Question Bank

Programme B.Tech.: Artificial Intelligence and Data Science

Course Image and Video Analytics
Course Code CCS349 Semester 07 Regulation AU R21

Unit I – INTRODUCTION

Part – A

Sl.
Question Blooms Level
No.

1 Define computer vision. L1

2 Why is computer vision difficult? L1

Differentiate low-level image processing and high-level image
3 L1
processing?
4 Define sampling. L1

5 What is Quantization? L1

6 Define pixel and voxel. L1

7 What are the types of noise in images? L1

8 Define Palette images. L1

9 Recall the term pseudo color. L1

Distinguish between M-pyramids (matrix-pyramids) and T-pyramids (tree-
10 L1
pyramids)?

Part – B

Sl.
Question Blooms Level
No.
Classify the computer vision after explaining the basic concepts of the
1 L2
same in detail.
Illustrate Image representation and image analysis tasks with neat
2 L2
diagrams.
Explain in detail about
i. image representation
3 L2
ii. Continuous image function
iii. Image digitization and sampling
4 Infer the Metric and topological properties of digital images. L2

5 Illustrate the algorithm for distance transform in an image. L2

6 Demonstrate how histogram and entropy is used in image processing. L2

7 Explain in detail about various Levels of image data representation. L2

Unit II – IMAGE PRE-PROCESSING

Page 1 of 5
 Course Code & Name CCS349- Image and Video Analytics

Part – A

Sl.
Question Blooms Level
No.

1 What are the two groups in local pre-processing? L1

2 What is meant by Convolution Mask? L1

3 Define Central Limit Theorem. L1

4 Tell the types of Edge Profile. L1

5 Outline the objective of Image Sharpening? L2

6 Differentiate Roberts and Laplace Operators. L2

7 Classify the Canny Edge Detection criteria. L2

8 Recall the formula for Bi-Cubic Facet Model? L1

9 List out some Examples of Spatial Filtering. L2

10 Outline the High pass Filter used in Homomorphic Filtering. L2

Part – B

Sl.
Question Blooms Level
No.

1 Illustrate the Median Filtering Algorithm with a neat flow. L2

Explain the following operators in detail.
i. Roberts operator
2 ii. Laplace operator L2
iii. Prewitt operator
iv.Sobel operator
3 Illustrate the concept of Zero-crossings of the second derivative. L2
How can you use the Canny Edge Detection algorithm to improve edge
4 L2
detection in a complex grayscale image?
Show the process of implementing parametric edge models in computer
vision applications. Discuss how these models differ from non-
5 L2
parametric approaches and provide examples of scenarios where
parametric edge models are advantageous.
Explain the mechanisms involved in line detection using local pre-
6 processing operators, and discuss their advantages and limitations L2
compared to other line detection methods.
Compare Inverse Filtering and Wiener Filtering in the context of image
restoration for the underlying principles of each method, their
7 advantages, limitations, and suitable scenarios for their application with L2
examples to illustrate the effectiveness of both techniques in restoring
degraded images.

Unit III – OBJECT DETECTION USING MACHINE LEARNING

Page 2 of 5
Course Code & Name CCS349- Image and Video Analytics

Part – A

Sl.
Question Blooms Level
No.

1 Define Object Detection. L1

Compare Object classification with object localization and object
2 L2
detection.
3 How are bounding boxes important for object detection? L1

4 What is IoU? L1

5 What are the Deep Learning architectures used for Object Detection? L1

6 What is Region-based CNN (R-CNN)? L1

7 Compare RCNN with faster RCNN. L2

8 What is the loss function in YOLO? L1

9 What ii the advantage of using two-stage methods? L1

10 Why do we use data augmentation? L1

Part – B

Sl.
Question Blooms Level
No.
What are the fundamentals of object detection, and how do various
1 L2
object detection methods differ in their approaches?
Illustrate a deep learning framework for object detection to identify
2 L2
objects in images.
What is the bounding box approach in object detection? Explain how it
works and its significance in computer vision tasks with examples to
3 L2
illustrate how bounding boxes are used to localize objects within
images.
What is Intersection over Union (IoU) in object detection? Explain its
4 calculation and importance in assessing bounding box accuracy, L2
with examples.
Apply Faster R-CNN for object detection in Python. Describe its steps
5 L2
and evaluate its performance across various images.
What are the salient features of You Only Look Once (YOLO) in object
detection, and how do they contribute to its effectiveness? Explain the
6 concept of loss functions in YOLO and their role in training the model. L2
Provide examples to illustrate how YOLO handles
object detection tasks efficiently.
What are the main parts of YOLO (You Only Look Once) architectures
for object detection, and how do they work together? Explain how
7 L2
YOLO's single neural network approach simplifies object detection. Give
an example to illustrate its effectiveness.
Discuss about Deep learning methods and its approaches:
i. Sliding window approach
8 L2
ii. Bounding box approach
iii. Intersection over Union (IoU)

Page 3 of 5
 Course Code & Name CCS349- Image and Video Analytics

Unit IV – FACE RECOGNITION AND GESTURE RECOGNITION

Part – A

Sl.
Question Blooms Level
No.

1 What is face detection? L1

2 What is face recognition? L1

List the deep learning algorithms for face recognition.
3 L1
DeepFace,VGGFace.
4 What are the various processes in face recognition systems? L1

5 List out the various applications of face recognition. L1

6 Brief the concept of Facenet for face recognition. L1

7 What is the concept of triplet loss? L1

8 Brief about the concept of DeepFace solution by Facebook. L1

9 What is gesture recognition? L1

10 What are the various use cases of gesture recognition? L1

Part – B

Sl.
Question Blooms Level
No.
How does a computer know whose face it's looking at when using face L2
1
recognition technology?
Explain face recognition technology utilized in enhancing security L2
2
measures within public transportation systems?
Illustrate facial recognition technology which enable smartphones to L2
3
unlock only when the owner's face is detected?
How does Facebook's DeepFace technology help identify people in L2
4
photos by recognizing unique features in their faces?

5 Elaborate the FaceNet technique for Face Recognition. L2

6 Explain in detail about gesture recognition. L2

Unit V – VIDEO ANALYTICS

Part – A

Sl.
Question Blooms Level
No.

1 Give 4 use cases of video analytics. L1

2 Define video processing. L1

3 What is vanishing gradient and exploding gradient problem? L1

Page 4 of 5
Course Code & Name CCS349- Image and Video Analytics

Sl.
Question Blooms Level
No.

4 Define Residual Networks and their role in video analytics. L1

5 How does ResNet solve the vanishing gradient problem? L1

6 Define identity mapping L1

7 List the improvements in inception v2. L1

What is the improvement in Inception v3 when compared to Inception v1
8 L1
networks?
9 What is inception v3. L1

10 What is the use of GoogleNet? L1

Part – B

Sl.
Question Blooms Level
No.

1 Outline the basic steps of video processing. L2

Explain with examples of how video analytics is used across industries
2 like retail, healthcare, and transportation, and explain its impact on L2
operations and decision-making?
Illustrate ResNet architecture applied to enhance image classification
3 L2
tasks with real-world examples of its effectiveness?
Infer skip connections in the ResNet architecture contribute to improving
4 the performance of deep neural networks, and give examples of how L2
this technique is applied in practical image recognition tasks?
How does Inception v2 improve performance in tasks like image
5 L2
classification or object detection.
What are the fundamental design principles underlying the GoogleNet
6 architecture, and how do they enable efficient processing of information L2
in deep neural networks?

Page 5 of 5