Computer Vision: How Machines See

Hello students, welcome to Kwickprep. You glance at your phone, and it unlocks. It recognised your face in less than a second. But a phone has no eyes, so how does it see? Today we learn how a computer sees an image, the steps of computer vision, where it is used, and its challenges and future.

Let us define this branch of AI. Computer vision is a field of AI that helps computers understand images and videos. Its input is a photo, or the frames of a video, which are still pictures shown quickly. Its output is information, such as what objects are present and where they are. The big difference is this: we see a face, but a computer only sees numbers.

Zoom far into any photo and you will see tiny coloured squares. Each square is a pixel, short for picture element, the smallest dot of an image. An image is simply a grid of pixels arranged in rows and columns. Resolution tells how many pixels there are, written as width by height. A photo of nineteen twenty by ten eighty has about twenty lakh pixels.

The computer stores every pixel as a number. In a simple black and white image, each pixel is zero or one. In a greyscale image, each pixel is a number from zero for black to two hundred fifty five for white. In a colour image, each pixel has three numbers, for red, green and blue, so two hundred fifty five, zero, zero is pure red.

Computer vision works in a series of steps. First, image acquisition means capturing the image with a camera or scanner. Second, preprocessing cleans it by resizing, removing noise and fixing brightness. Third, feature extraction finds useful details like edges, corners and shapes. Fourth, detection finds objects, and segmentation marks exactly which pixels belong to each object. Fifth, high-level understanding gives the final meaning, such as, this is Riya's face.

Inside these steps, computer vision performs a few common tasks. Classification answers, what is in this image, like cat or dog. Object detection answers, where is each object, by drawing boxes around cars on a road. Segmentation answers, exactly which pixels belong to the object, like outlining a tumour in a scan. Face recognition answers, whose face is this?

Let us see the process inside face unlock. The front camera captures an image of your face. The software measures features, like the distance between the eyes and the shape of the jaw. These features are compared with the face you saved during setup. If they match closely enough, the phone unlocks; otherwise, it stays locked.

Computer vision also helps doctors. It studies medical scans such as X-rays, CT scans and MRI scans. It can highlight possible signs of tuberculosis, broken bones or tumours. In India, such tools help screen patients in areas that have few specialist doctors. But the AI only assists, and a qualified doctor makes the final decision.

The most exciting use is the self-driving car. Several cameras watch the road on all sides. Computer vision detects lane lines, road signs, traffic signals, vehicles and people crossing. With other sensors, it estimates how far away objects are and how fast they move. Many new cars in India already use this idea for driver assistance, like warning you when you drift out of your lane.

Computer vision is already part of daily life. Traffic cameras read vehicle number plates to catch rule breakers. Farming apps spot crop diseases from a photo of a leaf. Warehouses and shops count stock on shelves automatically. Photo apps apply filters and let you search your gallery for words like beach or dog.

Computer vision is powerful, but it faces real challenges. Poor light or glare changes pixel values, so the same object can look different. An object seen from a new angle, or partly hidden, is harder to recognise. If the training photos show too few kinds of faces, the system can be unfair to some people. Cameras everywhere raise questions of privacy and consent. And the same technology can create deepfakes, which are fake but realistic videos.

What does the future hold? Phones and home devices will understand the world around them even better. Early disease screening through a simple camera could reach far more villages. Driver assistance and smart traffic signals can make roads safer. And laws like India's Digital Personal Data Protection Act will guide how face data may be collected and used.

Let us revise what we learned today. A computer sees an image as a grid of pixel numbers. Computer vision works in steps: acquisition, preprocessing, feature extraction, detection, and high-level understanding. We saw it in face unlock, medical scans and self-driving cars. It still struggles with lighting, bias, privacy and deepfakes. As future creators of AI, always use it responsibly. Notice three places you meet computer vision today, and name the task each one does.