1

CER
TIFI
CAT
E
FORVision has been completed by
This is to certify that the project work titled
Computer
Avyakt Malhotra of class 10 C under my
th

AI
guidance.

PRO
This project fulfills the term put forth by the

JECT
CBSE in terms of content, presentation, and
explanation.

(Mrs. Nakashi Bajaj)

3

Gui
dan
With a deep sense of gratitude, I acknowledge the
ce
invaluable guidance, motivation, and constructive
criticism rendered to me by my AI teacher.

Mrs.Nakashi Bajaj

Her constant inspiration made the completion of

this project titled Computer Science possible.

Name: Avyakt Malhotra

Class & Section: 10th C
Roll no: 06
 4

INTRODUCTION
◊What
is
Computer Vision is a field of Artificial Intelligence
comp
(AI) that enables computers and systems to
uter
interpret and understand visual data from the
world, such as images or videos. The primary goal
ofvision
computer vision is to replicate and enhance
human visual perception through machines. In
? words, it's the technology that allows
other
computers to "see" and interpret visual data in a
meaningful and useful way, similar to how humans
perceive and understand their surroundings.

Unlike traditional software, which relies solely on

numerical data, computer vision systems analyze
pixel-based information from images and videos,
transforming this raw data into structured insights,
predictions, or actions. As technology advances,
computer vision has become integral to many
applications, ranging from autonomous vehicles
and medical diagnostics to social media filters
and augmented reality (AR).
5

◊ Core & Concept of Computer Vision

To understand computer vision, it's helpful to break it
down into some core tasks that define the field. These
tasks are used to build systems that can perform visual
recognition, analysis, and decision-making:
1. Image Classification
Image classification is the task of categorizing an image
into a predefined category or class. For example, an
image might be classified as either "cat" or "dog." A
machine learning model trained on large datasets of
images can learn to recognize patterns and classify new
images accordingly. This process involves detecting
important features of an image, like colour, texture, and
shape, and mapping those features to known categories.
 Use Case Example: Automatically identifying
whether a picture contains a dog, cat, or person,
which is widely used in social media platforms for
tagging or sorting images.
2. Object Detection
Object detection goes beyond simple classification by not
only identifying what is in an image but also locating
where it is. This process typically involves drawing
bounding boxes around the detected objects in an image.
Object detection requires the system to consider both
spatial information and the characteristics of the objects.
 Use Case Example: Self-driving cars use object
detection to identify pedestrians, traffic signs, other
6

vehicles, and obstacles, helping them navigate safely

on the road.

3. Image Segmentation
Image segmentation refers to the process of dividing an
image into meaningful sections or segments, often based
on different features, like colour or texture. This task is
important in scenarios where the boundaries of objects
need to be determined very precisely.
 Use Case Example: In medical imaging,
segmentation helps identify tumours or other
abnormalities by outlining the specific regions that
are affected, aiding in diagnosis and treatment
planning.
4. Facial Recognition
Facial recognition is a specific type of object recognition
where the system is designed to identify human faces. It
involves analyzing facial features such as the distance
between the eyes, nose, mouth, and chin, then
comparing these features against a database of known
faces to find matches.
 Use Case Example: Used in security systems for
surveillance or unlocking smartphones by recognizing
the user's face.
7

◊ How Does Computer Vision Work?

Computer vision systems rely heavily on

machine learning, especially deep learning
techniques, to process and understand visual
data. These systems work through several key
steps:
1. Image Acquisition: The first step in
computer vision involves capturing an
image or video, often using cameras or
sensors. The data from this capture is
processed by the system.
2. Preprocessing: Before the image can
be analyzed, preprocessing steps may be
applied to enhance the quality or normalize
the data. This might include reducing noise,
resizing images, or adjusting contrast to
make key features more distinguishable.
3. Feature Extraction: During this step,
relevant information or features (such as
edges, corners, and textures) are extracted
from the image to help the machine
recognize objects or patterns. Traditional
computer vision techniques used filters like
Sobel filters and Harris corner
8

detection to extract these features.

