Computer Vision Introduction

Computer vision is a subfield of artificial intelligence that deals with acquiring,

processing, analyzing, and making sense of visual data such as digital images and
videos. It is one of the most compelling types of artificial intelligence that we
regularly implement in our daily routines.

Computer vision helps to understand the complexity of the human vision system
and trains computer systems to interpret and gain a high-level understanding of
digital images or videos. In the early days, developing a machine system having
human-like intelligence was just a dream, but with the advancement of artificial
intelligence and machine learning, it also became possible. Similarly, such
intelligent systems have been developed that can "see" and interpret the world
around them, similar to human eyes. The fiction of yesterday has become the
fact of today

Further, Artificial intelligence is the branch of computer science that primarily

deals with creating a smart and intelligent system that can behave and think like
the human brain. So, we can say if artificial intelligence enables computer
systems to think intelligently, computer vision makes them capable of seeing,
analyzing, and understanding.

History of Computer Vision

Computer vision is not a new technology because scientists and experts have been trying to
develop machines that can see and understand visual data for almost six decades. The
evolution of computer vision is classified as follows:

 1959: The first experiment with computer vision was initiated in 1959, where they
showed a cat as an array of images. Initially, they found that the system reacts first to
hard edges or lines, and scientifically, this means that image processing begins with
simple shapes such as straight edges.
 1960: In 1960, artificial intelligence was added as a field of academic study to solve
human vision problems.
 1963: This was another great achievement for scientists when they developed
computers that could transform 2D images into 3-D images.
 1974: This year, optical character recognition (OCR) and intelligent character
recognition (ICR) technologies were successfully discovered. The OCR has solved
the problem of recognizing text printed in any font or typeface, whereas ICR can
decrypt handwritten text. These inventions are one of the greatest achievements in
document and invoice processing, vehicle number plate recognition, mobile
payments, machine translation, etc.
 1982: In this year, the algorithm was developed to detect edges, corners, curves, and
other shapes. Further, scientists also developed a network of cells that could recognize
patterns.
 2000: In this year, scientists worked on a study of object recognition.
 2001: The first real-time face recognition application was developed.
 2010: The ImageNet data set became available to use with millions of tagged images,
which can be considered the foundation for recent Convolutional Neural Network
(CNN) and deep learning models.
 2012: CNN has been used as an image recognition technology with a reduced error
rate.
 2014: COCO has also been developed to offer a dataset for object detection and
support future research.

How does Computer Vision Work?

Computer vision is a technique that extracts information from visual data, such
as images and videos. Although computer vision works similarly to human eyes
with brain work, this is probably one of the biggest open questions for IT
professionals: How does the human brain operate and solve visual object
recognition?
On a certain level, computer vision is all about pattern recognition which
includes the training process of machine systems for understanding the visual
data such as images and videos, etc.

Firstly, a vast amount of visual labeled data is provided to machines to train it.
This labeled data enables the machine to analyze different patterns in all the
data points and can relate to those labels. E.g., suppose we provide visual data
of millions of dog images. In that case, the computer learns from this data,
analyzes each photo, shape, the distance between each shape, color, etc., and
hence identifies patterns similar to dogs and generates a model. As a result, this
computer vision model can now accurately detect whether the image contains a
dog or not for each input image.

Computer vision applications use input from sensing devices,

artificial intelligence, machine learning, and deep learning to
replicate the way the human vision system works. Computer
vision applications run on algorithms that are trained on massive
amounts of visual data or images in the cloud. They recognize
patterns in this visual data and use those patterns to determine
the content of other images.
How an image is analyzed with computer vision
 A sensing device captures an image. The sensing device is
often just a camera, but could be a video camera, medical
imaging device, or any other type of device that captures an
image for analysis.

 The image is then sent to an interpreting device. The

interpreting device uses pattern recognition to break the
image down, compare the patterns in the image against its
library of known patterns, and determine if any of the
content in the image is a match. The pattern could be
something general, like the appearance of a certain type of
object, or it could be based on unique identifiers such as
facial features.

 A user requests specific information about an image, and

the interpreting device provides the information requested
based on its analysis of the image.

