📖 Chapter 2 of 9 · AI For New Learner Advance 2 (Multi Indian Language) · Class 8
Chapter 2: AI That SEES — Computer Vision
In this chapter
- How AI ‘sees’ — it’s just numbers, lots of them!
- Dhruv at Aravind Eye Hospital — AI saves sight.
- Case File: Niramai (breast cancer) and ISRO Chandrayaan-3.
- Behind the Code: training AI to recognise a mango.
- Debate: should face recognition be everywhere?
The Big Idea
Computer Vision is AI that ‘sees’. But the AI doesn’t see like we do. To AI, a photo is just a giant grid of NUMBERS — each number is a tiny dot of colour. Vision AI is the magic of finding patterns in those numbers. From there comes face unlock, X-ray reading, self-driving cars, DigiYatra, and ISRO’s moon landing.
Did you know?
ISRO’s Chandrayaan-3 used Computer Vision AI to look at the Moon’s surface DURING landing — and dodge boulders in real time. India became the first country to land near the Moon’s south pole. The ‘eye’ that saw the surface — was AI.
Story 1: Dhruv at Aravind Eye Hospital
From: Trivandrum, Kerala · Dhruv, age 12
Dhruv’s grandmother needed an eye check. Aravind Eye Hospital — famous all over the world, started right here in Tamil Nadu — was packed. Dhruv came along.
He expected long waits. Instead, a doctor’s assistant placed a small camera near Dadi’s eye. CLICK. The screen lit up: a beautiful, swirling picture of her retina — the back of her eye. Then green text appeared: “AI detected: early-stage diabetic retinopathy. Confidence 96.2%. Action: refer to specialist.”
The doctor smiled. “AI saw the signs before I could. Dadi’s eyes will be saved, beta. We just have time.”
Dhruv was stunned. “How?” The doctor explained: this AI had been trained on more than 1 lakh retina images. It had learnt to spot diabetic damage faster than any human. Aravind Hospital uses it for thousands of patients a year — many of them poor villagers who can never see a city specialist.
On the way home, Dhruv kept thinking: “That AI can SEE. Really see. Better than the best doctor sometimes. And it saved my Dadi’s eyes.” That night, he searched: “How do I learn to BUILD AI like that?”
How AI ‘sees’ — and two big wins
To a computer, an image is not a picture at all — it is a huge grid of numbers, and Vision AI finds patterns in them.
📁 CASE FILE Niramai — AI Against Breast Cancer
What it is: AI-powered thermal imaging that detects breast cancer — no X-ray, no pain, no radiation.
Built by: Bangalore startup founded by Dr. Geetha Manjunath.
Why it matters: Indian women often skip cancer screening. Niramai is portable, painless, and accurate. It is now used in hundreds of hospitals.
📁 CASE FILE Chandrayaan-3 — AI That Landed on the Moon
What it was: AI vision system that read the Moon’s surface DURING landing and steered the lander away from rocks and slopes.
Built by: ISRO, India’s Space Research Organisation, Bangalore.
Why it mattered: 23 August 2023: India became the FIRST country ever to land near the Moon’s south pole. The AI ‘eye’ did what no human eye could.
The challenge: The Moon is 3.8 lakh km away. Signal takes 1.3 seconds each way — too slow to control from Earth. The AI had to think for itself.
💻 BEHIND THE CODE Training a Mango Vision AI
# How AI learns to recognise a ripe mango
TRAIN mango_AI:
LOAD 5000 mango photos
FOR each photo:
LABEL it 'RIPE' or 'UNRIPE'
SHOW all photos to AI
AI looks at: colour, spots, shape
AI finds patterns
TEST mango_AI:
LOAD 100 NEW mango photos
IF accuracy > 95%:
DEPLOY to farmer's app
⚖ DEBATE IT! Should Face Recognition Be Everywhere?
DigiYatra at airports, attendance in schools, payment in shops, CCTV in markets — should ALL of these use face recognition?