However, with deep learning, convolutional
neural networks (CNNs) are commonly used
for automated feature extraction.
4. Model Training: Machine learning
models, particularly deep neural networks,
are trained on large datasets of labelled
images. For instance, a neural network
might be trained to recognize faces by
showing it thousands of images of faces,
each labelled with the identity of the
person. The model learns the patterns and
features that are important for making
accurate predictions.
5. Prediction/Recognition: After training,
the model can predict or recognize objects
in new, unseen images by applying the
learned patterns and features to classify or
locate objects.
◊Applications of Computer Vision

Computer vision has a wide range of practical

applications across various industries:
1. Autonomous Vehicles: In self-driving
cars, computer vision systems process data
9

from cameras and sensors to detect other

vehicles, pedestrians, and road signs. This
enables the car to navigate safely without
human intervention.
2. Healthcare: In the medical field,
computer vision is used to analyze medical
images such as X-rays, MRIs, and CT scans.
It can identify tumours, fractures, or other
abnormalities more quickly and accurately
than human doctors in some cases. It also
aids in robotic surgeries and real-time
diagnostics.
3. Retail: Retailers are using computer
vision for inventory management, customer
tracking, and even cashier-less stores.
Systems can detect when products are out
of stock or when customers are looking at
specific items, helping to optimize
operations.
4. Surveillance and Security: In
security, computer vision systems can
monitor video feeds to detect unusual
activities or identify individuals using facial
recognition technology. This is becoming
common in public places like airports,
stadiums, and shopping centres.
10

5. Entertainment and social media:

Apps like Instagram and Snapchat use
computer vision to apply filters that track
facial movements and overlay effects. It
also enables features like automatic image

tagging by recognizing the content of the

photo.
6. Manufacturing and Quality Control:
In industries, computer vision can be used
to inspect products on assembly lines. It
can detect defects or quality issues in real-
time, improving efficiency and reducing
human error.
11

CO
MP
UT
◊ Healthcare and Medical Imaging

ER
 Computer vision in healthcare is a
transformative tool with the potential to
enhance diagnostic accuracy, speed up the
ViSmedical workflow, and offer personalized
care for patients. One of the most prominent

IO applications of computer vision in

healthcare is diagnosis through medical
imaging. Technologies like Convolutional
NNeural Networks (CNNs) are trained to
analyze medical images—such as X-rays, CT

AP scans, or MRIs—and detect abnormalities

that may indicate diseases like cancer,
fractures, or even subtle organ changes. For
PLIinstance, CNN-based models can detect
early-stage breast cancer by analyzing
mammograms with remarkable accuracy,
CA often outperforming human radiologists in
spotting microscopic signs that the naked

TIOeye may miss.

 Beyond diagnostics, computer vision plays a
critical role in robot-assisted surgeries.
NS Integrating high-resolution cameras with
real-time image processing allows surgeons
to operate with enhanced precision,
12

minimizing human error. A notable example

is the da Vinci Surgical System, which
utilizes computer vision to offer surgeons a
3D, magnified view of the surgery site. It
enables more precise movements by
filtering out hand tremors and providing
clearer visual feedback, thus improving
surgical outcomes and reducing recovery
times.
 Another growing application is in
telemedicine and remote patient
monitoring. By analyzing video footage,
computer vision systems can monitor
patients with chronic conditions or elderly
individuals at home. For example, patients
with Parkinson’s disease can be observed
for changes in their motor functions, such as
tremors or gait disturbances, and doctors
can adjust treatment plans based on these
observations without the patient needing to
visit the clinic frequently. In dermatology,
apps like Skin Vision use computer vision
algorithms to analyze images of skin lesions
and provide users with information about
the likelihood of skin cancer, encouraging
earlier detection and treatment.
 In addition to benefiting patients, computer
vision helps reduce hospital workload.
Tasks like organizing medical records,
tracking patient flow, and even disinfecting
hospital spaces are increasingly automated
13

with AI and computer vision, allowing

medical staff to focus on more critical tasks.
 Computer vision in healthcare is still
evolving, with exciting future possibilities,
such as personalized medicine, where AI
systems analyze genetic and physical data
to recommend tailored treatments.
However, challenges remain, particularly
regarding data privacy and ethical
concerns surrounding AI decision-making in
life-or-death scenarios. As computer vision
continues to develop, its role in
revolutionizing healthcare is undeniable.

◊ Autonomous Vehicles
 Autonomous vehicles, or self-driving cars,
represent one of the most futuristic and exciting
applications of computer vision. These cars rely
heavily on visual data to understand and
navigate their environment, making computer
vision an essential technology for achieving fully
autonomous transportation. At the core of self-
driving systems are cameras, lidar sensors, and
radar systems that continuously feed data into
machine learning models to interpret the
14

surroundings, make decisions, and take actions

in real time.