Deep learning and computer vision

 Modern computer vision applications are shifting away from
statistical methods for analyzing images and increasingly
relying on what is known as deep learning. With deep
learning, a computer vision application runs on a type of
algorithm called a neural network, which allows it deliver
even more accurate analyses of images. In addition, deep
learning allows a computer vision program to retain the
information from each image it analyzes—so it gets more
and more accurate the more it is used.

Computer vision capabilities

There are three main functions for how computer vision programs process
images and return information:

Object classification
Object identification
Object tracking
Optical character recognition
1. Object classification

The system classifies the objects in an image according to a defined

category. For example, with object classification, a computer could
distinguish people from objects in a photo and determine how many
people appear in the photo.

2. Object identification
The system identifies a particular object in a photo, video, or image. For
example, with object identification, the system would be able to not only
distinguish people in a photo, but also analyze their appearance to determine the
identity or traits of those people.
3. Object tracking

The system analyzes a video to process the location of a moving object over
time. For example, with object tracking, a parking lot surveillance camera could
identify cars in a parking lot and provide information about the location and
movements of those cars over time.
4. Optical character recognition

The system identifies letters and numbers in images and convert that text
into machine-encoded text that can be read by other computer
applications or edited by users.

How Computer Vision Is Used

Computer vision systems use machine learning and deep

learning models to train the system to recognize aspects of
an image or video and make predictions about them. Types
of computer vision models include:

 Image classification for inspecting an image and

assigning it a class label based on the content. For
example, an image classification model can be used to
predict which images contain a dog, cat, or angry
customer.
 Image segmentation for identifying objects and
extracting them from their background, such as
isolating a tumor from surrounding brain tissue in X-
ray results.
 Object detection for scanning images or videos and
finding target objects. Object detection models
commonly highlight multiple objects simultaneously
and can be used for tasks such as identifying items on
shelves for improved inventory management or
anomalies in items on a production line.
 Object tracking for tracking movements of detected
objects as they navigate in an environment. For
 example, object tracking can be used in autonomous
driving to track pedestrians on sidewalks or as they
cross the road.
 Feature extraction for isolating useful
characteristics captured in an image or video and
sharing them with a second AI algorithm, such as
search and retrieve image matching. For example,
feature extraction can be used to automate traffic
monitoring and incident detection.
 Optical character recognition for extracting and
converting text from an image into a format that a
machine can read. This is often used in banking and
healthcare to process important documents and
patient records.

Computer Vision Applications across Industries

Computer vision is one of the most advanced innovations of artificial
intelligence and machine learning. As per the increasing demand for AI and
Machine Learning technologies, computer vision has also become a center of
attraction among different sectors. It greatly impacts different industries,
including retail, security, healthcare, automotive, agriculture, etc.

Computer vision enables a range of new use cases, helping

companies across industries solve real-world problems like
reducing operational costs, unlocking business automation,
and creating new services or revenue streams. Here are
the top industries using computer vision and the exciting
ways they apply this technology.

Industrial Automation and Manufacturing

Manufacturers use computer vision to enable automation,

which helps make production processes more efficient,
reduce human error, improve worker safety, and produce
higher outputs at lower costs. Some common applications
of computer vision in manufacturing include:
  Automated product inspection: Visual product
inspections are critical to quality control. By
automating optical inspections using production line
cameras, AI models for defect classification and
anomaly detection, and edge computing,
manufacturers can improve quality assurance
accuracy and speed.
 Safety monitoring: Computer vision can be used to
monitor factory floors to help ensure employee safety.
For example, analysis of real-time video can help
identify and alert staff to accidents or spills or detect
access to restricted, hazardous areas.

Healthcare

From preventive care to cancer treatment planning,

computer vision is used by healthcare organizations in a
variety of ways to help improve patient outcomes, enhance
accuracy, accelerate disease detection, and more.
Examples of how computer vision is being applied
in healthcare include:

 Medical imaging: Equipping CT scanners, X-ray

systems, endoscopy cameras, and other medical
imagining technology with computer vision systems
can help enable rapid processing of massive amounts
of data, streamlined workflows, and accurate and
efficient image evaluation. Deep learning technology
is being applied to assist with whole slide imaging in
digital pathology.
 Remote patient monitoring: Cameras and sensors
equipped with computer vision applications can be
used to collect and analyze data about patient
movement, such as gait or body positioning, to
identify deviations from established norms and alert
care team members to possible urgent needs.
Retail

