, such as determining the optimal number of
clusters.
5.Explain the Visual Bag of Words (BoW) approach in the context of
image classification. Explain its significance in representing images 2 L2 2.3.1
using local features and global image descriptors.

16
 School of Computer Science and Engineering

Course Code and Title: 24ECSE317/Computer Vision

Chapter Number and Title: 4.Semantic Segmentation-II Planned Hours:03 hrs

Learning Outcomes:-
At the end of the topic the student should be able to:
Topic Learning Outcomes COs BL PI Code
1.Apply the clustering algorithms. 3 L3 2.3
2.Discuss the Visual Bag of Words (BoW) approach in the context of
3 L3 2.1
image classification

Lesson Schedule
Class No. - Portion covered per hour
1.Clustering: K-means, Mean shift;
2.Visual Bag of Words: Texture features
3.Visual bag of words…contd.

Review Questions
Sl.No. - Questions TLOs BL PI Code
1.Illustrate the clustering in the context of computer vision and its
significance in image segmentation. Explain how clustering 2.3.1
1 L3
algorithms group similar pixels together to form coherent regions in
images.
2.Describe the application of the K-means clustering algorithm in
image segmentation. Explain how K-means partitions the image
into clusters based on pixel similarities in color space. Discuss the 2.1.3
1 L3
challenges and considerations in using K-means for image
segmentation tasks, such as determining the optimal number of
clusters.
3.Explain the Visual Bag of Words (BoW) approach in the context of
image classification. Explain its significance in representing images 2 L2 2.3.1
using local features and global image descriptors.

17
 School of Computer Science and Engineering

Course Code and Title: 24ECSE317/Computer Vision

Chapter Number and Title: 5. Motion Planned Hours: 6 hrs

Learning Outcomes:-
At the end of the topic the student should be able to:

Topic Learning Outcomes COs BL PI Code

1.Discuss optical flow in the context of computer vision. 4 L2 2.1
2. Describe the Lucas-Kanade method and Horn-Schunk method for
4 L2 2.4
optical flow estimation.
3.Discuss feature tracking in the context of computer vision 4 L3 2.4
4. Explain the concept of pyramids in the context of optical flow
4 L3 2.3
estimation for handling large motion.
5. Describe the Lucas-Kanade Tracker (LKT) algorithm for feature
4 L3 2.3
tracking.
Lesson Schedule
Class No. - Portion covered per hour
1. Optical Flow,
2. Lucas-Kanade method,
3. Horn-Schunk Method,
4. Pyramids for large motion,
5. Tracking: Feature Tracking,
6. Lucas KanadeTomasi (KLT) tracker.

18
 School of Computer Science and Engineering

Review Questions
Sl.No. - Questions TLOs BL PI Code
1.Discuss optical flow in the context of computer vision. Explain its
1 L2 2.1.3
significance in estimating the motion of objects in consecutive
frames of a video sequence.
2.Illustrate the Lucas-Kanade method for optical flow estimation.
Explain how it computes the motion vectors of image patches by
solving a system of linear equations using spatial and temporal 2 L3 2.1.3
gradients. Discuss its assumptions and limitations, particularly
regarding small motion and spatial smoothness

3. Discuss the feature tracking in the context of computer vision.

Explain its significance in analyzing the motion of objects across 3 L3 2.3.1
consecutive frames in video sequences.

4. Discuss the advantages and limitations of the Lucas-Kanade

Tracker (LKT) algorithm for feature tracking. Explain scenarios
where LKT performs well, such as tracking rigid objects with small 5 L2 2.1.3
motion, and challenges it may encounter, such as occlusions and
non-rigid deformations.

5.Explain how LKT can be extended or modified to address some of

its limitations. Discuss techniques such as feature selection, multi- 2.3.1
5 L3
scale processing, and robust error metrics that can improve the
performance of LKT in challenging tracking scenarios.

19
 School of Computer Science and Engineering

Course Code and Title: 24ECSE317/Computer Vision

Chapter Number and Title: 6. Advanced Techniques Planned Hours: 6hrs

Learning Outcomes:-
At the end of the topic the student should be able to:
Topic Learning Outcomes COs BL PI Code
1.Discuss image stitching in the context of computer vision 5 L2 2.1
2. Describe the concept of image pyramids and their role in 5 L3 2.4
computer vision tasks.
3. Discuss object recognition in computer vision 5 L3 2.4

4. Explain dimensionality reduction in the context of machine 5 L3 2.3

learning and computer vision.
5. Describe face identification and detecting object by parts in 5 L3 2.3
computer vision.

Lesson Schedule
Class No. - Portion covered per hour
1.Image stitching,
2.Image pyramids ,
3.Object recognition,
4.Dimensionality reduction,
5.Face identification,
6.Detecting objects by parts

20
 School of Computer Science and Engineering

Review Questions
Sl.No. - Questions TLOs BL PI Code
1.Discuss the image stitching in the context of computer vision.
Explain its significance in creating panoramic images by combining 1 L2 2.1.3
multiple overlapping images.
2.Describe the concept of image pyramids and their role in
computer vision tasks. Explain how image pyramids are constructed
by recursively downsampling images to multiple scales and how 2 L3 2.3.1
they are used for tasks such as feature detection, motion
estimation, and image blending.
3.Explain dimensionality reduction in the context of machine
learning and computer vision. Discuss techniques such as principal
component analysis (PCA) and t-distributed stochastic neighbor 3 L3 2.3.1
embedding (t-SNE) used for reducing the dimensionality of high-
dimensional feature spaces while preserving relevant information.
4.Describe face identification in computer vision. Explain its
significance in biometric authentication and surveillance systems.
Discuss techniques such as eigenfaces, local binary patterns (LBP), 4 L2 2.4.1
and deep learning-based approaches used for face identification
tasks.