 One of the key applications of computer vision in

autonomous vehicles is object detection. Cars
must detect and classify objects in their
environment—pedestrians, cyclists, other
vehicles, road signs, and traffic lights. Using
algorithms like YOLO (You Only Look Once) or R-
CNN (Region-Based Convolutional Neural
Networks), these systems can identify objects in
real time and predict their movement, enabling
the car to react appropriately. For example, when
approaching a crosswalk, the car detects
pedestrians and calculates whether they will
cross the street. Based on this information, it can
either slow down or continue driving.

 In addition to object detection, environment

mapping is crucial for self-driving cars.
Autonomous vehicles use a combination of
cameras and lidar (light detection and ranging)
systems to generate 3D maps of their
surroundings. These maps help the car
understand the spatial layout, including the
location of buildings, lanes, and other vehicles.
For instance, Tesla’s Autopilot system uses a
combination of cameras and ultrasonic sensors to
create a comprehensive understanding of the
road environment, allowing it to follow lanes and
avoid collisions.
15

 Collision avoidance is another critical feature

made possible by computer vision. By constantly
analyzing the environment, the vehicle’s
computer vision system can predict potential
accidents and take preventative actions, such as
emergency braking or steering to avoid
obstacles. Systems like Mobileye’s ADAS
(Advanced Driver Assistance System) use
cameras to monitor road conditions and detect
hazards, warning the driver and, in some cases,
taking corrective actions automatically.

 The future of autonomous vehicles also includes

vehicle-to-vehicle (V2V) and vehicle-to-
infrastructure (V2I) communication, where
cars and traffic systems share information. For
example, if a car ahead suddenly brakes, nearby
vehicles receive this information instantly,
preventing accidents. This integration of
computer vision with other AI technologies
promises to create smarter, safer roads.

 However, the challenges to the widespread

adoption of self-driving cars are significant. There
are technical issues related to unpredictable
human behaviour, poor weather conditions (like
fog or snow), and incomplete data for edge cases
that autonomous vehicles still struggle with.
Moreover, there are legal and ethical questions
surrounding accountability in case of accidents.
16

Despite these hurdles, advancements in

computer vision and AI will eventually lead to
safer, more efficient transportation.

Security and Surveillance

 In the realm of security and surveillance,
computer vision plays a pivotal role in enhancing
safety measures, reducing crime rates, and
ensuring public safety. Security systems that rely
on computer vision are far more efficient and
accurate than traditional methods, allowing for
real-time monitoring, facial recognition, and
anomaly detection.
 One of the most widespread applications of
computer vision in security is facial
recognition. This technology can be seen in
airports, public venues, and even smartphones. It
works by analyzing facial features from video or
still images and matching them to stored
databases to identify individuals. For example, in
airports, facial recognition is used for identity
verification at security checkpoints, reducing the
need for manual passport checks and speeding
up the boarding process. It’s also used in law
enforcement to track down suspects or missing
persons by scanning CCTV footage. However, this
technology raises privacy concerns, as it can
be used to track individuals without their
17

consent, leading to ethical debates about

surveillance overreach.
 In addition to facial recognition, computer vision
is integral to anomaly detection. This involves
the real-time analysis of video feeds to identify
unusual or suspicious activities, such as loitering,
trespassing, or vandalism. AI-powered systems
can flag these activities and alert security
personnel, making surveillance much more
proactive. For example, in a shopping mall, an AI
system can detect a person behaving unusually,
such as repeatedly circling a particular store or
acting nervously, and alert security before an
incident occurs. This kind of predictive
surveillance can prevent crimes before they
happen.
 Crowd management is another application of
computer vision in security. In large public
gatherings, such as concerts or sporting events,
monitoring the flow of people and detecting
overcrowded areas is crucial to avoid accidents
or stampedes. Computer vision systems can
analyze video footage from multiple cameras to
understand crowd dynamics and send alerts
when areas become too congested. This can help
authorities redirect the crowd and prevent
dangerous situations.
 In the context of smart cities, computer vision
enhances public safety by monitoring traffic,
enforcing road rules, and detecting hazards. For
example, cameras equipped with vision
algorithms can automatically detect vehicles
running red lights or illegal parking, and issue
tickets without human intervention. Furthermore,
these systems can identify accidents, debris, or
18