From understanding where to place products to the optimal

time to restock inventory to in-store customer behavior
tracking, computer vision can help retailers discover
powerful insights about their operations for more-informed
business decision-making. Some applications of computer
vision in retail include:

 Loss prevention: Computer vision models can

analyze data from existing store cameras or self-check
kiosks to identify suspicious behaviour and send real-
time alerts to managers so they can intervene and
help stop fraud being committed.
 Touchless self-service checkout terminals:
Retailers looking to increase efficiency and enhance
the customer experience can leverage 3D smart scan
technology and computer vision models to capture,
detect, and recognize nonbarcoded food items,
enabling quick and convenient checkout with minimal
staff intervention.

Smart Cities

Smart city technologies can help gather video feeds from

street cameras so city leaders can make more-informed
operational decisions to help improve citizen safety,
mobility, and quality of life. Here are a few ways computer
vision can be applied in smart cities:

 Traffic management: City governments can

implement computer vision systems to monitor and
analyze street intersections and traffic patterns and
detect and track vehicles and pedestrians to optimize
traffic flow and help enhance safety at intersections.
 Infrastructure maintenance: Computer vision
models can be trained to recognize road and bridge
problems, such as potholes or cracked pavement,
 throughout a city or entire county and inform crews of
locations needing maintenance.

How Computer Vision Works

Computer vision combines components like edge
computing, cloud computing, software, and AI deep
learning models to enable computers to “see” data
collected from cameras and videos; quickly recognize
specific objects, people, and patterns; make predictions
about them; and take action if necessary.

The Role of Convolutional Neural Networks

Computer vision systems use deep learning models from a

family of algorithms known as convolutional neural
networks (CNNs) to guide image processing and analysis.
These deep learning models analyze the RGB values
embedded in digital image pixels to detect identifiable
patterns. CNNs can be developed to evaluate pixels based
on a wide range of features—including color distribution,
shape, texture, and depth—and accurately recognize and
classify objects.

Training Computer Vision Models

Before a computer vision system can be put to work, data

scientists and developers must train the system’s deep
learning model for its specific use case. This requires
inputting large amounts of application-specific data that
the model can use to recognize what it has been developed
to identify. For example, for a computer vision application
designed to recognize a dog, the model must first learn
what a dog looks like. It does this by being trained on
thousands, maybe even millions, of images of dogs of
different breeds, sizes, colors, and characteristics.
Most commonly, training takes place in data centers or
cloud environments. For especially complex training
initiatives, GPUs and AI accelerators can be applied to
expedite the process and better handle the increased
number of parameters involved. Once the model has
completed the training phase, it has the knowledge needed
to interpret and infer information from digital images. The
model may also be further fine-tuned or retrained over
time.
It’s also important to note that those seeking to build
computer vision solutions can use off-the-shelf,
foundational models as starting points for fine-tuning to
accelerate development times and avoid starting from
scratch.

Deploying Computer Vision Models

Once trained, computer vision models can be deployed to

computer systems to perform inference and interpret
conditions in the field—continuously assessing image and
video data to extract insights and information. While
computer vision solutions can run inferencing workloads in
the cloud or data center, many organizations today are
exploring edge AI applications, where computer vision
models run closer to where data is generated on
lightweight, optimized edge hardware or embedded
devices.
Moving AI inferencing capabilities closer to the edge can
offer several key benefits:

 Increased speed and lower latency: Moving data

processing and analysis to where it is generated helps
speed system response, enabling faster transactions
and better experiences that are vital in many
computer vision applications.
 Improved network traffic
management: Minimizing the amount of data sent
over a network to the cloud can reduce the bandwidth
and costs of transmitting and storing large volumes of
data.
 Greater reliability: The amount of data that
networks can transmit simultaneously is limited. For
 locations with subpar internet connectivity, storing
and processing data at the edge improves reliability.
 Enhanced security: With proper implementation, an
edge computing solution may increase data security
by limiting data transmission over the internet.
 Privacy compliance requirements: Some
governments, customers, or industries may require
that data being used for computer vision applications
remain in the jurisdiction where it was created. Edge
computing can help businesses stay compliant with
such rules and regulations.