21
 School of Computer Science and Engineering

Model Question Paper for Minor Examination (ISA-I)

Course Code: 24ECSE317 Course Title: Computer Vision
Duration: 60 mins Max. Marks: 30 (Scaled to 16)
Note: 1. Answer any two full questions 2. Each full question carries equal marks.
Q.No Questions PI
Marks CO BL PO
Code
1.a Explain the linear systems in the context of
8 CO1 L2 2 2.3.1
image processing.
b. Illustrate the properties of linearity and shift-
7 CO1 L3 2 2.3.1
invariance.
2.a Explain the purpose and application of each filter
type, such as Gaussian blur, Sobel edge 8 CO2 L3 2 2.3.1
detection.
b. Describe the concept of local feature descriptors
in computer vision. Explain their role in 7 CO2 L3 2 2.3.1
capturing distinctive information from image
patches.
3.a Discuss the advantages and limitations of cross- 8 CO1 L3 2 2.1.3
correlation-based methods in computer vision.
b. Compare and contrast the performances of
7
Gaussian, Sobel, and Canny edge detection CO2 L3 1 1.3.1
techniques.

22
 School of Computer Science and Engineering

Model Question Paper for Minor-II Examination (ISA-II)

Course Code:24ECSE317 Course Title: Computer Vision
Duration: 60mins Max. Marks: 30 (Scaled to 17)
Note: 1. Answer any two full questions 2. Each full question carries equal marks.
Q.No Questions PI
Marks CO BL PO
Code
1.a Discuss the strengths and limitations of the K-means
algorithm for image segmentation. Explain scenarios
where K-means performs well, such as segmenting 8 CO3 L2 2 2.1.3
images with distinct color distributions, and challenges
it may face, such as sensitivity to noise and initialization.
b. Describe the agglomerative clustering algorithm.
Discuss the parameters involved in agglomerative 7 CO3 L3 2 2.4.1
clustering and their impact on the segmentation results.
2.a Explain the concept of superpixels and their role in
image segmentation. Discuss how superpixels reduce
the computational complexity of segmentation 8 CO3 L3 2
2.3.1
algorithms by representing pixels as compact and
homogeneous regions..
b. Describe the key steps involved in the Visual Bag of
Words pipeline for image classification. Include in detail
the following:
 Feature extraction: Extract local features (e.g., SIFT,
7 CO4 L3 2 2.1.3
SURF) from image patches.
 Feature quantization: Quantize local features into
visual words using clustering algorithms like K-
means.
3.a Describe the Lucas-Kanade Tracker (LKT) algorithm for
feature tracking. Explain how LKT estimates the motion
of feature points by minimizing the error between the 8 CO3 L3 2 2.3.1
intensity values of corresponding pixels in consecutive
frames.
b. Describe face identification in computer vision. Explain 7 CO4 L3 2 2.4.1
its significance in biometric authentication and
23
 School of Computer Science and Engineering

surveillance systems.

Model Question Paper for End Semester Assessment (ESA)

Course Title: Computer Vision Course Code: 24ECSE317
Duration 120 Minutes Max. Marks:60
Note: Answer any 2 full questions from each UNIT I and Unit II
UNIT – I
Q.No Questions Marks CO PI Code BL
1a How does image resolution impact image quality and
processing tasks? Provide examples of scenarios
8 CO1 1.4.1 L2
where high and low resolutions are preferred.

b Compare and contrast RGB, HSV, and grayscale

7 CO1 1.4.1 L3
representations.
2a Explain the HSV significance in computer vision
applications. 8 CO4 1.4.3 L3

b Define linear systems in the context of image

processing. Explain the properties of linearity and 7 CO2 2.3.1 L2
shift-invariance.
3a Explain the Canny edge detection algorithm and its
8 CO2 2.2.2 L3
advantages over other edge detection techniques.
b Illustrate the semantic segmentation in the context of
7 CO4 2.3.1 L3
computer vision.
UNIT – II
4a Explain semantic segmentation significance in scene 8 CO4 2.3.1 L3
understanding and pixel-level classification tasks.
b Explain how agglomerative clustering merges similar 7 CO4 1.4.3 L3
image regions iteratively to form larger clusters.
5a Discuss the challenges posed by over segmentation 8 CO4 1.4.3 L3
and its impact on downstream tasks such as object
recognition and scene understanding.
b Describe the application of the K-means clustering 7 CO3 2.3.1 L3
algorithm in image segmentation.
Illustrate dimensionality reduction in the context of 8 CO5 2.2.2 L3
6a machine learning and computer vision.
24
 School of Computer Science and Engineering

b Provide examples of object recognition applications, 7 CO5 1.4.3 L2

such as autonomous driving and surveillance.

25