YES — safer and faster
Find missing children. Stop terrorists. Speed up queues. India is a big country; we need AI to keep order.
NO — privacy and freedom matter
If every camera knows your face, you can be tracked everywhere. What if the data leaks? What if used wrongly by powerful people?
Solved illustrations
Test your Vision AI understanding
1. How is a digital photo just ‘numbers’?
Show answer
Answer: Every photo is a grid of tiny dots called pixels. Each pixel is 3 numbers (red, green, blue) from 0 to 255. A normal selfie has about 10 million numbers. AI’s job is to find patterns in this huge sea of numbers.
2. Why did Niramai use THERMAL (heat) images, not normal X-rays?
Show answer
Answer: Cancer cells are slightly warmer than normal cells. A heat camera can see this — without radiation, without pain. AI finds the warm patches even when they are too tiny for human eyes.
3. Why did Chandrayaan-3 need AI to land?
Show answer
Answer: The Moon is 3.8 lakh km away, so the signal takes about 1.3 seconds each way — too slow to control the landing from Earth. The AI had to read the surface (rocks, slopes) and make hundreds of decisions per second by itself.
4. What is the first step in building a vision AI?
Show answer
Answer: Collect and label lots of photos. For the mango AI, that meant 5000 photos each labelled RIPE or UNRIPE. Only after training and testing above 95% accuracy do you deploy it to a real app.
5. Name three places Computer Vision AI is used in India.
Show answer
Answer: DigiYatra face gates at airports, retina and X-ray reading in hospitals (Aravind, Niramai), Chandrayaan-3’s Moon landing, mango and crop sorting on farms, and UPI/CCTV safety systems. Vision AI is everywhere.
Questions — Chapter 2
Test your Vision AI knowledge
A. Fill in the blanks
- A digital photo is a grid of tiny dots called .
- Each pixel has 3 numbers — for , green, and blue.
- Aravind Eye Hospital is famous for AI-powered scans.
- uses thermal imaging to detect breast cancer.
- India landed on the Moon’s south pole using AI from .
B. Multiple Choice
- Computer Vision AI sees photos as — (a) magic (b) numbers (c) light
- Chandrayaan-3 needed AI because — (a) ISRO was lazy (b) Earth-to-Moon signal is too slow (c) it was cheaper
- The first step in training a vision AI is — (a) deploy it (b) collect labelled photos (c) sell it
C. Short Answers
- Name any THREE places where Computer Vision AI is used in India today.
- Why did Niramai use heat cameras instead of X-rays? Give two reasons.
- Explain the ‘Debate It’ on face recognition in your own words. Which side do YOU pick, and why?
Builder Challenge
Visit a vegetable market. Note 5 fruits AI could be trained to identify. Sketch the pipeline.
Check your understanding — Chapter 2
Tap an answer to check it instantly. These are for practice — nothing is saved or sent.
Q1. How does a digital photo become just 'numbers' to AI?
- It doesn't; AI sees colours
- Every photo is a grid of pixels, and each pixel is 3 numbers (red, green, blue) from 0 to 255
- AI converts photos to words first
Show answer
B. Every photo is a grid of pixels, and each pixel is 3 numbers (red, green, blue) from 0 to 255
Q2. Why did Niramai use thermal (heat) images instead of X-rays?
- Heat cameras are prettier
- Cancer cells are slightly warmer, so a heat camera can spot them without radiation or pain
- X-rays don't exist in India
Show answer
B. Cancer cells are slightly warmer, so a heat camera can spot them without radiation or pain
Q3. Why did Chandrayaan-3 need on-board AI to land?
- ISRO was being lazy
- The Earth-to-Moon signal takes about 1.3 seconds each way — too slow to control landing from Earth
- AI made it cheaper
Show answer
B. The Earth-to-Moon signal takes about 1.3 seconds each way — too slow to control landing from Earth
Q4. What is the first step in training a vision AI?
- Deploy it to users
- Collect and label lots of photos
- Sell it to a hospital
Show answer
B. Collect and label lots of photos