other obstructions on roads, ensuring faster

emergency response times.
19

◊ Retail and E-commerce

 Computer vision is revolutionizing the retail
and e-commerce industries, enabling
businesses to offer more personalized,
efficient, and immersive shopping
experiences. As consumer demands
continue to shift toward digital-first
interactions, computer vision helps retailers
streamline operations, understand customer
behaviour, and enhance the online shopping
journey.
 One of the most prominent applications of
computer vision in retail is visual search.
Traditional keyword-based searches often
fail to capture the exact product a consumer
is looking for, especially when it's based on
appearance. With visual search, consumers
can upload an image of a product they like—
such as a pair of shoes, a jacket, or a piece
of furniture—and the computer vision
system will analyze the image to find similar
products in the retailer’s catalo. This is
especially useful in fashion and lifestyle
categories, where visual appeal is a major
factor in purchasing decisions. Large
retailers like Amazon and ASOS have
already integrated visual search features
into their apps, allowing shoppers to find
products by simply taking a picture.
 Inventory management is another area
where computer vision is making a
20

significant impact. Managing stock levels

accurately is crucial for retailers and manual
methods can be time-consuming and error-
prone. Computer vision systems can
automatically track inventory by using
cameras placed throughout a store or
warehouse. These systems can scan
barcodes or recognize products on the
shelves, notifying staff when stock is
running low or when items are misplaced.
For example, Walmart has experimented
with robots that use computer vision to
monitor inventory levels and restock shelves
as needed. This not only increases efficiency
but also reduces human error and labour
costs.
 In e-commerce, computer vision is essential
for improving the online customer
experience through features like virtual
try-ons. Shoppers can upload a photo of
themselves or use their webcam to virtually
"try on" clothing, accessories, or makeup,
giving them a better sense of how the
product will look on them before making a
purchase. Companies like Warby Parker and
Sephora have successfully implemented
these virtual try-on systems, allowing
customers to visualize products like glasses
or lipstick on their faces without visiting a
physical store. This technology is
particularly useful during the pandemic
21

when consumers have shifted to online

shopping, and in-store visits are limited.
 Computer vision is also being used to
analyze customer behaviour in physical
stores. Cameras equipped with AI can track
how customers move through a store, what
products they spend the most time looking
at, and even their emotional reactions to
certain items. By analyzing this data,
retailers can optimize their store layouts,
product placements, and marketing
strategies. For example, if a certain product
display attracts a lot of attention but results
in few sales, retailers can adjust pricing,
positioning, or promotions to increase
conversion rates.
 Checkout automation is another area
where computer vision is having a
transformative impact. Amazon Go stores
are leading the way with their "just walk
out" technology, where customers can pick
up items and leave the store without having
to go through a traditional checkout
process. Cameras and sensors track the
items as customers place them in their cart,
and the system automatically charges their
account when they exit the store. This
frictionless experience reduces wait times,
eliminates the need for cashiers, and
provides a seamless shopping experience
for consumers.
22

 Moreover, personalization in e-commerce

is driven by computer vision as well.
Algorithms analyze visual data to
recommend products based on a user’s past
purchases or browsing history. This allows
retailers to create tailored marketing
campaigns and recommend relevant
products. For instance, if a shopper
frequently browses for athletic wear,
computer vision-powered recommendation
engines can suggest similar items, thus
increasing the likelihood of a sale.
 While the benefits of computer vision in
retail and e-commerce are clear, there are
challenges to overcome, particularly
concerning data privacy and consumer
trust. The collection of vast amounts of
visual data, especially in physical stores,
raises questions about how this data is
stored, used, and shared. Retailers must
ensure that their computer vision systems
comply with privacy regulations, such as
GDPR, and are transparent with customers
about data usage.
 In conclusion, computer vision is
transforming retail and e-commerce by
enhancing both operational efficiency and
customer experiences. As technology
advances, it will continue to reshape how
consumers shop, interact with brands, and
receive personalized recommendations.
23

