Key differences between Computer Vision (CV) and human vision as highlighted in the

document:

1. Process of Understanding:

○Human Vision: Humans see objects when light reflects off them and enters the
eye. The brain then interprets the image by understanding color, shape, and
movement.
○ Computer Vision: Computers interpret images through mathematical algorithms
without any human assistance. They rely on training models to simulate visual
perception.
2. Accuracy and Detail:

○ Human Vision: Limited by factors like fatigue, subjective judgment, and varying
sensitivity to colors and details.
○ Computer Vision: Offers consistent accuracy and can analyze minute details
(like pixel values) that might be overlooked by humans.

Learning and Adaptation:

● Human Vision: Humans learn to recognize objects through natural experiences and
develop an understanding over time without explicit training.
● Computer Vision: Computers require extensive training with labeled datasets and
algorithms (e.g., neural networks) to recognize and classify objects, relying on
predefined categories and data samples.

Here are two important features of a pixel in a digital image as mentioned in the document:

1. Color Representation:
Each pixel represents a specific color in an image. For grayscale images, the pixel value
ranges from 0 (black) to 255 (white). For RGB images, a pixel contains three values
(Red, Green, and Blue), each ranging from 0 to 255, which combine to form a specific
color.Intermediate shades of gray are represented by equal brightness levels of the three
primary colours.

2. Smallest Unit of Image Data:

A pixel is the smallest unit of information in a digital image. It is the fundamental building
block, and the overall resolution and quality of an image depend on the number of pixels
it contains.
AI Ethics:

1. Data Privacy: AI systems require large amounts of user data, raising concerns about
data collection, storage, and misuse. Apps often collect data legally through user
permissions, but users should be cautious to protect their privacy.
2. Unemployment: AI automation in industries, customer service, and transportation can
replace human jobs, leading to increased unemployment.
3. AI Bias: AI systems can inherit biases from developers or datasets, leading to unfair
decisions in areas like hiring, loans, or advertising, which can amplify social inequalities.
4. AI Access: Advanced AI technologies are expensive, creating a divide between those
who can afford them and those who cannot, worsening inequality.
5. Ethical Usage: AI tools must be used responsibly, especially by children, to ensure their
benefits contribute to society's welfare without causing harm.

For a 5-mark answer on Problem Scoping, you can write:

Problem Scoping is the first and crucial stage of the AI project cycle where the problem is
identified and analyzed. It uses the 4Ws Problem Canvas to address:

1. Who: Identifies stakeholders impacted by the problem or benefiting from its solution.
2. What: Defines the problem's nature and gathers evidence to confirm its existence.
3. Where: Pinpoints the context, situation, or location where the problem is most
prominent.
4. Why: Explains the benefits of solving the problem for stakeholders and society.

This process ensures a clear understanding of the problem and lays a strong foundation for
achieving the project's goals.

For a 5-mark answer on Data Acquisition, including references to open-source portals:

Data Acquisition is the second stage of the AI project cycle, where relevant and accurate data
is collected to train the AI model.

Key points include:

1. Purpose: Data is crucial for training AI models to make accurate predictions or

decisions.
2. Features of Data: The data must be authentic, accurate, and relevant to the problem.
3. Sources of Data: Data can be gathered through:
○ Surveys: Feedback and reviews from users.
 ○ Web Scraping: Extracting data from websites.
○ Sensors: Tracking physical conditions in real-time.
○ Cameras: Collecting visual data from surveillance or live feeds.
○ APIs: Application-generated data.
○ Open-source Portals: Reliable data can be accessed from government websites
like data.gov.in and india.gov.in, which provide structured and authentic
datasets for various use cases.
4. Example: For predicting traffic flow, past traffic data for the same location and time
period is acquired and used for training the model.

The quality and relevance of the data directly affect the AI model's performance.

For a 5-mark answer on Data Exploration, you can write:

Data Exploration is the third stage of the AI project cycle, where the collected data is analyzed
to uncover patterns or trends. Key points include:

1. Purpose: To make sense of complex, unstructured data and determine its usefulness for
the AI project.
2. Activities:
○ Researching, filtering, and organizing data.
○ Identifying trends, patterns, and correlations within the data.
3. Visual Representation: Visualizing data through graphs, charts, or diagrams helps
simplify analysis and strategy development for the AI model.
4. Importance: This stage ensures the data aligns with project goals and helps decide the
type of model required for further stages.
5. Example: If analyzing laptop purchase trends, data exploration might involve
categorizing configurations like RAM, storage, and processor to identify popular choices.

Effective data exploration is critical to the success of AI projects, as it lays the groundwork for
building accurate models.

Data Modeling is the process of creating a mathematical model based on prepared data to
analyze and predict outcomes. It involves selecting algorithms and techniques to build AI
systems that can identify patterns and relationships within the data.