◊ Agriculture and Farming

 The agricultural industry is undergoing a
digital transformation, with computer
vision playing a critical role in making
farming more efficient, sustainable, and
profitable. By combining machine learning
and computer vision, farmers are now able
to monitor crop health, optimize resource
use, and even automate labour-intensive
tasks. These advancements are
particularly important as the global
population grows, putting more pressure
on food production systems.
 One of the most impactful applications of
computer vision in agriculture is crop
monitoring. Traditional methods of
assessing crop health—such as manual
inspections—are time-consuming, labor-
intensive, and prone to human error. With
computer vision, drones and satellite
imagery can capture high-resolution
images of large agricultural fields. By
analyzing these images, AI models can
detect patterns in plant health, identify
areas affected by diseases, pests, or
nutrient deficiencies, and predict yields.
For example, early detection of fungal
infections or insect infestations allows
24

farmers to take preventative measures

before the problem spreads, minimizing
crop loss. These systems can also monitor
the growth rate of crops, helping farmers
optimize planting and harvesting times.
 In addition to disease detection, computer
vision is used for precision agriculture,
which involves the efficient use of
resources like water, fertilizer, and
pesticides. Drones equipped with
multispectral cameras can monitor soil
moisture levels and determine which areas
of the field need irrigation. This ensures
that water is distributed where it’s needed
most, reducing waste and promoting more
sustainable farming practices. Similarly,
AI-powered vision systems can analyze
leaf colour and texture to identify nutrient
deficiencies and recommend targeted
fertilizer applications, preventing
the overuse of chemicals that could harm
the environment.
 Automated harvesting is another area
where computer vision is making
significant strides. Harvesting crops,
especially fruits and vegetables, is labour-
intensive and time-sensitive, as produce
needs to be picked at the right stage of
ripeness to maximize yield and quality.
Vision-guided robots are now being used
to pick crops with high accuracy, ensuring
only ripe produce is harvested. These
25

robots use cameras to assess the colour,

size, and texture of fruits, such as
strawberries or tomatoes, and use robotic
arms to gently pluck them from the plant.
This automation reduces labour costs,
speeds up the harvesting process, and
ensures consistency in the quality of the
produce.
 Beyond field crops, computer vision is
being used in livestock monitoring.
Farmers can monitor the health and
behaviour of their animals’ using cameras
and AI models that analyze video footage.
For example, vision systems can detect
signs of illness, stress, or injury in cattle by
observing their posture, gait, and eating
patterns. Early detection of health issues
allows farmers to intervene quickly,
reducing the need for costly veterinary
treatments and improving animal welfare.
These systems can also track the
movement of animals to prevent
overcrowding in certain areas and ensure
that all animals have access to food and
water.
 Weed control is another area where
computer vision is proving beneficial.
Traditionally, herbicides are applied
uniformly across entire fields, which can
be wasteful and environmentally
damaging. Computer vision systems,
however, can identify and target individual
26

weeds, allowing for precision herbicide

application. This not only reduces the
amount of chemicals used but also
minimizes the impact on the surrounding
crops and soil.
 While the potential benefits of computer
vision in agriculture are vast, there are
challenges to widespread adoption. Cost
remains a significant barrier, particularly
for small-scale farmers who may not have
the resources to invest in high-tech
equipment. Moreover, the complexity of
setting up and maintaining these systems
requires specialized knowledge, which
may be lacking in some rural areas.
 However, as computer vision technology
becomes more affordable and accessible,
its adoption in agriculture is expected to
grow. By enabling more precise, efficient,
and sustainable farming practices,
computer vision will play a crucial role in
ensuring food security and meeting the
demands of a growing global population.
27

Tools used here

Computer vision models rely on a range of tools
and techniques to process, interpret, and
understand visual data. Here are some key tools
and libraries often used:
1. Programming Languages
 Python: Most common for computer vision due
to the variety of libraries.
 C++: Often used for performance-critical tasks.
2. Libraries and Frameworks
 OpenCV: A popular library for real-time
computer vision applications.
 TensorFlow & Keras: Deep learning
frameworks widely used for building and
training vision models.
28

 PyTorch: Another deep learning framework

popular for image classification, segmentation,
and object detection.
 sci-kit-image: A Python library focused on
basic image processing.
 Dlib: Used for facial recognition, object
detection, and more.

3. Deep Learning Models

 Convolutional Neural Networks (CNNs):
Specialized for processing images (like VGGNet,
ResNet, Efficient Net).
 YOLO (You Only Look Once): Real-time
object detection system.
 Mask R-CNN: Used for instance segmentation
(detecting objects and their boundaries).
 UNet: Often used in
Thanks